hash stringlengths 32 32 | doc_id stringlengths 7 13 | section stringlengths 3 121 | content stringlengths 0 2.2M |
|---|---|---|---|
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 15.6.2 Link Service Establishment | Figures 15.3 and 15.4 show examples of link service establishment with two different outcomes. Re-negotiation of service profile is not supported. If RCST B responds to the Mesh Control Establishment request with the lower service profile then the one it received from the Mesh Controller, the link service will be torn down. Figure 15.3 also illustrates the capacity assignment kick-start initiated from the Mesh controller that is needed to reduce the initial packet round-trip time. Without capacity requests from the Mesh controller which are sent on behalf of the RCSTs involved, the length of the initial packet round-trip time would inevitably caused retransmissions. Figure 15.4 shows an example of link service establishment failure. In this case the link service establishment was rejected because the RCST B was not logged on. There are many other reasons for link service establishment rejection. It can be due to an unauthorized request from RCST A, or a request containing erroneous data, it can be due to reject from RCST B because of lack of resources or some other. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 134 Figure 15.3: Example of DCP message exchange during link establishment RCST_B Mesh Controller RCST_A NCC NMS If none of Mesh Acks arrives after short timeout and no data is received i in that period, RCST B can send link Release request and tear down the link "Link complete", be ready to receive data If the Mesh Controller does not receive any if the ACKs, it will tear down the link. ACK is sent twice to reduce the loss probability DCPLoggedOn DCPLoggedOn Link authorization & routing MSC DCP_Establishment DCPLinkControlAck DCPLinkControlAck DCPLinkControlAck DCPLinkControlAck DCPLinkResponse DCPLinkRequest RCSTB_CapRequest DCPLinkResponse DCPLinkRequest RCSTA_CapRequest ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 135 Figure 15.4: Example of DCP Link Establishment Failure |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 15.6.3 Link Supervision | A RCST (A) that initiates a link service should also drive the supervision of the connectivity and quality of the corresponding link. It should do this by sending a keep-alive message to the peer RCST nominally 2 times per idle/fixed timeout interval, independent of the level of traffic. The peer RCST should immediately respond with a corresponding supervision message and should restart its peer link supervision timer. If RCST A does not get a response to its keep- alive signal, it should initiate link service release towards the MC. If RCST B does not receive a keep-alive signal before supervision timer expiry it should initiate link service release towards the MC. These internal keep-alive signals should not contribute to keep the connection alive, only externally initiated traffic should do that. RCST_B Mesh Controller RCST_A NCC NMS Link authorization & routing After minimum attempt hold-off, the RCST will again try to establish Link Link establishment is temporarily rejected Timer for minimum attempt hold-off starts Link authorization & routing DCPLoggedOn MSC DCP_Establishment_Failure DCPLinkServiceResponse with Reject DCPLinkServiceRequest DCPLinkServiceResponse with Reject DCPLinkServiceRequest ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 136 Figure 15.5: Example of DCP message exchange during link supervision |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 15.6.4 Link Service Release | A RCST that takes down a link service should do this by requesting the MC to take down both this link service and the corresponding link service used for the opposite direction. The RCST will wait for the response and acknowledge the response, and will release mesh if there is no response from the MC. The other RCST will acknowledge the link service release. The MC will release either RCST from mesh if there is not acknowledgement from the respective RCST. RCST_B Mesh Controller RCST_A Repeats keep-alive until Ack is received, max 10x, Then releases link. Link MSC DCP_Link_Supervision DCPLinkKeepAlive DCPLinkKeepAlive Supervises reception of Keep-alive Releases link if missing Link ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 137 Figure 15.6: Example of DCP message exchange during link service release |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 15.6.5 Link Service Keep Alive | A RCST operating a TX link service sends keep-alive regularly to the MC as long as the link service is in use. The MC should autonomously release the TX link service if it does not receive these keep-alive messages in time. The RCST should send keep-alive to the MC at an interval less than half of the shortest timeout interval for the link service, independent of the traffic activity. The MC should maintain a supervisory timer that autonomously initiates release of a link service if it is not kept alive by the RCST. The MC should simultaneously release the pair of link services used to serve a two-way mesh connection. RCST_B Mesh Controller RCST_A Repeats release until Ack is received, max 10x, Then releases RCST. LinkService MSC DCP_Link_Service_Release DCPLinkControlAck DCPLinkControlAck DCPLinkServiceRelease DCPLinkServiceRelease LinkService DCPLinkServiceRelease Initiate Link Service Release Repeats release until Response is received, max 10x, Then releases mesh ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 138 Figure 15.7: Example of Link Keep Alive The RCST maintains one keep-alive timer per TX link service, and issues keep alive when the timer expires. Reception of MC feedback for the specific link service should reset the link specific timer. RCST Mesh Controller MSC DCP_Mesh_Link_Keep_Alive DCPLinkServiceKeepAlive LinkService DCPLinkServiceKeepAlive Respond to Keep Alive Request Repeats keepalive until Response is received, max 10x, Then releases mesh. Release link service If MC does not acknowledge link. Release mesh if MC does not respond ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 139 Annex A: Interworking with the NGN service layer A.1 Policy and Charging Control (PCC) Architecture Despite the fact that the NGN design is focused to integrate different access networks, satellite networks have many singular characteristics that require further analysis and specification. This clause provides recommendations in regards to the integration of DVB-RCS2 satellite access networks with the Service Layer. Next Generation Networks (NGN) are all-IP based where all services are supported over IP-backbone (see Figure A.1). A DVB-RCS2 network is viewed in this clause as an example access network that can be integrated in the NGN architecture. In NGN, traditional circuit switched services (e.g. voice) and more feature rich multimedia service are supported over IP-backbone. One main aspect of NGN networks is a clear separation between the lower transport network layers and the upper service layers. This enables a common service environment to be used across different access networks. IP Multimedia Subsystem (IMS) is such a common service environment that is well specified in 3GPP [i.45] and has wide industry support. IMS provides the service requirements that are used by elements in the Policy and Charging Control (PCC) architecture to control the bearers in the underlying access network. Simply stated; IMS is the service layer, PCC is the control and DVB-RCS2 is the transport layer. The supported IMS services can for example be voice telephony services, video on demand, interactive IPTV, or video surveillance. New services can be developed with available IMS tooling. IP multimedia applications are, as a principle, not standardized, allowing rapid service creation and deployment using standard service capabilities. Figure A.1: Next Generation telecommunication networks are moving away from stove pipe architectures to multi-access / multi-service networks Policy and charging control (PCC) rules can be derived using information/requirements provided by the application function (AF). The AF represents applications that require dynamic policy and QoS control in the access network. IMS- based applications provide policy information to the AF. If for example an IMS-multimedia session need to be set-up and maintained, the AF (see Figure A.2) will feed the necessary policy information to the PCC elements (PCRF/PCEF). This clause describes the standard interfaces between the network transport layer and the service control layer as defined in the 3GPP specifications. The PCC (Policy and Charging Control) architecture is specified by 3GPP [i.46], the architecture specifies both 3GPP access (e.g. UMTS, LTE) as well as non-3GPP (Wifi, Wimax) access networks. A DVB-RCS2 satellite network is viewed as a non-3GPP access network that may use the PCC architecture. Within the PCC architecture it is possible to set up and control a session with multiple media streams. For many of these media streams, a specific QoS may be required. Especially in a capacity constrained access network like Satellite it is important that for services like voice communications, sufficient bandwidth is reserved and guaranteed. The PCC architecture enables the necessary control of underlying bearers. IMS uses the PCC architecture to control the QoS of bearers and / or IP flows in the IP access network. Figure A.2 depicts the PCC architecture. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 140 Gy Gz Subscription Profile Repository (SPR) Rx AF Sp Gx Offline Charging System (OFCS) Gxx BBERF PCEF Sd TDF Policy and Charging Rules Function (PCRF) PCEF Gateway Sy Online Charging System (OCS) NOTE: PCEF function in the network that enforces policies rules set in the PCRF to control services supported in the underlying all-IP network [i.46]. Figure A.2: PCC architecture The Policy Control architecture consists of the following functional components related to the integration with DVB-RCS2 access network; see Figures A.2 and A.3 where the components and interfaces are detailed: • Policy Control and Charging Rules Function (PCRF). The PCRF enables the provisioning of policy decisions to policy enforcement functions using PCC/QoS rules. The PCRF performs policy rule authorization for each policy request and assigns a QoS class (QCI) and QoS parameters to each rule for prioritization. • Policy and Charging Enforcement Function (PCEF). The PCEF has the capability of policing packet flow into an IP network or other transport network (e.g. by controlling a network router, and GWs) • 3GPP also defines a functional element to control non-3GPP accesses and serving Packet GWs through Gxx interface (Gxa, Gxb or Gxc): "Bearer Binding and Event Reporting Function" (BBERF). BBERF maps the policy decisions from the PCRF to access network specific parameters. • TDF (Traffic Detection Function) is defined in 3GPP Release 11. The TDF is a functional entity that performs application detection and reporting of detected application and its service data flow description to the PCRF. Using Sd interface, the PCRF may instruct the TDF on which applications to detect and report to the PCRF by activating the appropriate ADC (Application Detection Control) rules. The TDF may be also pre-configured on which applications to detect and report. It is very useful to provide policy control to traffic that is not based on IMS signalling. The TDF to PCRF reference point, listed as Sd (see Figure A.2), have strong similarities to the 3GPP system specific Gx reference point, because the Sd is a subset of the Gx. TDF was introduced to allow implementations where the traffic detection and enforcement functions are separated (e.g. from different vendors). Note that a significant number of operators implement a single vendor Packet Gateway (PDN gateway) that includes the PCEF function and traffic detection function. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 141 Figure A.3: QoS Rules in PCC architecture The second function in the PCC acronym, after Policy, is Charging Control. Charging the users for services is important in order to support the satellite operator's business. Satellite network operators may benefit from available BSS (Business Support Systems) and already developed systems in the mobile and wireline domain. The available PCC architecture standards, with standard interfaces Gy (for online charging and usage monitoring), or Gz (for off-line charging) should be used to interface with billing systems. For the purpose of charging correlation between application level (e.g. IMS) and service data flow level, applicable charging identifiers should be passed along within the PCC architecture. The operator should be able to off-line or online charge the users, via standards interfaces to BSS. The GSM/UMTS core network-charging architecture and principles are specified in [i.47], which provides an umbrella for other charging management documents that specify: • The content of the CDRs per domain and subsystem (offline charging). • The content of real-time charging messages per domain / subsystem (online charging). • The functionality of online and offline charging for those domains and subsystems. • The interfaces that are used in the charging framework to transfer the charging information (i.e. CDRs or charging events). Subscription information contained in the SPR (Subscription Profile Repository) or HSS (Home Subscriber Server), see Figure A.2, is used to set policy rules for a particular user. For example the QoS subscription information may be used to derive a policy rule that is used to enforce the maximum data rate of a service data flow that the user has subscribed to. A.2 Integrating DVB-RCS2 Access Network into the PCC architecture Figure A.4 shows a generic multi-service network scenario where the UE (User Equipment) is using a satellite network (OVN defined in HLS document) to access to different services. In the figure some UEs are subscribed to the Internet service provided by the Internet Service Provider (ISP), other UEs are subscribed to a wholesale ISP, and others to 3GPP/NGN/IMS services. The UEs are attached to "Customer Premises Networks (CPN)", typically Ethernet LANs. And each CPN uses a "Customer Network Gateway (CNG)" as access device. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 142 Figure A.4: General multi-service Satellite Access Scenario Several CNGs can be connected to a satellite terminal (ST – e.g. DVB-RCS2 RCST). The satellite terminal may function as the CNG itself. Different SVNs (Satellite Virtual Networks) may be used to segregate the traffic associated to different services. In Figure A.4, each SVN (with an individual colour) is one-to-one mapped to end service provider. In this scenario, the role of the OVN in the complete PCC architecture may be different depending on the integration approach. Two different approaches have been identified: 1) PDN GW (Packet Data Network GateWay) integrated in the OVN (Sat GW/NCC): This is a very satellite- centric approach where OVN is the only access network for users. In this approach, the OVN implements the PCEF and TDF functional components, interacting with the rest or the PCC components through Gx, Gy, Gz and Sd interfaces. In this approach, OVN implements both policy and charging control. Figure A.5 highlights the components and control interfaces of the OVN with this approach. Figure A.5: PDN GW integrated in the OVN (Sat GW) 2) OVN as a trusted access network of a general multi-service/multi-carrier network: In this approach, the OVN shares most of the PCC components with other access networks and service providers. This second approach is much simpler in terms of functionality and interfaces; and it is fully specified in PCC standards [i.48]. Also, it provides a real integration of the OVNs with the already existing networks. Figure A.6 highlights the components and control interfaces of the OVN with this approach. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 143 Figure A.6: OVN as a trusted access network For this second approach, different integration schemes have been defined in [i.48] for trusted IP access networks; all of them use Gxx (Gxa) as control interface (QoS rules provisioning and Event reporting) and S2a or S2c as data interfaces. In this guideline document the simplest scheme has been selected; it provides: • Full QoS control (through Gxa) for IMS sessions traffic and also for non-IMS traffic detected by TDF. In this scheme, the DVB-RCS2 network (OVN) should implement the BBERF functionality. • User mobility is based on "PMIPv6 Network Mobility" where UEs do not need any modification to support mobility. DVB-RCS2 network should implement the S2a PMIPv6 interface defined in [i.48]. Figure A.7 highlights the data and control interfaces of the OVN behaving as a trusted Non-3GPP IP Access. Figure A.7: Data and Control interfaces for OVN as Trusted Access ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 144 A.3 Interfaces and Reference Points The Reference Points in Table A.1 (which have the properties of interfaces) are those between DVB-RCS2 network entities and NGN entities within a DVB-RCS2 System and assume that certain entities are integrated into DVB-RCS2 entities (e.g. the BBERF/PCEF) as indicated in previous clauses. The interfaces are all based on the 3GPP definitions for these reference points. Table A.1 Reference Point Entities Use Gx (Diameter) PCEF - PCRF Policy enforcement and control Gxa (Diameter) PCRF and the BBERF Policy enforcement and control [i.46] Gy (Diameter) PCEF – OCS (online charging system) Online charging, online usage meeting for gating and/or throttling Gz PCEF – OFCS (Offline Charging System) Call data records (CDRs) FTP file transfer Ro (Diameter) CSCF – OCS Used to exchange online charging information with OCS Rx (Diameter) PCRF – CSCF Used to exchange policy and charging related information between P-CSCF and PCRF ISC (SIP) CSCF – AF Notify the AF of registration state, UE capabilities, etc. Sp HSS or SPR provide subscription data to PCC Sd TDF – PCRF Policy and Charging control S2a 3GPP-PDN – Trusted Non-3GPP-PDN 3GPP interface to Trusted Non-3GPP IP access network. It supports of mobility management of mobile devices [i.48] A.4 Interactions with DVB-RCS2 network A.4.1 Interaction between the PCEF/BBERF and PCRF The interface between the PCRF and PCEF/BBERF is via standard Diameter based interfaces. The policies that are defined in the PCRF are sent to the PCEF over the Diameter Interface. Note that it is not necessary to manage all session via the PCEF/BBERF, this depends on the enforcement rules set on the PCEF/BBERF. The policies are defined in AVPs (Attribute Value Pairs) that are exchanged between the PCRF and PCEF/BBERF. The full set of AVPs is specified in [i.47] and [i.49]. This is a standardized interface allowing different PCRF systems from different vendors and operators to connect over this interface. This is common practice in mobile networks where different service providers share the same network and each Service Provider (SP) is able to set policies (within an agreed set) for its own subscribers. PCRF provisions QoS rules to BBERF component (or PCEF if the first approach previously defined is used, where QoS rule is contained into a more general PCC rule) implemented into the OVN. A.4.2 Mapping of BBERF/PCEF to DVB-RCS2 controls For the policies to be enforced through the BBERF/PCEF function there should be a mapping that is DVB-RCS2 internal. This mapping should ensure that – for example – a required guaranteed bit rate for a voice service is enforced. The BBERF/PCEF has to configure all the satellite components involved in the IP flows contained in the QoS rule as SDF filters (Service Data Flows, See Figure A.3); both forward and return flows: NCC, GW and ST. To perform this configuration, new control interfaces needs to be defined internally in the satellite network as Figure A.8 shows. A SDF is an aggregate set of packet flows that matches a service data flow template (IPs, ports, etc.). ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 145 Figure A.8: DVB-RCS2 network Internal Policy Control interfaces First, BBERF/PCEF should bind the UE´s IP address to the ST it is sending traffic through. Afterwards, NCC needs to allocate enough return bandwidth to the ST based on the Return Maximum & Guarantied Bit Rates (MBR/GBR), and interact with the ST and GW to provision the return and forward QoS Rule. The terminal and the GW should use the rule to dynamically configure the Traffic Classification and Per Hop Behaviours for the corresponding IP flows. These two new internal control interfaces (called in the present document as Gxt and Gxg) need to be fully defined. The proposal for these new interfaces should be based on the "all IP" interfaces already defined for Policy Control, such as Gxx. Both Gxt and Gxg interfaces provision QoS rules according to [i.49] using diameter AVPs over TCP or SCTP connections between NCC and STs/GW. A more detailed definition of these new interfaces is provided below. A.4.3 Policy control on the RCST&GW Many UEs can be connected to a RCST, via for example a LAN interface. When an UE wants to setup a session with a particular QoS, the RCST will need to be aware of the QoS requirements for the return traffic of the requested session (e.g. Video Call). The "current approach" to control the traffic in a RCST is shown in Figure 7.4 in clause 7 where user plane and satellite bearer control functions are completely separated. The RCST classifies and schedules packets using a static configuration (e.g. managed through SNMP). It also controls the satellite resources using BoD techniques based on "Traffic Snooping" and "Buffers Monitoring". Because the RCSTs are unaware of service logic and PCC protocols, they should determine QoS parameters based on static configured rules and traffic snooping. Figure A.9 shows a new enforcement module (T-BBERF) and interfaces in the RCST as an extension to the "current approach" in Figure 7.5. T-BBERF module uses the QoS rules and the current available channel capacity to dynamically configure the classification, shaping/policing and packet scheduling (in general, Traffic Classification & Per Hop Behaviour, TC&PHB). A binding between the QCI contained in the QoS rule and a DiffServ Class should be configured and applied in the T-BBERF. Also, T-BBERF may inform the NCC about the status of the different aggregated flows per SVN and quality class as event reports. NCC capacity assignment can take into account both, the QoS Rules and the status reports (note that status reports are also useful for non policy controlled traffic). ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 146 Figure A.9: New T-BBERF Component in RCST Control Plane The internal behaviour of the new T-BBERF component is out of the scope of the guideline document; different implementations are possible and they do not affect the ST interoperability if it is compliant with the Gxt interface defined below. The same functional QoS enforcer module (called GW-BBERF in this case) should be implemented into the satellite GW to control forward traffic QoS. Both T-BBERF and GW-BBERF provide dynamic control over the user plane traffic handling and encompasses the functionalities defined in [i.48], section 4a.4.2 for the BBERF component. These functionalities are mainly: • It should ensure that the service data flow under QoS control is carried over the return or forward satellite bearer with the appropriate QoS class. The ARP, GBR, MBR and QCI parameters in the QoS Rules (see Figure A.3) are used for selecting the appropriate PHB (e.g. Weights of the packet scheduler). • Event reporting: It should report events to the NCC based on the event triggers installed by the NCC using the Gxt/Gxg procedures defined below. A.5 Example of a SIP call Figure A.10 shows an example of the steps required to complete a SIP call (e.g. Video Call) when the DVB-RCS2 Satellite Network (IP-CAN: IP Connectivity Access Network. A general term used to denote an Access Network that provides IP connectivity) is integrated with the PCC architecture. We assume that the BBERF has already established a Gateway control session with the PCRF as specified in [i.49]. Detailed control sequences can be found in [i.50]. These steps can be summarized as follows: 1) An UE requests a session (SIP call) to the "Call Session Control Function (CSCF)" of the IMS system acting as "Application Function (AF)". 2) The CSCF use the Rx reference point to exchange application level session information with the Policy and Charging Rules Function (PCRF). This information is part of the input used by the PCRF for the Policy and Charging Control (PCC) decisions. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 147 3) After admission control, the PCRF generate the corresponding QoS rule. It uses the standard Gxa interface to provision the QoS rules in the BBERF implemented in the NCC. This rule is bind to the Satellite IP-CAN connection (one per ST and SVN). 4) The rule enforcement required of the two new control interfaces internal to the Satellite Network (Gxt and Gxg) to enforce the rule in both the Sat-GW and the ST; where the enforcement functions should be implemented as discussed above to provide dynamic TC/PHB configuration. Figure A.10: Example, Steps complete a SIP Call with PCC The concrete implementation of an IP-CAN connection and a Bearer Service in the satellite network can be implementation dependent. As an example, the IP-CAN connection can be the portion of a DVB-RCS2 "connectivity channel" used by a SVN, and the Bearer Services can be the DiffServ Class associated to the QCI. Next clause details the signalling flows involved in this example; see Figures 14.16 and A.17. A.6 Gxt and Gxg Reference Points The Gxt reference point is located between the NCC and the T-BBERF (Satellite Terminal Bearer Binding and Event Reporting Function). The Gxg reference point is located between the NCC and the GW-BBERF (Satellite Gateway Bearer Binding and Event Reporting Function). The Gxt and Gxg reference points are used for: • Provisioning, update and removal of QoS rules from the NCC to the T/GW-BBERF. • Transmission of traffic plane events from the T/GW-BBERF to the NCC. These reference points are proposed to be fully compliant with the Gxx reference point defined in [i.49], (section 4a) where the NCC has the functionality of Policy Controller (PCRF) and the T/GW-BBERF has the functionality of BBERF. We provide below a description of the procedures and signalling flows involved in the policy control in a DVB-RCS2 network. The procedures and signalling flows for session termination and session modification are not provided in the present document, but all of them are compliant with Gxx procedures and protocol defined in [i.49]. Alternatively to the use of Diameter sessions for the Gxt/Gxg reference points, the provisioning of QoS rules and status reporting could be carried out using DCP. Clause 13.2.5 of the present document specifies other possible DCP functionalities, including dynamic QoS provisioning. Some adaptation to the already defined IEs, new IEs or new DCP messages may be necessary to include dynamic traffic classification rules. This adaptation should be possible since DCP has been specified to allow this degree of flexibility. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 148 A.6.1 Initial Satellite Terminal and Gateway Attachment procedure When a Satellite Gateway or Terminal starts-up it should establish a Diameter connection with the NCC. This connection will be used to send and receive all the Diameter messages related to Policy Control. Document [i.49], section 5a.2 details the Gxx procedures of "Initialization, maintenance and termination of connection and session" that we apply for Gxt and Gxg interfaces. With regard to the Diameter protocol defined over the Gxt/Gxg interface, the NCC acts as a Diameter server. The T-BBERF or GW-BBERF acts as the Diameter client. Figure A.11 shows the signalling flow required when the Sat-GW starts-up. Figure A.11: Initial Satellite Gateway Attachment Where: 1. The GW-BBERF (Sat-GW) establishes the transport connection. The initialization of the connection between the GW-BBERF (Sat-GW) and NCC is defined by the underlying transport protocol: TCP port 3864. 2. and 3. After establishing the transport connection, the NCC and the GW-BBERF should advertise the support of the Gxg specific Application using the CER (Capabilities Exchange-Request) and CEA (Capabilities-Exchange-Answer) commands specified in the Diameter Base Protocol [i.51]. "P1. Gateway Control Session Establishment": The GW-BBERF initiates this procedure in order initiate the policy control with the NCC; in this session establishment the default QoS rules and event triggers for all the SVNs that the Sat-GW handles may be deployed. This procedure "P1. Gateway Control Session Establishment" is defined in the next clause. Figure A.12 shows the signalling flow required when a ST starts-up. Figure A.12: Initial Satellite Terminal Attachment ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 149 Where: 1. After ST logon, the T-BBERF (ST) establishes the underlying TCP connection. 2. and 3. After establishing the transport connection, the NCC and the T-BBERF should advertise the support of the Gxt specific Application using the CER and CEA. "P1. Gateway Control Session Establishment": The T-BBERF initiates this procedure in order initiate the policy control with the NCC; in this session establishment the NCC may deploy the default QoS rules and event triggers for all the SVNs that the ST handles. This procedure "P1. Gateway Control Session Establishment" is defined in the next clause. "P2. Gateway Control & QoS Rules provision": The NCC deploys in the Sat-GW the default QoS rules and event triggers required for the traffic with the ST. This procedure "P2. Gateway Control & QoS Rules provision" is defined in the next clause. The procedures P1 and/or P2 are used in most of the policy control signalling flows, and they are defined in a general way in the next clause. A.6.2 Gateway Control Session Establishment Procedure on Gxa, Gxt and Gxg The Gateway Control Session Establishment Procedure on Gxx interface is fully defined in [i.49], section 4a.5.1 and [i.50], section 4.4.1. Note that the procedure has been simplified, not including the roaming scenarios fully defined in [i.49]. Figure A.13 shows this procedure that can be also applied on Gxt and Gxg interfaces: Figure A.13: P1. Gateway Control Session Establishment 1) The BBERF initiates a Gateway Control session with the PCRF by sending a CCR to the PCRF with the CC-Request-Type AVP set to the value INITIAL_REQUEST. The BBERF provides equipment identity and other information as defined in [i.49]. For the new T-BBERF component, the equipment identity may be the logon_Id of the ST when it is attached to the network; or it may be the UE MAC address when the UE is attached. The mapping between this identities and the IMSI subscriber identification required in Gxa interface should be performed by the NCC based on configuration data. 2) The NCC or PCRF performs the following actions: - It stores the information received in the CCR. - If it requires subscription-related information and does not have it, it requests such information. - It prepares for the installation of QoS rules if available. - It stores the selected QoS Rules and PCC Rules. On UE attachment, the NCC stores the binding of the Gxt session with the associated Gxa session. The PCRF may correlate the UE identity information with already established Gx sessions for the same UE. 3) The NCC or PCRF acknowledges the Gateway Control Session by sending a CCA to the T/GW-BBERF. It includes the available QoS rules and the event triggers. 4) The T/GW-BBERF installs and enforces the received QoS Rules. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 150 A.6.3 Gateway Control & QoS Rules Provision Procedure on Gxa, Gxt and Gxg The Gateway Control & QoS Rules Provision on Gxx interface is fully defined in [i.49], section 4a.5.2 and [i.50], section 4.4.3. Note that the procedure has been simplified, not including the roaming scenarios fully defined in [i.49]. Figure A.14 shows this procedure that can be also applied on Gxt and Gxg interfaces: Figure A.14: P2. Gateway control and QoS Rules Provision 1) The NCC or PCRF receives an internal or external trigger to update QoS Rules and event triggers for a gateway control session. The NCC/PCRF may decide to operate on QoS Rules without obtaining a request from the T/GW-BBERF, e.g. in response to information provided to the NCC via the Gxa reference point, or in response to an internal trigger within the NCC/PCRF. 2) The NCC or PCRF sends a Diameter RA-Request message (RAR) to request that the T/GW-BBERF installs, modifies or removes QoS Rules and/or updates the event triggers. 3) The T/GW-BBERF installs, modifies or removes the identified QoS Rules. The T/GW-BBERF also enforces the authorized QoS and enables or disables service flow according to the flow status of the corresponding QoS Rules. 4) The BBERF sends a Diameter RA-Answer message (RAA) to the NCC/PCRF to acknowledge the RAR and informs it about the outcome of the QoS rule operation. If the corresponding resource cannot be established or modified, then the T/GW-BBERF should reject the activation of a QoS rule as specified in [i.49]. A.6.4 User Equipment (UE) Attachment procedure There are many possible procedures to complete the attachment of a UE to the DVB-RCS2 network with support of policy control. The selected PMIPv6 S2a [i.48] interface permits that any UE, with no protocol modifications, can be attached to NGN network through using the satellite access, including IP mobility features. Section 4.7.2 of [i.48] defines different UE IP addressing schemes that can be applied, from the static allocation to dynamic allocation based on DHCPv4/v6. Also, section 6.2 of [i.48] defines the ignition attach procedure on S2a interface. As an example, Figure A.15 summarized the signalling flows of the UE attachment based on DHCPv4 and PMIP compliant with S2a interface: ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 151 Figure A.15: Example of UE attachment procedure 1. The UE sends a DHCPv4 Discovery message in broadcast to the network to find available servers. "P1: Gateway Control Session Establishment". The ST initiates the Gateway Control Session Establishment Procedure with the NCC and the NCC (BBERF) with the PCRF, as already defined. The DVB-RCS2 access network provides the information to the PCRF to correctly associate it with the IP CAN session to be established in step "P3". 2. Applying the PMIP architecture [i.52], the ST behaves as the MAG (Mobile Access Gateway) and the PDN GW as the LMA (Local Mobility Anchor). The ST sends a Proxy Binding Update (PBU) message to the PDN GW in order to request the new IPv4 address and update the current registration. Upon receiving the PBU message from the ST, the PDN GW allocates an IPv4 address for the UE in accordance with the operator's policies. "P3: IP CAN Session Establishment". The PDN GW initiates the IP CAN Session Establishment Procedure with the PCRF, as specified in [i.46]. The PDN GW provides information to the PCRF used to identify the session and associate Gateway Control Sessions established in "P1" correctly. The PCRF creates IP CAN session related information and responds to the PDN GW with PCC rules and event triggers. 3. The PDN GW responds with a PMIP Binding Acknowledgement (PBA) message to the ST with the assigned IPv4 Address. 4. The ST acting as a DHCPv4 server sends the DHCPv4 Offer with the assigned UE IPv4 address received in the PBA message in previous step. When the UE receives the lease offer, it sends a DHCPREQUEST message containing the received IPv4 address. The ST sends a DHCPACK packet to the UE. This message includes the lease duration and any other configuration information that the client might have requested. When receiving the DHCPACK message, the UE completes TCP/IP configuration process. "P2: GW Control & QoS Rules Provision". The PCRF updates the QoS rules in the DVB-RCS2 access network by initiating the GW Control & QoS Rules Provision Procedure. The NCC also updates the QoS rules in the ST and GW by initiating the GW Control & QoS Rules Provision Procedure. SIP and Applications traffic can now be sent through the configured PMIP tunnel (e.g. GRE) between ST and PDN GW, using the required service provider addressing plan. A.6.5 Signalling flows for IMS In [i.49] ("Annex B: Signalling Flows for IMS" and "Section 4.3.1: Network-Initiated IP-CAN Session Modification") we can find how IMS signalling is integrated with the PCC procedures that we have integrated with the proposed DVB-RCS2 policy control. Figure A.16 shows the PCC Procedures for IMS Session Establishment at originating CSCF and PCRF, where provisioning of service information is derived from SDP offer and answer. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 152 Figure A.16: Signalling flow for IMS SIP call Where: 1. The CSCF receives the first SDP offer for a new SIP dialogue within a SIP INVITE request. 2. The CSCF extracts service information from the SDP offer (IP address of the down link IP flow(s), port numbers to be used etc.). 3. The CSCF forwards the derived service information to the PCRF by sending a Diameter AAR over a new Rx Diameter session. It indicates that only an authorization check of the service information is requested. 4. The PCRF checks and authorizes the service information, but does not provision PCC/QoS rules at this stage. 5. The PCRF replies to the CSCF with a Diameter AAA. 6. The CSCF forwards the SDP offer in SIP signalling. 7. The CSCF receives the negotiated SDP parameters from the terminating side within a SDP answer in SIP signalling. 8. The CSCF extracts service information from the SDP answer (IP address of the up-link media IP flow(s), port numbers to be used etc.). 9. The CSCF forwards the derived service information to the PCRF by sending a Diameter AAR over the existing Rx Diameter session. 10. The PCRF replies to the CSCF with a Diameter AAA. 11. The PCRF selects the PCC Rule(s) to be installed, modified or removed for the IP-CAN Session. The PCRF may also update the policy decision by defining an authorized QoS and enable or disable the service flow(s) of PCC Rules. The PCRF may add or change QoS information per QCI applicable to that session. The PCRF may update the ADC decisions and select the ADC rules to be installed, modified or removed for the session. PCRF stores the updated PCC Rules, and ADC rules. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 153 "P2. Gateway Control & QoS Rule Provision". The PCRF initiates "Gateway Control and QoS rules provisioning procedures" following signalling flows described in Figure A.14. 12. The PCRF sends a Diameter RAR to request that the PCEF installs, modifies or removes PCC Rules and updates the policy decision. In the case of PCEF supporting Application Detection and Control feature, the PCRF may also request the PCEF to install, modify or remove the ADC rules by updating the ADC decisions for the session. 13. The PCEF installs, modifies or removes the identified PCC Rules. The PCEF also enforces the authorized QoS and enables or disables service flow according to the flow status of the corresponding PCC Rules. If QoS information is received per QCI, PCEF should set/update the upper limit for the MBR that the PCEF assigns to the non-GBR bearer for that QCI. In the case of PCEF supporting Application Detection and Control feature, when the solicited application reporting applies, the PCEF may also install, modify or remove the provided ADC Rules. 14. The PCEF sends a Diameter RAA to acknowledge the RAR. The PCEF informs the PCRF about the outcome of the PCC rule operation 15. In case of TDF, solicited application reporting, PCRF initiates the TDF session establishment, modification, or termination. 16. Upon successful authorization of the session, the SDP parameters are passed to the UE in SIP signalling. Figure A.17 is the same as Figure A.16 but it expands all the signalling flows. From this figure we can obtain the performance impact of the proposed PCC integration scheme. The complete SIP call requires only two additional satellite hops, having a total call establishment delay 1 second approximately. Figure A.17: Detailed Signalling flow for IMS SIP call ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 154 Annex B: COMSEC recommendations This clause presents the list of technical issues that occur from using Virtual Private Network (VPN) technologies in broadband satellite systems. The issues are described from the perspective of the satellite system integrator/operator point of view. The present document also proposes solutions to the technical issues, and provides recommendation and guidelines for efficient deployment of secure VPNs in broadband satellite systems. Three control cases are defined: • Case 1: The satellite system integrator/operator has control on both ends of the VPN or at least is able to recommend the VPN type, or the installation of features (e.g. Performance Enhancing Proxies, PEPs), or appropriate configurations on both sides of the VPN. • Case 2: The satellite system integrator/operator has no control on both VPN sides. This case means that the satellite system integrator/operator cannot choose the VPN technology, or cannot configure or modify the VPN devices, or cannot install or recommend the installations of PEPs before VPN processing. • Case 3: Case 3 is a mixture of Case 1 and Case 2. Here, the satellite system integrator/operator controls one end of the VPN but not the other. Hence, he is usually not able to recommend/chose the VPN technology but on one side of the VPN he is able to install or recommend the installation of (integrated) PEPs before VPN processing, or recommend configurations, etc. B.1 Issues with Performance Enhancing Proxies in secure VPNs Figure B.1 illustrates the normal TCP operation as well as the interception performed by a TCP acceleration PEP. The deployment of transport layer PEPs is not an issue for TLS/SSL-based protection, which leaves the transport layer accessible. In normal TCP operation, TCP data is acknowledge by the receiver after successful reception, meaning that it takes a round trip RTT1 until data is acknowledged. Since RTT1 is high in case a geostationary satellite link is involved, the bandwidth delay product limitation of TCP could be reached as mentioned previously. A PEP is only able to perform TCP acceleration in case it can send a faked TCP ACK packet successful to the TCP sender well before the original TCP ACK, resulting in a round trip time of RTT2, which is lower than RTT1. This has to be the function of the PEP independent whether the PEP splits the TCP session or just performs TCP ACK spoofing. In the following, we will give reasons why this PEP function is not possible on IPsec-protected data: • IPsec encryption: In case the TCP data is IPsec-encrypted, the PEP is unable to see the TCP header. Hence, it is not able to generate a TCP ACK message belonging to the respective TCP data. The PEP is even not able to determine the right TCP port and whether it is TCP data at all. • IPsec without encryption: In case the TCP data is not encrypted but IPsec integrity protection is deployed, the PEP is able to see the TCP header. It is able to generate an appropriate TCP ACK message but the PEP is unable to perform IPsec integrity protection without knowing the IPsec key. When sending the fakes TCP ACK message towards the sender, the IPsec GW will drop it since, due to normal security policies, only IPsec protected data is allowed. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 155 Figure B.1: Manipulation of IPSec data B.1.1 Possible solutions B.1.1.1 Positioning the distributed PEPs outside the VPN channel Provided having control case 1and having IPsec in tunnel mode, a straight forward solutions is to place the PEP functions in the path not subject to VPN protection, e.g. before VPN processing at the sender side and after VPN processing at the receiver side. Deploying the PEP process outside the VPN channel allows the PEP functions to access the headers and payload data in scope to enhance performance. The network architecture for using TCP acceleration via a distributed PEP solution is illustrated in Figure B.2. The PEPs on both sides have full access to the TCP layer and are able to split the TCP connection to use an enhanced transport protocol over the satellite link. In some network architectures PEPs do payload compression before the data enters the VPN tunnel. Hub Hub Satellite Server PC Servers VPN channel Standard TCP Standard TCP Enhanced transport protocol PCs PEP PEP VPN GW VPN GW Headquarter Remote station Figure B.2: Distributed PEP positioned outside the VPN channel ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 156 B.1.1.2 Positioning the integrated PEP outside the VPN channel Here we have to distinguish between an application layer PEP (e.g. HTTP cache or DNS cache) and a transport layer PEP, i.e. a PEP that is splitting TCP connections: Application layer PEP: Application layer PEPs like a web cache or a DNS cache are realized as integrated PEP. They are usually placed close to the hosts using the PEP to have short transmission times in case of a cache hit. Since the PEP usually does not know the VPN secret key, the only option is to place the PEP outside the VPN channel so that the respective protocol headers are accessible. The PEP terminates the application session between client and server and establishes a new session to the server. This network architecture is given in Figure B.3. Hub Hub Satellite PCs Servers VPN channel Standard protocol (e.g. HTTP or DNS) PEP VPN GW VPN GW Headquarter Remote station Figure B.3: Integrated PEP positioned outside the VPN channel Integrated transport layer PEP: An integrated transport layer PEP is usually based on a TCP splitting approach. It acts as the legal TCP receiver towards the TCP sender and terminates the TCP session by intercepting the TCP establishment. Afterwards, it establishes a new TCP session with the original TCP receiver. Hence, the TCP session is split into two parts. The usual deployment of an integrated PEP is at the satellite hub, as illustrated in Figure B.4. This has some advantages: • In case the PEP is located at the satellite hub, the path between TCP sender and PEP is usually based on a high-speed terrestrial link so the data transfer towards the PEP can be high-speed without the issues of satellite links (high RTT, higher packet loss). • Splitting the TCP session means that both TCP sessions have a lower RTT than the original TCP session. • In case the second part between PEP and TCP client is a satellite connection, an enhanced TCP version can be used as long as this is compatible with the TCP version used at the TCP clients. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 157 Satellite Clients Servers Standard TCP PEP Hub Enhanced TCP but compatible with receiver TCP Figure B.4: Integrated transport layer PEP at the hub Since operating at the transport layer, there are no issues with the deployment of VPNs based on TLS/SSL. However, in case of the deployment of a network layer VPN end-to-end the PEP is unable to access the transport layer without knowing the secret key, which is usually not acceptable for the user (from a security and management point of view). Hence, the only possibility is to position the PEP outside the VPN tunnel, which is just possible in case of IPsec is used in tunnel mode. Because the transport mode is used for end-to-end encryption, so there is no way to deploy the PEP outside the VPN tunnel. Two deployment options are possible: • Positioning of the integrated PEP at the Headquarter: In case TCP downloads are performed from the headquarter (e.g. the headquarter of a company or organization) to the remote station, the PEP can be installed at the headquarter. • Positioning of the integrated PEP at the Remote Station: In case TCP transfer is performed from the remote station to the headquarter (e.g. uploading of documents, videos, pictures, etc.), the PEP can be installed at the remote station. B.1.1.3 Deployment of SSL/TLS-aware proxies There are some solutions available to accelerate applications even when the VPN channel is protected by SSL/TLS. However, these solutions require control case 1 or 3. The deployment of SSL/TLS does not prevent TCP acceleration. However, some other performance improvements useful in satellite networks are not possible, e.g. caching and compression. In this clause, the deployment of caching in case of using HTTPS is discussed. A HTTP caching proxy caches the web content (e.g. just certain media or complete web pages) when a webpage is requested the first time to have it available when the same media or webpage is requested once again, either by the same or a different user connected to the proxy. There are HTTP caching proxy implementations available that support HTTPS/TLS. Thereby, the HTTPS/TLS connection between web browser and web server is split at the proxy. There are at least two preconditions given for this solution: 1) This function is not transparent for the user, i.e. the user has to explicitly configure the address of the caching proxy in its browser. 2) The user has to trust the proxy and the organization controlling it since in principle the proxy would be able to redirect requests to malicious servers. Hence, in most cases the proxy will be installed and controlled by the organization the user belongs to. Taking these preconditions into account, this solution is just possible in case of having control case 1 and control case 3. Of course, this solution is the more beneficial the more often the same web content is requested: either a single user is requesting web content several times or several users are connected to the proxy and have a similar browsing behaviour and interest. In order to save the time required for transmitting cached content over the satellite link, the best place for the HTTPS cache is at the remote side, as illustrated in Figure B.5. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 158 Hub Hub Satellite PCs Servers HTTPS HTTPS Cache Figure B.5: HTTPS caching proxy B.1.1.4 Selection of transport layer or application layer VPN methods Security mechanisms that protect the content above the transport layer (e.g. TLS/SSL or application layer security) allow PEPs on the protected path to perform TCP acceleration. Hence, in order to better support PEP deployment in satellite networks, a principle solution would be to choose VPN methods that operate above the transport layer instead of network layer VPNs. However, there are various constraints that influence the choice of VPN method and the user is usually not free to select a VPN method that fits best. Therefore, it is usually not possible to switch from a network layer VPN solution to a transport layer or application layer VPN method just to allow PEP deployment. In the following, some reasons are given for keeping a network layer VPN solution: • Higher level of security: A network layer VPN solution protects the fields of the transport layer and upper layer, e.g. port numbers. Furthermore, in case of IPsec AH, even some fields of the IP header are protected against manipulation. Changing to SSL/TLS means to weaken security. • Missing security support in applications: A network layer VPN solution is usually deployed to have a single secure channel for all applications independent of the application. Switching to a SSL/TLS-based solutions or application layer security requires having security support in all applications of interest, which may not be given. • Client/server model: TLS/SSL and also some application specific security features are based on client/server models, where a client starts the connection with one server. IPsec does not demand for a client/server model but is based on a peer to peer relationship. Depending on the scenario in scope, a client/server model may not be usable. In summary, transport and higher layer security mechanisms are appropriate when possible to be deployed, but usually it is not possible to replace network layer VPNs by transport layer VPNs or application layer security. B.2 QoS enforcement issues in secure VPNs With TLS and with IPsec in transport mode, QoS enforcement is not affected by VPN processing. As in IPsec tunnel mode the packet's original 5-tuple flow identifier is now replaced by the one of the VPN GW, QoS enforcement of IPsec-protected packets using those fields is not possible. This is because the original IP header (including the DSCP field) is replaced by the IP header generated at the VPN GW. Figure B.6 illustrates the problem of Quality of Service (QoS) enforcement for tunnel mode IPsec-protected traffic. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 159 Figure B.6: QoS enforcement issue with IPsec in tunnel mode B.2.1 Possible solutions B.2.1.1 Copying DSCP field from inner to outer header In the construction of the outer IP header, [i.53] specifies that the contents of DS field in the inner header should be copied to the outer header of a tunnel mode IPsec packet. It is applicable for both IPv4 and IPv6. Figure B.7 illustrates this process for ESP in tunnel mode. IP TCP/ UDP Payload Payload Outer IP header ESP header ESP Trailer ESP ICV encrypted Inner headers D S C P D S C P IP TCP/ UDP D S C P Figure B.7: Copying of DSCP value from the inner to the outer header In order to classify packets based on the DSCP field, the inner DSCP field should be marked as close to the traffic source as possible, such that the correct value is visible to the VPN GW for further mapping to the outer header. This could either be directly at the end node/application or at the VPN GW before the packet is protected by IPsec. For example, the user VoIP application sets the DSCP value that corresponds to DiffServ Assured Forwarding (AF) traffic class (see [i.30]). Alternatively, the VPN GW could also set this field before it applies IPsec processing to the packet. As the outer DSCP value reflects the original (the one of the inner header), it in turn reflects the intended QoS treatment of the packet. The QoS enforcement point at the hub or at the remote satellite terminal can then use the DSCP value to classify packets and deliver them into the proper transmission queues. Some remarks are worth noting: • Although this solution is mandated by IPsec, it might be undesirable in certain scenarios due to the security requirements. Copying the DSCP to the outer header means disclosing some information on the traffic flow characteristic, and thus potentially enabling a malicious party to perform traffic-analysis-based attacks. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 160 • Because the solution has to be implemented at the user end-points (either at the user terminal or at the VPN GW), its implementation requires a control case 1. • The mapping between the PHB / QoS policy and the DSCP value has to be agreed between the satellite operator and the end user. This can be achieved either by using the standard-defined DSCP values and PHB, or through a dedicated Service Level Agreement (SLA) established between the satellite operator and the user. • For packets that have to pass through the Internet before arriving at the satellite operator's QoS enforcement point, it is important to ensure that none of the intermediate router modifies the DS field such that it would cause a different treatment of the packet in the satellite link. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 161 Annex C: Impact of random access on TCP behaviour C.1 TCP delay variation and packet misordering A system that switches traffic flows from RA to DA channel could result in a change of the delay, or introduce variation of delay. This is especially the case for size-based network queuing. These considerations are most important for A-DAMA Top-Up and Back-Up use of RA with DAMA, since it is in these cases that the traffic may be divided between multiple physical layer transmission queues. Sudden changes in delay could adversely impact the TCP RTT measurements, potentially resulting in expiry of the RTO and hence an unwanted congestion response. This is not expected to be a significant effect when using a modern TCP implementation. Where the change in delay is not accompanied by loss, the effect of spurious retransmissions may be reduced using methods such as the Eifel algorithms [i.54], [i.55] in the TCP sender or Forward RTO-recovery [i.56]. The impact of delay variation depends on the application. Most TCP applications, such as web browsing, are tolerant to small (<RTT) delay variations [i.57]. On the other hand, performance of real time applications such as VoIP can be affected by delay variation [i.58], resulting in loss at the input to a speech codec, or adversely impacting the round-trip estimator used to scale the playout buffer. The switch of traffic (or unbalanced loading) from a RA to DA channel could result in reordering of packets at TCP receiver when the return path implements a queuing method that does not preserve per-flow order. Such methods could be motivated by the desire to use the RA channel for short packets, where it has the best possibility of reducing the queuing delay. Examples of such methods include size-based queuing algorithms, such as ACKs-first scheduling and variants of the shortest-packet first algorithm. These methods have been used in early routers to reduce the queuing delay that can result when ACKs are queued behind larger data segments sent over low capacity links. The methods can produce unusual pathologies when used with TCP, and their impact depends on the traffic pattern, as described in the examples below. In these examples size-based network queuing is considered for a path that comprises a capacity-constrained return link and high-speed forward link using standard queuing: • When the return link carries only ACKs, the forward path does not benefit from these methods. • When the return link predominantly carries ACKs with occasional Data segments and ACKs, the forward link will observe decreased delay. Bi-directional flows, but may experience slower cwnd growth (misordered ACKs in Data segments do not inflate cwnd in modern TCP). Overall there may be benefit for forward transfers, which decreases as the volume of Data increases. The return link data performance may not be appreciably impacted, since ACKs are generally much smaller than data. • When the return link predominantly carries Data segments with occasional ACKs, the forward link will benefit. But Shortest-First queuing can result in reordering of bursts when there are variable-sized Data segments, which can trigger Fast Retransmission and reduce performance. • A return link that carries only Data, the forward path does not benefit from these methods, and the return path is not significantly impacted by ACK-First queuing. The impact of Shortest-First queuing depends on the viability of Data segments, but is not recommended, since it can result in unpredictable behavior with specific applications (e.g. block-oriented data transfers that typically send full-sized segments, but periodically send small segments at the end of each block). The use of size-based queuing was common in early packet networks, but is not recommended for use in the general Internet, since it can lead to erratic application performance. Reordering can also significantly impact the opening of the cwnd at the sender, which is important to the performance of short-lived flows. Excessive reordering beyond the SuPACKThreshold (currently 3 segments) will trigger fast retransmission and fast recovery, with resulting impact on the cwnd and ssthresh, and should generally be avoided in networks supporting TCP. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 162 The effects of reordering may be mitigated by adapting the queuing algorithm to avoid simultaneous use of the RA and DA channels by a packet flow. Table C.1: Impact of integrated RA-DAMA at higher layers MAC Layer Network (IP) Layer Transport (TCP) Layer Application layer RA Channel Benefit at Higher Layers No access delay Lower Round Trip Delay (RTD) – Fast network response Faster delivery & acknowledgement of application data Better QoS performance for short interactive applications RA-DAMA Impact/Issues at Higher Layers 1. Random packet losses will occur on RA channel 1a. Lost packets are not recovered at IP layer 1a. TCP RTO mechanisms are triggered. Delay depends on initial RTO value, prompt & accurate estimation of RTT. 1a. QoS depends on how fast TCP can deliver the connection & data requests. 1b. Random packet losses cannot be differentiated from congestion losses 1b. A spurious congestion signal is triggered, affecting TCP sender initial cwnd & ssthresh values. The impact depends on how conservatively TCP responds to congestion. 1b. TCP congestion control affects QoS as small values of initial cwnd & ssthresh increase response time due to more round trips. 2. Maximum RA bitrate is low (due to high cost) compared to DAMA 2a. Packet reordering occurs if short packets are sent on RA & large packets on DAMA 2a. TCP prematurely triggers fast retransmit/fast recovery if serious reordering occurs. TCP mechanisms are available to detect spurious retransmissions. 2a. Packets are reordered by TCP. Delay & jitter components are experienced by application. 2b. Variable packet delay if large packets are switched/transmitted on RA and DAMA 2b. TCP RTT estimation may be inaccurate leading to premature RTO. The impact may be insignificant if RTT is much longer than transmission time. 2b. Some applications are sensitive to jitter (e.g. VoIP) but short interactive applications are more tolerant. C.2 Responsiveness of standard TCP The core principle for TCP congestion control is that loss of packets is regarded as a potential source of congestion. When packet loss is detected, TCP therefore activates its congestion control algorithms, as defined in [i.59]. Operating TCP over RA can result in loss of control or request packets at the beginning of a transmission. This may trigger overly conservative behavior, even though there is no congestion. In this case, response time is mostly affected by the state of RTO and IW variables, which are dynamically set according to TCP conservative principles. C.2.1 Reduced initial RTO It is recommended in [i.60] reducing the initial RTO of TCP from a previous value of 3 seconds to 1 second, unless the SYN or SYN-ACK is lost, in which case the default RTO is reverted to 3 seconds before data transmission begins. The lower RTO value was found to be sufficient for more than 97,5 % of connections, while implication of spurious retransmissions for few connections with RTT longer than 1 second is modest. More significantly, the new value is small enough to ensure timely recovery from packet losses occurring before an RTT sample is taken. Hence this new standard enhances TCP response time in case of initial packet loss. However, RTO loss recovery activates congestion control thus causing the TCP sender to be overly conservative during non-congestion periods. In particular, the following two state variables are affected: • Congestion Window (cwnd) - This is set to Loss Window. Slow start restarts with cwnd of only 1 segment. • Slow-start Threshold (ssthresh) - The ssthresh is set to around 2 segments i.e. max (FlightSize/2, 2*SMSS). C.2.2 Early loss recovery Modern TCP uses algorithms to detect and recover from loss within the shortest possible time, usually before an RTO has expired. These mechanisms include: ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 163 C.2.2.1 Fast Retransmit and Fast Recovery The Fast Retransmit/Fast Recovery algorithm allows a TCP receiver to send an immediate duplicate ACK when it receives an out-of-order segment that confirms data is held waiting for a particular byte number. The TCP sender uses the Fast Retransmit algorithm [i.59] to detect and repair loss, based on incoming ACKs. The arrival of 3 duplicate ACKs acts as an indication that a segment has been lost. Hence fast recovery of lost segment can be performed without incurring an RTO. SACK is a widely employed enhancement to this method – but has little impact on the first few packets of a flow. C.2.2.2 Limited transmit When the flight size is less than 4 segments, Fast Retransmit cannot be used, because there will never be more sufficient Dup ACKs to trigger the method. The Limited Transmit algorithm [i.61] allows an additional outstanding segment to be sent upon receiving each Dup ACK (increasing flight size). This eventually triggers Fast Retransmit, when 3 Dup ACKs may be induced after a loss. C.2.2.3 Early retransmit The Limited Transmit algorithm cannot trigger Fast Retransmit if TCP sender does not have additional outstanding segments to send up to required amount (e.g. a burst is limited to 3 segments or less, such many web page requests). To solve this problem, the Early Retransmit algorithm [i.62] calculates a new value (ER_thresh) that determines number of DUP ACKs needed to trigger Fast Retransmit based on outstanding unsent data. In the case of a burst of 3 segments, this method reduces the number of DUP ACKs required to trigger Fast Retransmit to only 2. C.3.3 Redundant TCP SYNs The TCP standard specifies sending a single initial SYN packet and waiting for an ACK. The SYN is only retransmitted after the RTO period, when a loss is assumed thus delaying actual start of data transmission. SYN duplication is a proposed technique that could improve TCP responsiveness when the initial SYN packet is lost. One way to achieve this is by setting initial RTO smaller than the actual path RTT. It has been argued that since general-purpose networks are designed for large traffic flows, it is reasonably safe to be aggressive when sending short flows [i.63]. The RTO retransmit timer can be set low e.g. 100ms, or even less, if packets are not too close together to share the same fate. (In doing so, it is important to verify that the implementation does not reset ssthresh when performing a SYN retransmission.) In a satellite system, duplication of the initial SYN could save time at the expense of using an additional transmission burst (e.g. in RA channel). However the overhead of SYN duplication may not be a significant, because the additional SYN packet is only 40B (without compression). In general, TCP responsiveness is affected primarily by how fast a client is able to deliver the connection and data requests. Thus, additional redundancy should be employed for duplicated SYN or request packets rather than subsequent confirmations. Delayed ACK or other proposed ACK Congestion Control mechanisms [i.64] may offset the possible extra load due on the RA channel. There are however concerns with resending a SYN, since some clients use this as method for detecting whether the end host supports a particular function. For example, only the initial SYN may carry some TCP options, and loss of this SYN could significantly change the operation of the remaining connection. One notable example is dual-stack systems, where an IPv6 sender may revert to IPv4 for the second SYN, since it is assumed that the server failed to respond to the initial network layer request. Care should be taken to avoid such effects impacting the user. C.3.4 Changing TCP RTT/RTO estimation [i.65] specifies a method that does not sample the RTT during the three-way handshake (3WHS) when using a large IW, because delay changes can result once a session is established. One example is when there is a significant time to serialize a data packet on a narrowband link, where seeding the RTO based on an RTT of a small SYN or SYN-ACK packets would likely underestimate the RTT for larger data packets. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 164 A proposal from Google [i.66] recommends sampling the RTT during the 3WHS and seeding the RTO regardless of the size of the IW. The main reasoning for this proposed reversal of practice is the prominence of faster links in the Internet suffering noticeable latency while waiting for an RTO compared to the benefits of a shorter RTT. Seeding the RTO with correct RTT sampled after SYN and SYN-ACK exchange has been suggested to improve TCP responsiveness in the case of loosing subsequent packets during the handshake. However, it is important to note that this proposal was made before [i.60] became the standard for computing TCP retransmission timer. Reduction of the initial RTO from 3 seconds to 1 second may reduce the urgency of this particular proposal. The proposal could still be useful for links with very short RTT, but could raise issues on slow links or links that rely on DA or RA methods. Further research is required to judge the applicability to the general Internet. C.3.5 Sending data with TCP SYN TCP Fast Open (TFO) [i.67] is another proposal that would allow data to be carried in the SYN or SYN-ACK packets and consumed by the receiving end during the initial connection handshake. This provides a saving of up to one full RTT compared to standard TCP, requiring a 3WHS to complete before data can be exchanged. Data on SYN behavior was allowed in [i.68] but TFO would additionally allow data to be delivered to the application before the 3WHS has completed. In the proposed method, the server side uses a security cookie to authenticate a client initiating a TFO connection thus addressing previous data integrity concerns caused by dubious SYN packets. This avoids the pitfalls of earlier methods, such as T/TCP. However, it requires an additional exchange between client and server at the beginning of a connection for requesting the fast open cookie, which should also be expired by the server after some time. TFO is somewhat limited, as it is more applicable for applications that have temporal locality on client and server connections. There are concerns with sending data on SYN such as a client choosing IP version (IPv6 or IPv4) that is not supported at the server, or starting with an unknown size of the Maximum Segment Size (MSS) for the link. Additionally, there is no sequence number protection hence the packet is more vulnerable to attacks. For the moment this remains a topic of research, if accepted this proposal would significantly increase the size of a TCP SYN, which may impact usage of the RA channel. C.3.6 Increasing TCP Initial Window An increase of IW from 1 to 3 segments has been widely deployed, motivated by the desire to improve Fast Retransmit. A recent proposal from Google argues for increasing IW further to at least ten segments (about 15KB) for speedy completion of short TCP transfers in one RTT. Furthermore, reduction in total transfer time for data greater than 4KB up to 4 RTTs is possible. Preliminary experiments by Google show benefits in reducing object transfer times at moderate cost in terms of increased congestion and associated packet losses. This analysis did not explore the potential collateral damage on other flows that share a bottleneck where the large IW is continuously used. Google has also recommended that TCP implementations refrain from resetting IW to one segment unless there have been multiple SYN or SYN-ACK retransmissions, or true loss detection has been made [i.62]. The current standard [i.65] specifies resetting IW to 1 on loosing even a single control packet. However, considering [i.60] reduction of initial RTO from 3 seconds to 1 second, it is possible to unnecessarily penalize connections with high RTT values (e.g. satellite links). A key argument to be assessed is that there is little or no experience of using a larger IW on other flows that share a constrained path. The likely impact on real-time flows (voice, video) may be significant if many flows use a larger IW. This is therefore an area of current research, and a topic where standards are expected within the TCPM working group of the IETF. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 165 Table C.2: Recent proposals to enhance responsiveness of standard TCP TCP Mechanism Standard RFC Proposed Enhancement Retransmission Timeout (RTO) RTO not seeded during three-way handshake [i.60] Google (Seeding RTO with RTT sampled during three-way handshake) Initial Congestion Window (IW) Maximum initial window of 3 segments [i.65] Google (Increasing TCP IW from 3 to 10 segments) Loss Window (LW) Reduce IW to 1 segment on loss of packet during three-way handshake [i.65] Google (Refrain from resetting IW to LW upon loss of packet during the three-way handshake) Initial SYN control packet TCP sender sends one initial SYN to start connection [i.68] Damon Wischik (Setting initial RTO smaller than RTT e.g. to duplicate SYN) ACK control Delayed ACK [i.59] [i.64] (ECN-marked ACK packets) Data on SYN [i.68] forbids the receiver to deliver the data to the application until 3WHS is completed. Google TCP Fast Open (allows the receiver to deliver the data to the application during 3WHS) ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 166 History Document history V1.1.1 April 2014 Publication |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 1 Scope | The present document provides implementation guidelines for equipment and systems intended to comply with [i.1]. It also provides designs that may be used to supplement the normative specifications provided in [i.1]. Such designs could evolve into being a part of the normative specifications in the future. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 2 References | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 2.1 Normative references | Normative references are not applicable in the present document. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 2.2 Informative references | References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies. NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee their long-term validity. The following referenced documents may be useful in implementing an ETSI deliverable or add to the reader's understanding, but are not required for conformance to the present document. [i.1] ETSI EN 301 545-2: "Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 2: Lower Layers for Satellite standard". [i.2] ETSI EN 302 307-1: "Digital Video Broadcasting (DVB); Second generation framing structure, channel coding and modulation systems for Broadcasting, Interactive Services, News Gathering and other broadband satellite applications; Part 1: DVB-S2". [i.3] ETSI TS 101 545-1: "Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 1: Overview and System Level specification". [i.4] ETSI TS 101 545-3: "Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 3: Higher Layers Satellite Specification". [i.5] Guidelines for 64-bit global identifier (EUI-64) Registration Authority. [i.6] ETSI EN 301 459: "Satellite Earth Stations and Systems (SES); Harmonised Standard for Satellite Interactive Terminals (SIT) and Satellite User Terminals (SUT) transmitting towards satellites in geostationary orbit, operating in the 29,5 GHz to 30,0 GHz frequency bands covering the essential requirements of article 3.2 of the Directive 2014/53/EU". [i.7] ETSI EN 301 428: "Satellite Earth Stations and Systems (SES); Harmonised Standard for Very Small Aperture Terminal (VSAT); Transmit-only, transmit/receive or receive-only satellite earth stations operating in the 11/12/14 GHz frequency bands covering the essential requirements of article 3.2 of Directive 2014/53/EU". [i.8] ETSI EN 301 427: "Satellite Earth Stations and Systems (SES); Harmonised Standard for low data rate Mobile satellite Earth Stations (MES) except aeronautical mobile satellite earth stations, operating in the 11/12/14 GHz frequency bands covering the essential requirements of article 3.2 of the Directive 2014/53/EU". [i.9] ETSI EN 302 186: "Satellite Earth Stations and Systems (SES); Satellite mobile Aircraft Earth Stations (AESs) operating in the 11/12/14 GHz frequency bands; Harmonised Standard for access to radio spectrum". ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 15 [i.10] ETSI EN 302 340: "Satellite Earth Stations and Systems (SES); Harmonised Standard for satellite Earth Stations on board Vessels (ESVs) operating in the 11/12/14 GHz frequency bands allocated to the Fixed Satellite Service (FSS) covering the essential requirements of article 3.2 of the Directive 2014/53/EU". [i.11] ETSI EN 302 448: "Satellite Earth Stations and Systems (SES); Harmonised Standard for tracking Earth Stations on Trains (ESTs) operating in the 14/12 GHz frequency bands covering the essential requirements of article 3.2 of the Directive 2014/53/EU". [i.12] ETSI EN 302 977: "Satellite Earth Stations and Systems (SES); Harmonised Standard for Vehicle-Mounted Earth Stations (VMES) operating in the 14/12 GHz frequency bands covering the essential requirements of article 3.2 of the Directive 2014/53/EU". [i.13] FCC Part 25-Satellite Communications, § 25.221: "Blanket Licensing provisions for Earth Stations on Vessels (ESVs) receiving in the 3700-4200 MHz (space-to-Earth) frequency band and transmitting in the 5925-6425 MHz (Earth-to-space) frequency band, operating with Geostationary Satellites in the Fixed-Satellite Service. § 25.222 Blanket Licensing provisions for Earth Stations on Vessels (ESVs) receiving in the 10.95-11.2 GHz (space-to-Earth), 11.45-11.7 GHz (space-to-Earth), 11.7-12.2 GHz (space-to-Earth) frequency bands and transmitting in the 14.0-14.5 GHz (Earth-to-space) frequency band, operating with Geostationary Satellites in the Fixed-Satellite Service". [i.14] Recommendation ITU-R M.1643: "Technical and operational requirements for aircraft earth stations of aeronautical mobile-satellite service including those using fixed-satellite service network transponders in the band 14-14.5 GHz (Earth-to-space)". [i.15] ETSI EN 301 358: "Satellite Earth Stations and Systems (SES); Satellite User Terminals (SUT) using satellites in geostationary orbit operating in the 19,7 GHz to 20,2 GHz (space-to-earth) and 29,5 GHz to 30 GHz (earth-to-space) frequency bands". [i.16] ETSI EN 301 489-12: "ElectroMagnetic Compatibility (EMC) standard for radio equipment and services; Part 12: Specific conditions for Very Small Aperture Terminal, Satellite Interactive Earth Stations operated in the frequency ranges between 4 GHz and 30 GHz in the Fixed Satellite Service (FSS); Harmonised Standard for ElectroMagnetic Compatibility". [i.17] ETSI ETS 300 784: "Satellite Earth Stations and Systems (SES); Television Receive-Only (TVRO) satellite earth stations operating in the 11/12 GHz frequency bands". [i.18] EN 61319-1: "Interconnections of satellite receiving equipment Part 1: Europe", (produced by CENELEC). [i.19] EN 50083 series: "Cable networks for television signals, sound signals and interactive services", (produced by CENELEC). [i.20] DiSEqC Bus Specification, Version 4.2, EUTELSAT: "DiSEqC Bus Specification". [i.21] NIST Special Publication 800-38A, 2001 edition: "Recommendation for Block Cipher Modes of Operation - Methods and Techniques". [i.22] IETF RFC 4945: "The Internet IP Security PKI Profile of IKEv1/ISAKMP, IKEv2, and PKIX". [i.23] IETF RFC 4880: "OpenPGP Message Format". [i.24] Recommendation ITU-T X.509: "Information technology - Open Systems Interconnection - The Directory: Public-key and attribute certificate frameworks", August 2005 and later corrigenda". [i.25] Rivest, R., A. Shamir, L. Adleman: "A Method for Obtaining Digital Signatures and Public-Key Cryptosystems", Communications of the ACM 21 (2): 120-126. [i.26] IEEE 802.3™: "Information technology - Telecommunications and information exchange between systems - Local and metropolitan area networks - Specific requirements - Part 3: Carrier Sense Multiple Access with Collision Detection (CSMA/CD) Access Method and Carrier Sense Multiple Access with Collision Detection (CSMA/CD) Access Method and Physical Layer Specifications". [i.27] IETF RFC 5911: "New ASN.1 Modules for Cryptographic Message Syntax (CMS) and S/MIME". ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 16 [i.28] ETSI TS 101 162: "Digital Video Broadcasting (DVB); Allocation of identifiers and codes for Digital Video Broadcasting (DVB) systems". [i.29] George Marsaglia: "Xorshift RNGs", Journal of Statistical Software (Vol.8 Issue 14), July 2003. [i.30] IETF RFC 3629: "UTF-8, a transformation format of ISO 10646", November 2003. [i.31] RSA Laboratories: "PKCS #5 v2.1: Password-Based Cryptography Standard", October 5, 2006. [i.32] National Institute of Standards and Technology: "Secure Hash Standard", FIPS Pub. 180-2, August 1, 2002. [i.33] IETF RFC 4648: "The Base16, Base32, and Base64 Data Encodings". [i.34] Federal Information Processing Standards Publication 140-2: "Security Requirements for Cryptographic Modules", May 25, 2001 and later amendments. [i.35] Recommendation ITU-R P.1623-1: "Prediction method of fade dynamics on Earth-space paths", 2005-03r. 1988. [i.36] ECC/DEC(06)02 (2006): "Electronic Communications Committee, ECC Decision of 24 March 2006 on Exemption from Individual Licensing of low e.i.r.p. satellite terminals (LEST) operating within the frequency bands 10.70 - 12.75 GHz or 19.70 - 20.20 GHz Space-to-Earth and 14.00 - 14.25 GHz or 29.50 - 30.00 GHz Earth-to-Space". [i.37] ECC/DEC(06)03 (2006): "Exemption from Individual Licensing of High e.i.r.p. satellite terminals (HEST) operating within the frequency bands 10.70 - 12.75 GHz or 19.70 - 20.20 GHz Space-to- Earth and 14.00 - 14.25 GHz or 29.50 - 30.00 GHz Earth-to-Space". [i.38] Rapp: "Effects of the HPA-nonlinearity on a 4-DPSK/OFDM signal for a digital sound broadcasting system", in: Second European Conf. on Sat. Comm., 22. - 24.10.91, Liege, Belgium, 1991. [i.39] B. E. Rimoldi: "A decomposition approach to CPM", IEEE™ Trans. Inform. Theory, vol. 34, pp. 260-270, March 1988. [i.40] A. Barbieri and G. Colavolpe: "Simplified soft-output detection of CPM signals over coherent and phase noise channels", IEEE™ Trans. Wireless Commun., vol. 6, pp. 2486-2496, July 2007. [i.41] L. Bahl, J. Cocke, F. Jelinek and J. Raviv: "Optimal decoding of linear codes for minimizing symbol error rate", IEEE™ Trans. Info. Theory, vol. IT-20, no. 2, pp. 284-287, March 1974. [i.42] D.C. Rife and R.R. Boorstyn: "Single tone parameter estimation from discrete-time observations", IEEE™ Trans. Inform. Theory, vol. 20, pp. 591-598, September 1974. [i.43] Directive 2014/53/EU of the European Parliament and of the Council of 16 April 2014 on the harmonisation of the laws of the Member States relating to the making available on the market of radio equipment (RED). [i.44] Ichikawa M., Hara T., Saito M., Okada M. and Yamamoto H.: "Evaluation of BER degradation due to phase nonlinearity of SSPA on MC-CDMA signals", Personal Wireless Communications, 2005. ICPWC 2005. Page(s): 533 - 536. [i.45] M. Angelone, A. Ginesi, E. Re and S. Cioni: "Performance of a Combined Dynamic Rate Adaptation and Adaptive Coding Modulation Technique for a DVB-RCS2 System", Proceedings of ASMS/SPSC Conference 2012, Baiona, Spain, Pages 124-131. [i.46] ETSI EN 301 790: "Digital Video Broadcasting (DVB); Interaction channel for satellite distribution systems". [i.47] ETSI TR 101 790 (V1.4.1): "Digital Video Broadcasting (DVB); Interaction channel for Satellite Distribution Systems; Guidelines for the use of EN 301 790". [i.48] M. Holzbock, A. Jahn, O. Gremillet, E. Lutz: "Aeronautical channel characterization measurements at K Band", in Proceedings 4th Ka Band Utilization Conference", Venice, Italy, pp. 263-269, November 1998. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 17 [i.49] S. Scalise, H. Ernst and G. Harles: "Measurement and Modeling of the Land Mobile Satellite Channel at Ku-Band", IEEE™ Transaction on Vehicular Technology, Vol. 57, No. 2 March 2008. [i.50] E. Kubista, F. Perez Fontan, M. A. Vazquez Castro, S. Buonomo, B. R. Arbesser-Rastburg, J.P.V. Poiares Baptista: "Ka-Band Propagation Measurements and Statistics for Land Mobile Satellite Applications", IEEE™ Transaction on Vehicular Technology, Vol. 49, No. 3 May 2000. [i.51] F. Perez Fontan, M. Vazquez Castro, C. Enjamio Cabado, J. Pita Garcia, and E. Kubista, "Statistical modelling of the LMS channel", IEEE™ Transactions on Vehicular Technology, vol. 50, pp. 1549-1567, November 2001. [i.52] P.A. Laurent: "Exact and approximate construction of digital phase modulations by superposition of amplitude modulated pulses (AMP)", IEEE™ Trans. Commun., vol. 34, pp. 150-160, February 1986. [i.53] A. Barbieriand, G. Colavolpe: "Simplified Soft-Output Detection of CPM Signals Over Coherent and Phase Noise Channels", IEEE™ Trans. on Wireless Comm., vol. 6, n. 7, July 2007. [i.54] G. Colavolpe and R. Raheli: "Reduced-complexity detection and phase synchronization of CPM signals", IEEE™ Trans. Commun., vol. 45, pp. 1070-1079, September 1997. [i.55] N. Abramson: "The Throughput of Packet Broadcasting Channels", IEEE™ Trans. Communications, vol. COM-25, no. 1, pp. 117-128, January 1977. [i.56] G. L. Choudhury and S. S. Rappaport: "Diversity ALOHA - A Random Access Scheme for Satellite Communications", IEEE™ Trans. on Comm., vol. COM-31, March 1983, pp. 450-457. [i.57] E. Casini, R. De Gaudenzi and O. del Rio Herrero: "Contention Resolution Diversity Slotted Aloha (CRDSA): an Enhanced Random Access Scheme for Satellite Access Packet Networks", IEEE™ Transactions on Wireless Communications, vol. 6, no. 4, pp. 1408-1419, April 2007. [i.58] Oscar del Rio Herrero and Riccardo De Gaudenzi: "A High-Performance MAC Protocol for Consumer Broadband Satellite Systems", In the Proc. of 27th AIAA International Communications Satellite Systems Conference, June 1-4 2009, Edinburgh (United Kingdom). [i.59] R. De Gaudenzi and O. Del Rio-Herrero: "Advances in Random Access protocols for satellite networks", International Workshop on Satellite and Space Communications, 2009, IWSSC 2009, Siena, Italy September 9-11, 2009, pp. 331-336. [i.60] ESA Study: "Advanced Modem Prototype for Interactive Satellite Terminals". [i.61] Del Rio Herrero and R. De Gaudenzi, US Patent 7,990,874: "Methods, Apparatuses and System for Asynchronous Spread Spectrum Communication", August 2, 2011. [i.62] G. Liva: "Graph-Based Analysis and Optimization of Contention Resolution Diversity Slotted ALOHA", IEEE™ Trans. On Comm., Vol. 59, Issue 2, February 2011, pp. 477-487. [i.63] Final Report: "Motorised antenna mount for bi-directional satellite terminals", ESA Study Contract No. 21632/08/NL/AD. [i.64] ETSI TS 136 211: "LTE; Evolved Universal Terrestrial Radio Access (E-UTRA); Physical channels and modulation (3GPP TS 36.211 version 8.9.0 Release 8)". [i.65] Digital Video Broadcasting (DVB) - Next Generation Handheld (NGH): "Frame structure channel coding and modulation". [i.66] H. G. Myung, J. Lim and D. J. Goodman: "Single carrier FDMA for uplink wireless transmission", IEEE™ Vehicular Tech. Magazine, pp. 30-38, June 2009. [i.67] M. Morelli and U. Mengali: "An improved frequency offset estimator for OFDM applications," IEEE™ Commun. Letters, Vol. 3, No. 3, Mar. 1999, pp. 75-77. [i.68] Giambene Giovanni (Ed.): "Resource Management in Satellite Networks Optimization and Cross-Layer Design", 2007, Springer. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 18 [i.69] Recommendation ITU-T G.1030 (2005): "Estimating end-to-end performance in IP networks for data applications". [i.70] ANSI/IEEE Standard 754™ (1985): "IEEE Standard for Binary Floating-Point Arithmetic". [i.71] IETF RFC 4326: "Unidirectional Lightweight Encapsulation (ULE) for Transmission of IP Datagrams over an MPEG-2 Transport Stream (TS)". [i.72] IETF RFC 5163: "Extension Formats for Unidirectional Lightweight Encapsulation (ULE) and the Generic Stream Encapsulation (GSE)". [i.73] Recommendation ITU-R S.728.1: "Maximum permissible level of off-axis e.i.r.p. density from very small aperture terminals (VSATs)". [i.74] ETSI EN 303 978: "Satellite Earth Stations and Systems (SES); Earth Stations on Mobile Platforms (ESOMP) communicating with satellites in geostationary orbit, operating in the 27,5 GHz to 30,0 GHz and 17,3 GHz to 20,2 GHz frequency bands; Harmonised Standard for access to radio spectrum". [i.75] ETSI TS 102 602: "Satellite Earth Stations and Systems (SES); Broadband Satellite Multimedia; Connection Control Protocol (C2P) for DVB-RCS; Specifications". [i.76] DVB BlueBook C107 (2023-07): "Commercial Requirements for the use of DVB-RCS2 in Geostationary and Non-Geostationary Systems". [i.77] ETSI TR 102 376-2: "Digital Video Broadcasting (DVB); Implementation guidelines for the second generation system for Broadcasting, Interactive Services, News Gathering and other broadband satellite applications; Part 2: S2 Extensions (DVB-S2X)". [i.78] ETSI TS 103 179 (V1.2.1) (2023-06): "Satellite Earth Stations and Systems (SES); Return Link Encapsulation (RLE) protocol". [i.79] ETSI TS 102 606-1: "Digital Video Broadcasting (DVB); Generic Stream Encapsulation (GSE); Part 1: Protocol". [i.80] ETSI EN 302 307-2 (V1.4.1) (2024-02): "Digital Video Broadcasting (DVB); Second generation framing structure, channel coding and modulation systems for Broadcasting, Interactive Services, News Gathering and other broadband satellite applications; Part 2: DVB-S2 Extensions (DVB-S2X)". [i.81] Consultative Committee for Space Data Systems: Recommended Standard CCSDS 502.0-B-3: "Orbit Data Messages". [i.82] CelesTrak®. [i.83] ETSI EN 300 468: "Digital Video Broadcasting (DVB); Specification for Service Information (SI) in DVB systems". [i.84] ETSI TS 102 606-2: "Digital Video Broadcasting (DVB); Generic Stream Encapsulation (GSE); Part 2: Logical Link Control (LLC)". [i.85] Ruben Morales-Ferre, Elena Simona Lohan, Gianluca Falco, Emanuela Falletti: "GDOP-based analysis of suitability of LEO constellations for future satellite-based positioning", 2020 IEEE™ International Conference on Wireless for Space and Extreme Environments (WiSEE). [i.86] Fabricio S. Prol et al.: "Position, Navigation, and Timing (PNT) Through Low Earth Orbit (LEO) Satellites: A Survey on Current Status, Challenges, and Opportunities", IEEE™ Access, August 2022. [i.87] N. Levanon: "Quick position determination using 1 or 2 LEO satellites", IEEE™ Transactions on Aerospace and Electronic Systems, vol. 34(3), pp.736-754, 1998. [i.88] N. Levanon: "Instant active positioning with one LEO satellite", Journal of the Institute of Navigation, vol. 46, pp. 87-95, 1999. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 19 [i.89] P. Delbeke, D. Duyck: "5G-NR GNSS independent time and frequency synchronization in NTN scenarios", Ka Conference 2023, Bradford, UK, October 2023. [i.90] P. Delbeke, B. Reynders, D. Duyck: "GNSS-Independent 5G-NR Tracking for Non Terrestrial Networks", Ka Conference 2024, Seattle, WA, USA, September 2024. [i.91] A. Freedman, L. Erup, D. Peilow, P. Nayler, V. Mignone: "DVB Standard Support of NGSO Systems", Ka Conference 2024, Seattle, WA, USA, September 2024. [i.92] H. Al-Hraishawi, H. Chougrani, S. Kisseleff, E. Lagunas, S. Chatzinotas: "A Survey on Nongeostationary Satellite Systems: The Communication Perspective", IEEE™ Communications Surveys & Tutorials, vol. 25, no. 1, first quarter 2023. [i.93] N. Pachler, I. del Portillo, E. F. Crawley, B. G. Cameron: "An Updated Comparison of Four Low Earth Orbit Satellite Constellation Systems to Provide Global Broadband", 2021 IEEE™ International Conference on Communications Workshops (ICC Workshops), 4-23 June 2021. [i.94] K. Dakic, C. C. Chan, B. Al Homssi, K. Sithamparanathan, and A. Al-Hourani: "On Delay Performance in Mega Satellite Networks with Inter-Satellite Links", IEEE™ Global Communications Conference, GLOBECOM 2023. [i.95] D. Duyck et al.: "Demonstrating end to end standards-based beam-hopping with commercial equipment derisking the physical layer challenges", Ka Conference 2023, Bradford, UK, October 2023. [i.96] Recommendation ITU-R P.618: "Propagation data and prediction methods required for the design of Earth-space telecommunication systems". [i.97] Inigo del Portillo: "ITU-Rpy: A python implementation of the ITU-R P. Recommendations to compute atmospheric attenuation in slant and horizontal paths", 2017. [i.98] A. Freedman, L. Erup, F. Díaz Canales, D. Peilow, P. Nayler, V. Mignone, K. Kaario, T. Huikko, V. Rönty: "DVB Standard Support of NGSO Systems", International Journal of Satellite Communication and Networks, 2025. [i.99] Weiwen Liu D. G. Michelson: "Fade Slope Analysis of Ka-Band Earth-LEO Satellite Links Using a Synthetic Rain Field Model", IEEE™ Transactions on Vehicular Technology (Vol. 58 issue 8), October 2009. [i.100] TM-S0593r2: "Simulation Report on Technology Evaluation: DVB-S2X Beam Hopping in NGSO Systems". |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 3 Definition of terms, symbols and abbreviations | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 3.1 Terms | For the purposes of the present document, the terms given in ETSI EN 301 545-2 [i.1] apply. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 3.2 Symbols | For the purposes of the present document, the following symbols apply: α Roll-off factor A, B Input sequences to the turbo encoder C1 Circulation state of the turbo encoder in the natural order C2 Circulation state of the turbo encoder in the interleaved order Eb/N0 Ratio between the energy per information bit and single sided noise power spectral density Es/N0 Ratio between the energy per transmitted symbol and single sided noise power spectral density f0 Carrier frequency fN Nyquist frequency ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 20 H(f) Raised Cosine filters frequency transfer function I, Q In-phase, Quadrature phase components of the modulated signal K/N GSPC code rate Nb GSPC sub-blocks number NR,max Number of replicas in a frame Nrand 12-bit random number used as a random seed value during CRDSA frame decoding Nslots Number of the slots in the frame p1, p2, …, pNR,max Vector that contains the NR,max indices of the slots containing the burst replicas 0 1 , , p p j d L − GSPC code parity bits R, k/n Burst code rate Rs Symbol rate corresponding to the bilateral Nyquist bandwidth of the modulated signal S State of the turbo encoder Sx Symbol Ts Symbol period ux Bits X GSPC code information word X(D) GSPC code information polynomial 0 1 , , x xK L − GSPC code information bits Z1 Output sequence of the puncturing for the encoder in the natural order Z2 Output sequence of the puncturing for the encoder in the interleaved order |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 3.3 Abbreviations | For the purposes of the present document, the following abbreviations apply: 16QAM 16-ary QAM 8PSK 8-ary PSK AC Allocation Channel ACC Acquisition Ciphertext Channel ACI Adjacent Channel Interference ACK ACKnowledgement ACM Adaptive Coding and Modulation AES Advanced Encryption Standard AF Assured Forwarding ALPDU Addressed Link Protocol Data Unit AM/AM Amplitude Modulated to Amplitude Modulated characteristic (Power Transfer) AM/PM Amplitude Modulation/ Phase Modulation (Phase Transfer) AMSS Aeronautical Mobile Satellite Service ANT Antenna (Subsystem) APSK Amplitude and Phase Shift Keying (modulation) ASCII American Standard Code for Information Interchange ASIC Application Specific Integrated Circuit ATM Asynchronous Transfer Mode AVBDC Absolute VBDC AWGN Additive White Gaussian Noise BA Behaviour Aggregate BB Base-Band BBFRAME BaseBand FRAME BCH Bose - Chaudhuri - Hocquenghem (code) BCJR Bahl, Cocke, Jelinek and Raviv (algorithm) BCT RL Broadcast Configuration Table BE Best Effort BHTP Beam Hopping Time Plan BPSK Binary PSK bslbf bit string, left bit first BTP Burst Time Plan BTP Burst Time Plan BTU Bandwidth-Time Unit BUC Block Up Converter ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 21 CA Connectivity Aggregate CA/KG Certificate Authority/ Key Generator CAZAC Constant Amplitude Zero Auto Correlation CBC Cipher Block Chaining CBR Constant Bit Rate CC Convolutional Coding CC-CPM Convolutional Code - Continuous Phase Modulation CCITT (The) International Telegraph and Telephone Consultative Committee CCM Constant Coding and Modulation CDF Cumulative Distribution Function CEPT European Conference of Postal and Telecommunications Administrations CFB Cipher Feedback CMF Control and Monitoring Functions CMOS Charge-coupled Metal Oxide Silicon CNR Carrier to Noise power ratio CP Cyclic Prefix CPE Continuous Phase Encoder CPM Continuous Phase Modulation (or Modulator) CR Capacity Request CRA Constant Rate Assignment CRC Cyclic Redundancy Check CRDSA Contention Resolution DSA CW Continuous Wave DA Dedicated Access DA-AC Dedicated Access Allocation Channel DAAF Digital Anti-Aliasing Filter DAC Digital to Analog Converter DAMA Demand Assigned Multiple Access DC Direct Current DCC Dynamic Ciphertext Channel DCP Dynamic Connectivity Protocol DFL Data Field Length DFT Discrete Fourier Transform DNS Domain Name Server DOP Dilution Of Precision DRA Data Rate Adaptation DSA Diversity Slotted Aloha DSCP Differentiated Service Code Point DSSS Direct Sequence Spectrum Spreading DVB Digital Video Broadcasting DVB-RCS2 Digital Video Broadcast Return Channel via Satellite 2nd generation ECB Electronic Code Book ECC Electronic Communication Committee (of the CEPT) EF Expedited Forwarding EIRP Effective Isotropic Radiated Power EMC ElectroMagnetic Compatibility ESAA Electronically Steerable Antenna Array ESV Earth Stations (on board) Vessels FCA Free Capacity Assignment FCC Federal Communication Commission FCT2 Frame Configuration Table 2 FEC Forward Error Correction FER Frame Error Ratio FFT Fast Fourier Transform FIR Finite Impulse Response FL Forward Link FLS Forward Link Signalling FPDU Frame PDU FPGA Field-Programmable Gate Array FRSS Frame Repetition Spectrum Spreading FS Fixed Service FS Fixed Service ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 22 FSS Fixed Satellite Service GPS Global Positioning Systems GS Generic Stream GSE Generic Stream Encapsulation GSO Geostationary Satellite Orbit GSPC Generic Sub-block Polynomial Code GW GateWay HL Higher Layer HLC Higher Layer Classification HLS Higher Layer Specifications HSM Hub Security Module HW HardWare IANA Internet Assigned Numbers Agency IB Installation Burst IBO Input Back Off ICT Interleaver Configuration Table ID IDentifier IDU Indoor Unit IERS International Earth Rotation and Reference System Service IF Intermediate Frequency IFDMA Interleaved Frequency Division Multiple Access IFFT Inverse Fast Fourier Transform IFL Inter-Facility Link ISI Input Stream Identifier ISL Inter-Satellite Links ISSYI Input Stream SYnchronizer Indicator IV Initialization Vectors KEK Key Encrypting Keys L2S Lower Layer Signalling LAN Local Area Network LDPC Low Density Parity Check LFDMA Localized Frequency Division Multiple Access LL Link Layer LL Lower Layer LLC Lower Layer Classification LL-FEC Link Layer Forward Error Correction LLS (DVB-RCS2) Lower Layer Specification LM Linear Modulation (or Modulator) LMSS Land Mobile Satellite Service LNB Low Noise Block LO Local Oscillator LOS Line Of Sight LPF Low Pass Filtering LSB Least Significant Bit LUT Look-up Table MAC24 24 bit MAC address MAP Maximum A Posteriori MATYPE Mode Adaptation TYPE MCD Multi Carrier Demodulator MECH MECHanical subsystem MES Mobile satellite Earth Stations MF-TDMA Multi-Frequency TDMA MIB Management Information Base MM Memoryless Modulator MMSS Maritime Mobile Satellite Service MODCOD MODulation and CODing MPEG Moving Pictures Expert Group MSB Most Significant Bit MSS Mobile Satellite Service MTU Maximum Transmission Unit NACK Negative ACKnowledgment NCC Network Control Centre ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 23 NCC/GW Network Control Centre/GateWay NCR Network Clock Reference NGSO Non Geostationary Satellite Orbit NIT Network Information Table NIU Network Interface Unit NLOS Non-Line-Of-Sight mobile NMC Network Management Centre NMS Network Management System NOC Network Operator Centre NPD Null Packet Deletion NRM Network Resource Model NSC Network Security Controller NTP Network Time Protocol OBO Output Back-Off OBP On Board Processing ODU OutDoor Unit OFDMA Orthogonal Frequency Division Multiple Access PAM Pulse Amplitude Modulation PAPR Peak to Average Power Ratio PCI Peripheral Component Interconnect PCR (MPEG-2) Program Clock Reference PDU Protocol Data Unit PER Packet Error Ratio PFD Power Flux Density PHB Per Hop Behaviour PHY Physical (Layer) PL PHYsical Layer PLL Phase Lock Loop PLR Packet Loss Ratio PNT Positioning, Navigation and Timing PPDU Payload-adapted PDU PRG Pseudo-Random Generator PSD Power Spectral Density PSI Program Specific Information PSK Phase Shift Keying PSU Power Supply Unit PWD Password PWK Pulse Width Keying QAM Quadrature Amplitude Modulation QoS Quality of Service QPSK Quadrature Phase Shift Keying RA Random Access RA-AC Random Access Allocation Channel RAS Radio Astronomy Service RA-SIG Random Access SIGnalling RBDC Rate Based Dynamic Capacity RC Raised-Cosine or Request Class RCS Return Channel over Satellite RCST RCS Terminal (compliant with DVB-RCS2) REC Rectangular RF Radio Frequency RFC Request For Comments RL Return Link RLE Return Link Encapsulation RMS Root Mean Square RMT RCS Map Table RNG Random Number Generator RO Roll-Off RRM Radio Resource Management RSC Recursive Systematic Convolutional (code) RSM RCST Security Module RSMS Regenerative Satellite Multimedia System ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 24 RSS Received Signal Strength RT Real Time RX Receiver SA Slotted Aloha SAT Satellite SCADA Supervisory Control And Data Acquisition SCT Superframe Composition Table SDU Service Data Unit SF Spreading Factor (in Annex E), or Super-Frame (in Clause 6) SGW Satellite Gateway SHA Secure Hash Algorithm SI Service Information SISO Single Input Single Output SLA Service Level Agreement SNIR Signal to Noise (plus) Interference Ratio SNR Signal to Noise Ratio SOF Start Of Frame SOSF Start of SuperFrame spfmsbf single precision floating-point, most significant bit first SPT Satellite Position Table SRM Satellite Remote MODEM SRS Space Research Service SSB Single Side Band ST Satellite Terminal SVN Satellite Virtual Network SW SoftWare SYNC SYNChronization SYNCD SYNC Distance TBTP2 Terminal Burst Time Plan 2 TC Turbo Coding TC-LM Turbo-Coded Linear Modulation (scheme in DVB-RCS2) TCP Transport Control Protocol TD Total Distortion TDM Time Division Multiplex TDMA Time Division Multiple Access TDOA Time Difference Of Arrival TIM-B Terminal Information Message Broadcast TIM-U Terminal Information Message Unicast TOD Time Of Day TRANSEC TRANSmission SECurity TRF Traffic TS Transport Stream TS/GS Transport Stream/Generic Stream TSS Time Slot Sharing TWT Travelling Wave Tube TWTA Travelling Wave Tube Amplifier TX Transmitter uimsbf unsigned integer most significant bit first UIU User Interface Unit UPL User Packet Length UW Unique Word VACP™ VSAT Antenna Control Protocol VBDC Volume Based Dynamic Capacity VCM Variable Coding and Modulation VMES Vehicle-Mounted Earth Stations XPD cross Polarization Discrimination ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 25 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4 Reference Model | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.0 Introduction | An overview of DVB-RCS2 system scenarios and network topologies is described in [i.3]. Reference models to realize such satellite networks could include different interconnections among Network Control Centre, Traffic Gateway(s), Feeder(s) and Terminals. In practice not all these interconnections will be implemented. Also, some functional blocks may be co-located. This clause describes therefore the network architectures that are more likely to be implemented for the service provision. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.1 Architecture with co-located NCC, Gateway and Feeder | The simplest architecture is an interactive satellite network with a single Traffic Gateway and a single Feeder co-located in an Earth Station (see Figure 4.1). The Network Control Centre is possibly also collocated. NCC Gateway Feeder Terminal controls User data return Interactive Network Adapter Broadcast Network Adapter Figure 4.1: Architecture with a single gateway and feeder (collocated) This Earth Station has both an Interactive Network Adapter and a Broadcast Network Adapter. It generates the forward link signal, including user data and the control and timing signals needed for the operation of the Satellite Interactive Network. It receives the RCST return signals, provides interactive services and/or connections to external service providers and networks and it provides monitoring, accounting and billing functions. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.2 Architecture with multiple feeders | When more Feeders exist in the interactive satellite network, the terminals should be able to switch from one to another, without losing network synchronization (see Figure 4.2). In order to achieve this, the following network architecture is envisaged. Terminals are equipped with at least two receivers. One receiver is continuously tuned to the forward link from a "primary" Feeder, the one which includes the control and timing signals and which provides monitoring, accounting and billing. The other receiver(s) can be tuned to different signals transmitted by "secondary" feeds to receive user data. The capability of the ODU to receive separate signals is the only limitation. In this configuration, terminals tuned to different "primary" Feeders (most likely belonging to different networks), might receive information from the same "secondary" Feeder(s). ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 26 NCC Gateway Feeder Terminal controls User data return Interactive Network Adapter Broadcast Network Adapter Feeder Broadcast Network Adapter Figure 4.2: Architecture with more than one feeder |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.3 Architecture with transparent mesh connectivity | By incorporating one or more TDMA burst receivers the RCST will be capable of receiving TDMA bursts as well as transmitting them. This allows RCSTs to communicate directly over a bent-pipe satellite, as indicated in Figure 4.3, as well as simultaneously operating according to the architectures of Figures 4.1 and 4.2. Figure 4.3 shows the architecture of Figure 4.1 extended with transparent mesh capability. NCC Gateway Feeder Terminal Forward controls Transparent Star User data Return controls Interactive Network Adapter Broadcast Network Adapter Terminal Transparent mesh User data Figure 4.3: Architecture with transparent mesh connectivity |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.4 Architecture with regenerative satellites and NGSO | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.4.0 General | Figure 4.4 highlights a reference architecture for a regenerative satellite network where On-Board Processing capability (OBP) of the satellite allows for an efficient peer-to-peer connectivity of RCSTs. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 27 Figure 4.4: Regenerative Satellite Network Reference Model A reference scenario for a Non-Geostationary Orbit (NGSO, also referred to as Non- Geosynchronous Orbit) scenario is given in Figure 4.5. In this scenario, each satellite is managed by an OBP as above, however it is optionally connected by Inter Satellite Links (ISL) to other satellites, carrying user, control and management traffic to and from the OBP, connecting it to gateways and terminals which are not served directly by it. The communication protocol of the ISL is out of scope of the DVB satellite communications specifications, ([i.1] to [i.4] and [i.80]), but it is assumed that higher layers SDU are available to the OBP, together with routing/switching information enabling the OBP to route them to proper destination. The term "mesh network" below would refer, in the case of NGSO constellation with ISL, to the entire network supported by that constellation. Figure 4.5: NGSO Satellites Network Reference Model The Onboard Processors are classified as: • Regenerative with onboard switching that can provide full traffic re-arrangement for point-to-point connections between terminals in a mesh network. The onboard processor can also be configured to support point-to-multipoint, multi-point-to-point connections and/or concentration /multicasting /broadcasting through flexible routing/switching between input and output ports. RCS2T Interactive Network Adapter RCS2T GATEWAY 2 STATION RCS2T Interactive Network Adapter Interactive Service Provider RCS2T RCS2T RCS2T Active gateway link Make-before- break gateway link Active user link Make-before- break user link Inter-satellite link Baseband link NCC GATEWAY 1 STATION ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 28 • Regenerative without onboard switching which is particularly attractive when the number of uplink and/or downlink beams is relatively small and the requirements for onboard traffic arrangement are moderate. In such cases the requirements for concentration and/or multicasting/ multiplexing type of connectivity prevail. • Regenerative in conjunction with transparent repeater which is based on a hybrid payload including both transparent and regenerative onboard switching repeaters. The terminals are connected to the RSMS network through the transparent repeater. Point-to-point connectivity between terminals is provided by the OBP Processor, hereafter called "mesh processor". The functional requirements of the OBP processor are: • Receive all traffic and control data sent by the terminals. • Receive all traffic and control data sent by the NCC. • Extract the traffic data to be sent on the downlink within DVB- S2/S2X format and route them to the appropriate output(s) towards the receiving terminals. • Generate/extract the control data to be sent to the NCC and route them to the appropriate output(s). • Format downlink streams including all the necessary downlink signalling messages in DVB-RCS2 compatible DVB-S2/S2X format and route/switch them to the appropriate output(s). Different OBP implementations result from the apportionment of the MAC functions between onboard processor and on-ground entities; the RCS terminals and NCC. These are described in Table 4.1. Table 4.1: MAC Functions Partitioning in case of regenerative OBP satellite systems MAC Function Network Entity Comments Data encapsulation / de-capsulation OBP See OBP switching / routing modes in clause 4.4.1. Routing_Label Extraction OBP See OBP switching / routing modes in clause 4.4.1. Frame Format OBP See OBP switching / routing modes in clause 4.4.1. Synchronization and power control OBP/NCC On board measurements NCR generation and insertion OBP/NCC For NGSO, NCR may be common to the entire constellation or independent to each satellite. Resource control and management RCS (Capacity requests)/NCC/OBP Performed on ground in case of Hybrid regenerative P/L with "mesh processor". Performed also on board in case Traffic manager is implemented on board RCS configuration and management NCC On ground Logon NCC/OBP On board measurements Network Configuration Network Configuration On ground |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.4.1 On board switching requirements | A regenerative system requires onboard routing or switching of signals between input and output ports. Different on board switching architectures can be used; circuit-switched, frame-switched, packet or cell switched architecture. The On-Board Processor in a regenerative system uses a DVB-RCS2 air interface on the uplink (return link), a DVB-S2/S2X in the downlink (forward link), and, for NGSO with ISL, an independent protocol over the ISL. Over the satellites to earth links, the OBP will be providing the modulation, coding and framing functions. Different architectures are possible on the additional functions (multiplexing, table-switching, routing, etc.) or the QoS support. In case of regenerative payload performing packet or cell switching, there are two types of information identified as necessary to perform routing/switching: • Addressing information to identify the destination RCST; and ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 29 • Control (routing) information to the Onboard Processor (OBP) to perform routing/switching. The increasing need of addressing and controlling/routing information is attributable to specific system design assumptions: • The multi-beam coverage and the multi-port onboard switching matrix increase the number of possible routes. • The support of multiple levels of Quality of Service. • The multi-operator context: the inter-operability is increasing the routing possibilities and limits the reuse of the identifiers. • The increasing number of simultaneous connections managed by a terminal (the number of customer premises equipment connected to this terminal). • The complexity increases for multiple- satellite multi gateway constellation. The mesh traffic between terminals is only handled by the terminals but controlled, monitored and allocated by the NCC. There are cases of regenerative satellite systems where the traffic manager functions are split between the NCC and the on-board processor. In this latter case, the on board traffic manager plays an active role for the control, monitoring and allocation of resources to terminals. A regenerative satellite network can be configured to support both star and mesh topologies based on DVB-RCS2 specification for the return channel and DVB-S2/S2X specification for the forward channel. As for a transparent satellite network, the regenerative satellite network in star topology supports access traffic to/from a gateway. In a regenerative architecture this star topology features some enhanced access network flexibility, such as interconnecting terminals to multiple gateways and/or multiplexing the star traffic with mesh traffic on a same given downlink carrier for the destination beam. The regenerative satellite network is characterized by the capability of star and mesh single-hop communications and are composed of one or more beams. For multi-beam regenerative systems, cross-connectivity can be supported. Multi-satellite NGSO constellation can be considered an extension of multi-beam regenerative systems in this sense. Following RLE/GSE RCS2 encapsulation, different OBP switching modes can be observed in a regenerative satellite system: • Carrier/timeslot switching: used mainly for NCC signalling and prosumer traffic, with no multiplexing gain • Burst switching: - Based on RLE in uplink and downlink: no decapsulation/re-assembly needed - All RLE packets in one burst should share destination (and ACM/QoS) • Packet switching: - Burst label not needed - No fragmentation of packets is considered (high impact in OBP) - Packet length should be chosen to fit in burst length • Fragment label switching: Fragment label is used for switching and sender identification, with two profiles: - UL-RLE/DL-RLE the most efficient and simple - UL-RLE/DL-GSE requires more complexity at OBP level • Layer 3 switching: - OBP capable of fully de-encapsulation - encapsulation up to IP layer ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 30 Figure 4.6: OBP switching / routing modes in RCS2 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.5 Encapsulation Scenarios | According to the payload architecture (regenerative or transparent) and the type of connectivity (star or mesh), there will be several possibilities for the encapsulation scheme of the different satellite links. These methods can be clarified with the help of the following scenarios: • Star-only encapsulation scenario: this is the mandatory mode. RLE encapsulation is used in the return link and GSE in the DVB-S2/S2X forward link. The RCST terminals compatible with this mode has an RLE encapsulator and a GSE decapsulator. Figure 4.7: Star-only encapsulation scenario • Star and Mesh transparent encapsulation scenario: this scenario is mandatory for mesh transparent systems. RLE encapsulation is present in the downlink due to the presence of mesh user terminals with DVB-RCS2 demodulator (in addition to the DVB-S2/S2X demodulator). GSE is also in the downlink for reception of the forward link for signalling and star traffic. It is foreseen the presence of star-only and mesh terminals in the same satellite system. Mesh overlay RCSTs need two decapsulators, for RLE (user traffic) and GSE (DVB signalling) encapsulations. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 31 Figure 4.8: Star and Mesh transparent encapsulation scenario • Mesh regenerative (option 1) encapsulation scenario: this scenario uses RLE for the uplink and GSE in the downlink, for all of the RCSTs and also for the NCC modem ODU (NCC-RCST). The OBP is in charge of translation of the encapsulation mode from RLE uplinks to GSE downlinks. This mode is intended for the use of standard RCSTs with little changes in SW (only related to DCP functionality) with regard to the transparent star terminals used in the star-only scenario. Regenerative RCST Regenerative RCST NCC Regenerative Satellite (RLE to GSE conversion) RLE GSE RLE RLE GSE GSE Figure 4.9: Mesh regenerative encapsulation scenario 1 • Mesh regenerative (option 2) encapsulation scenario: only RLE encapsulation is used, since the OBP does not perform any change in the encapsulation of uplink packets. The RCST supports RLE encapsulation and decapsulation functions, plus the DCP functionality for dynamic connectivity. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 32 Figure 4.10: Mesh regenerative encapsulation scenario 2 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6 Reference model for mobile Scenarios | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6.0 General | A reference model for mobile DVB-RCS2 system architecture, as envisaged in the normative and system documents is depicted in Figure 4.10. Figure 4.11: Overall system architecture for mobile interactive services via satellite ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 33 The space segment includes one or more GEO-satellites with a single or multi-beam configuration per satellite and with performances equivalent to those used for classical Ku-band or Ka-band fixed satellite services. Two service coverage scenarios are identified: one for regional case (one country or part of continent covered); and one for global case (land coverage complemented by oceans wide coverage, e.g. transatlantic). Mobile RCSTs are in most cases mounted on a mobile platform operating as an access point for multiple users. The RF characteristics are adapted to the service requirements (in particular, antenna minimum/maximum size depending on the applications, regulatory and accommodations constraints). The IDU component, compliant to a mobile profile, covers the different applications. The ground segment sub-system consists of the Network Operator Centre (NOC), the Network Control Centres (NCC) compatible with the definition of the NCC and the gateways, providing access to the terrestrial networks. The NOC is in charge of sharing the satellite resource among network operators, bandwidth allocation to NCC's and centralized management of satellite handover if relevant. The RCST is managed by one single NCC within the satellite coverage area. Scenarios envisaged by a mobile system can be classified as follows: • Line-Of-Sight (LOS) scenarios: these correspond to low-fading scenarios, which are almost always in LOS or close to LOS conditions. The aeronautical and maritime are the two main scenarios in this category. • Non-LOS scenarios: these correspond to land-based strong-fading scenarios characterized by frequent/deep/long signal blockages and shadowing. The railway and the land vehicular are the two main scenarios in this category. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6.1 Line-Of-Sight Scenarios | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6.1.1 Maritime | The maritime scenario comprises mainly: passenger transportation ships (like ferries and cruises); commercial ships (like cargos and tankers); and private transportation ships (like sailing boats). Two coverage scenarios are particularly considered: a global one, corresponding to cruise routes (like transatlantic cruises); and a regional one, corresponding to ferry routes (for example, in Europe near the coast). Regarding the modelling of the LOS channel conditions encountered in the maritime scenario, no experimental results concerning propagation measurements at the frequencies of interest are available in the literature. Nevertheless, considering the usage of high directive antennas, LOS conditions can be assumed and the channel can be regarded as a pure Ricean channel with very high Rice factor, i.e. very close to a purely AWGN channel. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6.1.2 Aeronautical | The aeronautical applications include providing telecommunication access to: passenger aircrafts (consisting in wide body and single aisle aircrafts); and private aircraft (like executive jets) or Unmanned Aeronautical Vehicles. Two coverage scenarios are particularly considered: a global one, corresponding to long haul flights (like transatlantic flights); and a regional one, corresponding to short to medium haul flights (for example, over Europe). For Line-of-Sight channel conditions encountered in the aeronautical scenario, experimental results at 18,6 GHz [i.48] show that the channel can be modelled as a Ricean during normal flight situations and manoeuvres, with a Rice factor well above 20 dB. Some signal fading, in the order of 3 dB, was observed for manoeuvres with roll angles up to 20°, whereas only in case of extreme manoeuvres, with roll angles up to 45°, the influence of the aircraft structure resulted in deep fades in the order of 15 dB. In conclusion, in aeronautical scenarios, LOS conditions can typically be assumed, and the channel can be fairly approximated by a purely AWGN channel. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 34 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6.2 Non-LOS Scenarios | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6.2.1 Railway | The mobility effects, such as multipath, shadowing and blockage, encountered due to the local environment in the vicinity of the mobile RCST, such as adjacent buildings, vegetation, bridges, and tunnels, result in non-LOS conditions of severe fading. Several propagation measurements at Ku band [i.49] and Ka band [i.50], [i.51] were performed, based on which reference statistical channel models have been established: • For Ku-band, the behaviour of the land mobile satellite channel can be modelled using a 3 state (namely LOS, shadowed and blocked) Markov chain based model, where each state is further characterized by a Rice distribution. • For Ka-band, the behaviour of the land mobile satellite channel can be modelled using a 3 state (namely LOS, shadowed and blocked) Markov chain based model, where each state is further characterized by a Loo distribution. The railroad satellite channel is in LOS state for most of the time. However, short blockages due to power arches as well as long blockages due to obstacles, such as buildings, vegetation, bridges, and tunnels, are also possible leading to non-LOS effects |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 4.6.2.2 Vehicular | The land vehicular scenario comprises mainly: passenger vehicles, commercial vehicles and private vehicles. Only the regional coverage scenario is considered since vehicles remain within one continent. For the modelling of the non-LOS channel conditions, the same statistical channel models described above for the railway scenario apply here as well except for deterministic and recurring signal attenuation (due to power arches for example) that is not applicable in non-LOS channel |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5 Forward Link | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.0 Forward Link Implementation | Figure 5.1 shows one way to implement the forward link signalling from the NCC (the SI-information for RCS) to an existing DVB-S2/S2X forward-link. The SI-tables with signalling data to the RCST from the NCC are represented as binary data (for example in binary files). These binary data are sent to the Gateway and put into a PSI/SI-inserter, together with the PSI/SI binary signalling data for DVB-S2/S2X. It can also be transported over a generic stream, as specified in clause 5.1.1 of ETSI EN 301 545-1 [i.1]. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 35 MOD M U X PSI/SI Inserter audio, video, data etc. DVB-S2 link RCST PSI/SI Network Control Center (NCC) SI for RCS Gateway ISP Figure 5.1: Implementation of SI signalling from NCC in DVB-S2 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1 Forward Link Considerations | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.1 General Notes on ACM Operation | A DVB-S2/S2X ACM forward link will typically be configured by the network operator with a "minimum" (most robust) and "maximum" (most spectrally efficient) MODCOD. The minimum MODCOD will be used to transmit critical data such as TBTPs and NCR synchronization. The selection of the minimum MODCOD will be based on the link budget analysis and is chosen to ensure that all RCSTs are able to maintain forward link synchronization. All RCSTs in the network will periodically relay their reception capability back to the hub using the return link data channel (see clause 8.3.2 of ETSI EN 301 545-2 [i.1]). The NCC will select a new MODCOD every time it transmits a BBFRAME. The "ACM Gain" of the system is the effective spectral efficiency of the forward link divided by the spectral efficiency that would have been achieved if only the "minimum" MODCOD had been used. In order to maximize the ACM Gain of the system, the NCC should attempt to send all unicast data to RCSTs using their best achievable MODCOD. To do this, the NCC needs to maintain a table of the current MODCOD reception capability of each RCST in the network. The system should allow the MODCOD to potentially change with every BBFRAME to maximize the ACM Gain. It is important to note that in a DVB-S2/ACM system, any given RCST may be incapable of receiving many/most of the BBFRAMEs on the forward link. This depends on the location of the RCST within the footprint of the satellite beam. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.2 FEC Frame Size Selection | In DVB-S2/S2X ACM, there is a fundamental trade-off between the increased coding gain achieved by normal size DVB-S2/S2X frames, and the reduced latency achieved by short DVB-S2/S2X frames. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 36 For many interactive applications, reducing the latency with short frames is more important than the 0,2 - 0,3 dB in coding gain (see clause 6 in ETSI EN 302 307 [i.2] and [i.80]) with normal frames at a given MODCOD. The small loss of efficiency can usually be recovered with fine-grained MODCOD selection using ACM as described in clause 5.1.1. Experience has shown that there is a performance gain when using short frames, compared to normal frames, given the same target latency for data packets. Normal frames with short target latencies result in many large BBFRAMEs being transmitted with padding, thus wasting bandwidth. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.3 Receiver Operation | A non-viable MODCOD is one that the receiver knows it cannot demodulate because the signal quality, measured in Signal to Noise plus Interference Ratio (SNIR), of the received signal is too low. A background software control loop in the RCST can periodically determine the SNIR and set the demodulator to filter out these MODCODs. This prevents known-bad input from entering the higher layers. Depending on the demodulator implementation, it can also provide an opportunity for the LDPC decoder to perform more iterations on the viable MODCODs. RCSTs should attempt to decode data on all viable MODCODs. There is no explicit signalling from the hub to indicate which MODCODs a terminal should demodulate or which MODCOD the NCC considers as the lowest viable MODCOD for the RCST. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.4 Pilot Symbols | When using ACM, the demodulator can only rely on the presence of PLHEADER and pilot symbols. Terminals at the edge of a beam may not have many viable MODCODs to detect. The use of time interval between PLHEADERs alone can be insufficient to allow the RCST to remain locked to the DVB-S2/S2X signal. Frames with pilot symbols have negligible overhead (see clause 5.5 of ETSI EN 302 307 [i.2] and [i.80]). The efficiency ranges from 99,3 % to 99,7 % when using normal frames; 97,3 % to 98,9 % when using short frames. It is therefore recommended that pilot symbols always be used for DVB-S2/S2X ACM deployments. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.5 Restrictions on Transmitting DVB-S2 Dummy Frames | Some existing DVB-S2/S2X demodulator ASIC implementations could dysfunction when DVB-S2/S2X signals contain dummy frames, even though such chipsets are otherwise ACM-capable. As a work-around for the above mentioned issue, it is recommended that empty (DFL=0) 32APSK short frames, with pilots, be used in lieu of dummy frames. A DVB-S2 dummy frame is 36 SLOTs, plus 1 SLOT for the PLHEADER (37 × 90 = 3 330 symbols). This is the same length as a short 32-APSK frame with no pilot symbols. A short 32APSK frame with pilots is 3 402 symbols, an increase of 2 %. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.6 GSE Fragmentation | A data packet transmitted using GSE may be fragmented prior to transmission. The resulting fragments may be transmitted over multiple BBFRAMEs. The system should allow GSE fragments to be transmitted on BBFRAMEs of any viable MODCOD that the (set of) target RCSTs is capable of receiving. This helps to improve the overall ACM gain of the system. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.7 Explicit Integrity Protection of SDU | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.1.7.1 Background | Clause 5.1.1.2 of ETSI EN 301 545-2 [i.1] indicates that the RCST supports a modified format and syntax of the BBFRAME data field for transport of the GSE PDUs by inserting a CRC32 as the last four bytes of the BBFRAME data field. A background and rationale for such protection is provided below. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 37 The GSE protocol relies on error free reception of PDU's contained in each BBFRAME for proper GSE operation [i.3]. While this is a valid assumption for certain scenarios, such as CCM transmission to fixed receivers in DVB-S2 broadcast systems, it may not hold in other circumstances that are important for interactive services. In particular, transient propagation conditions in ACM operation and the occurrence of signal interruptions, which are common in land-mobile scenarios, are examples of situations where further error detection capability may be required. These two cases are further analysed below. ACM Systems For continued, error-free operation ACM systems need to include a margin that allows reception while a MODCOD change is being carried out. In order to maximize the capacity, it is of course desirable to minimize this margin. The margin needs to account for C/(N+I) measurement uncertainty as well as for additional fade that occurs in the switchover latency interval. In the following analysis, an undetected frame error probability of 10-7 (quasi error free) is targeted. For the worst-case BCH detection performance of 2 × 10-4, the actual frame error probability should therefore be kept below 5 × 10-4 at all times. When relying on the BCH decoder, the link margin should therefore ensure that the probability of a frame being received below the actual detection threshold for the current MODCOD is smaller than this value. A typical RMS C/(N+I) measurement uncertainty is 0,1 dB. This can be achieved relatively quickly even at low C/(N+I), for example by data-aided estimation on approximately 4 000 symbols. This corresponds to the frame headers and pilots of about 13 short QPSK frames. Assuming a Gaussian distribution (which is reasonable for a data-aided estimate), the error magnitude exceeded with a probability of 5 × 10-4 is ~3,5 σ or 0,35 dB. Fading during the system reaction time depends strongly on the current fade level as well as on implementation details. A model of the statistical distribution of fade slope is given in [i.35]. This model predicts the probability that a given fade slope (in dB/s) is exceeded for a given level of fading. Figure 5.2 is computed according to that model and shows the fade slope exceeded for 0,05 % of the time (= 5 × 10-4) as a function of the instantaneous fade. It can be noted that the model is frequency-independent; propagation measurements have shown a very similar behaviour in all bands. However, the probability of actually having a fade of a given level of course depends on the band. The system reaction time is very implementation-specific. The physical lower bound is one round-trip time (500 ms). The total margin needed is the sum of the measurement uncertainty and the worst-case additional fade in the reaction interval. Assuming statistical independence of the two terms, the values for 5 × 10-4 were used for each to predict the overall value for the same probability. For a 1-second reaction time near clear weather (2 dB fade), the margin is 0,44 dB. For a 10 dB fade and 5 seconds reaction time, it is 2,7 dB. This corresponds to a significant loss of capacity, so it is highly desirable to operate with smaller margins. This however increases the probability of errored frames during the transient. While some frame errors can sometimes be tolerated, it is important that the errors are detected reliably. Figure 5.2: Fade slope exceeded 0,05 % of time 0 5 10 15 0 0.1 0.2 0.3 0.4 0.5 0.6 Instantaneous fade level, dB Fade slope exceeded 0.05% of time, dB/s ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 38 Land-Mobile Systems Signal blockages in Non-Line-Of-Sight mobile (NLOS) systems are equivalent to fades that occur very rapidly. A blockage can occur at any time. Depending on the receiver implementation, it is quite possible that the frame may have been detected before the loss of signal occurred. In other words, the on-set of a signal blockage is likely to be seen as a frame error. It is important for the operation of the link-layer FEC NLOS countermeasures that these frame errors are detected properly so that the corresponding symbols can be marked as erasures for the LL-FEC decoders. Consider a high-speed train moving at 350 km/hr (97 m/s). With power poles spaced at 50 m, there will be 1,94 blockages/s just from these. Further assume that each blockage causes one detected-but-errored frame and that the probability of undetected frame errors is given by the performance of the BCH decoder (2 × 10-4 = 1 in 5 000, worst case). This will result in an undetected frame error on average every 5 000 / 1,94 seconds or 43 minutes. This is clearly inadequate. Experimental results presented confirm that the error detection capability of the BCH decoder is not sufficient to achieve the desired robustness. Analytical results however, confirm that an explicit 32-bit CRC check will provide the necessary robustness. In order to allow the system implementer the flexibility to trade off the margin against the risk of frame errors during the transient, it is recommended that the encapsulation should be robust against such frame errors; in particular since this can be achieved with the minimal overhead incurred by a CRC-32 BBFRAME carrying GSE PDUs as shown in Figure 5.3. Figure 5.3: Recommended use of CRC32 per BBFRAME carrying GSE PDUs |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.2 Forward Link L2S Considerations | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.2.1 L2S Insertion | Most broadcast L2S messages are fundamental to RCS2 system operation, and should therefore be transmitted on the lowest system MODCOD. These messages include: NIT, SCT, FCT2, BCT, SPT and TBTP2. Unicast L2S messages may be transmitted on any MODCOD that can be received by the targeted RCST. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.2.2 NCR Insertion | It is recommended that NCR timestamps be inserted at a fixed, periodic interval. An RCST can detect the transmission interval and optimize its NCR tracking loop accordingly. This is not a hard-realtime requirement; however, the NCC should aim to transmit NCR timestamps at a reasonable fixed interval. BB HEADER DATA FIELD ModulatorPA DDING MATYPE (2 bytes) UPL (2 bytes) DFL (2 bytes) SYNC (1 byte) Deint erlea CRC -8 (1 byte) DFL Base Band Frame GSE HDR GSE Data GSE HDR GSE Data MATYPE - 2 (1 byte) TS/GS ( 2bits) SIS/MIS ( 1bit) CCM/ ACM ( 1bit) ISSYI ( 1bit) NPD ( 1bit) RO ( 1bit) CRC32 ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 39 ETSI EN 301 545-2 [i.1] specifies as a minimum the NCR Insertion rate of 10 times per second (see clause 6.3 of [i.1]). In order to minimize the impact of the L2S on the forward link overhead it is desirable to maintain the NCR insertion rate close to the minimum. However, due to operation and channel conditions, such as support of mobile services, it may be necessary to use a higher NCR insertion rates. In certain system scenarios NCR transmission rates of up to 32 times per second has been reported. It is therefore recommended to carefully trade off desired efficiency with the required frame loss resilience in the operational environment when selecting the NCR insertion rate. Because the NCR is fundamental to RCS2 system timing, it should always be transmitted on the lowest system MODCOD. The NCR transmission frequency will therefore affect the overall ACM gain achievable in the system. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.2.3 FL L2S Components | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.2.3.0 Overview | ETSI EN 301 545-2 [i.1], clause 6 provides the description and use of lower layer signalling components. In this clause, some guidelines on the use of L2S components are provided. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.2.3.1 NIT | It is recommended that the following descriptors are present in NIT: • Linkage descriptor (for finding the RCS service information) • Multiplex Stream Specification containing: - System delivery descriptor - S2/S2X System delivery descriptor |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 5.2.3.2 SCT/FCT2/BCT/TBTP2 | It is assumed that the SCT, FCT2 and BCT contents do not change frequently, so the issues related to handling of changing superframe configurations dynamically are considered outside the scope of the guidelines. The TBTP2 is used to allocate time slot allocations to single (DAMA) or groups (RA) of terminals. The flexibility in the TBTP2 allows dynamic superframe to superframe changes of the frame allocation including changes in: • MODCOD • Symbol rate • Burst duration changes (by aggregating continuous allocated BTUs • Access method (DAMA or RA) • Burst type content (Traffic, signalling or both) It should be noted that the frame duration in FCT2 has the format of NCR counter as integer, while the same value in the FCT from ETSI EN 301 545-2 [i.1] used the PCR base and extension format. Figure 5.4 shows across time and frequency dimensions the information included in the SCT. Two superframes are included in the SCT in this example, covering different portions of the frequency. It should be emphasized that this illustration deliberately shows the information that can be deduced from the SCT only. The complete definition of superframes will need joint interpretation of SCT, FCT2, BCT, and TBTP2 tables. Primarily, the superframe definition within the SCT provides information regarding: • the start time, the duration, and the centre frequency of a superframe. • the start time and the centre frequency for each frame within the superframe. Each frame in the superframe is identified by an implicit frame numbering within the superframe, each frame referring to a frame type specifying its structure. FCT2 contains the specification of each frame type. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 40 As stated in the diagram, the SCT also contains information regarding the superframe sequence identification, superframe counter, uplink polarization, and whether or not logon with large timing uncertainty is supported in this superframe sequence. Each superframe sequence has its own superframe specification in the SCT. Figure 5.4: Visualization of SCT Content Figure 5.5 shows five different frame type examples. The BTU grid, which is completely specified in the FCT2, is pictorially shown for each frame type. In addition, FCT2 may contain a section loop for each frame type. The red dashed arrows in Figure 5.5 presents an example timeslot definition within the frame type, p. The first section in the section loop corresponds to the first timeslot in the frame (lowest frequency, lowest start time). These red dashed arrows show the correct order in which different elements in the FCT2 should be read to define the first timeslot, which spans 6 BTUs in the lowest frequency and first BTU in time. The example tx_type is 128. Also shown in the figure are the tx_content_type and tx_format_class that correspond to tx_type:128; this information would be available in the BCT. For tx_type:128, the example shows a tx_content_type:1 and a tx_format_class:1. As shown in the tables in the example, tx_content_type:1 indicates that the timeslot is to be used for logon and tx_format_class:1 indicates that the timeslot is to carry a Linear Modulation burst. Back in the section loop, the fixed_access_method parameter for this timeslot is 0xF, which indicates that the timeslot is for random access. Had the fixed_access_method been 0 for this timeslot, then the timeslot would have been a dedicated access timeslot, the owner which would have been indicated in a follow-up TBTP2 table. The frame type may combine a number of consecutive BTUs into a timeslot. For each such combination, FCT2 may indicate a default_tx_type and a fixed_access_method. The default_tx_type =0 indicates that the tx_type is determined in TBTP2. Also, the non-zero default_tx_type allocations may be overridden by a tx_type specified for the timeslot in TBTP2. The transmission in a timeslot with a given tx_type is completely determined by the tx_type specification contained in the BCT table. Figure 5.5 below shows a simplified BCT example to illustrate how FCT2 and BCT jointly specify the stationary structure of each frame type. The BCT determines the timeslot size for each tx_type in BTU units. Thus, the timeslot duration in absolute time depends on the BTU specification. Timeslots using the same tx_type may have different duration when appearing in frames of different frame types. Figure 5.6 completes superframes that would emerge by joint interpretation of the example SCT, FCT2, and BCT contents. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 41 Figure 5.5: Visualization of FCT2 Content ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 42 Figure 5.6: Superframe Composition, as specified in SCT, FCT2, BCT Figure 5.7 shows an example encapsulation of single user traffic SDU and one L2 signalling SDU in two FPDUs. The first FPDU is transmitted in a timeslot that is associated with a tx_content_type: '3' in the BCT (see Figure 5.5 for tx_content_types), since it carries both user traffic and L2 signalling. The second FPDU is sent in a timeslot with tx_content_type: '3' or '4', since it only carries user traffic. Thus, the encapsulation procedure should take into account the content type allowed for a given timeslot before building the payload. Figure 5.7: User Traffic SDU Encapsulation Timeslots that only carry L2 signalling are associated with tx_content_type '1' (for Logon) or '2' (for Control) in the BCT. Figure 5.8 shows an example of L2 signalling being encapsulated in a FPDU. With such timeslots, the L2 signalling SDU need not ALPDU and PPDU headers. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 43 Figure 5.8: L2 Signalling Encapsulation It should be noted that the SCT/FCT2/BCT/TBTP2 are sufficient to identify RA timeslots in a superframe. RA timeslots with tx_content_type '1'or '2' are for L2 signalling only, and Slotted ALOHA is the random access mechanism to be used with these timeslots. RA timeslots with tx_content_type '3' or '4' may carry user traffic. The specific access mechanism for these slots (Slotted ALOHA vs. CRDSA) is indicated in periodic Random Access Traffic Method descriptors. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6 Return Link | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.1 Fixed-RCST Physical Layer Synchronization | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.1.0 Ovdrview | For administrative and technical reasons the location of an RCST providing fixed satellite services should be known to the network operator. An RCS system can be designed assuming an accuracy of the location (longitude, latitude and altitude) of the RCST of no more than a few kilometres. Some network operators may require a better accuracy. It is recommended that commonly available high precision localization systems be used during the RCST installation or any re-installation. If possible, the NCC should correct for satellite translation error and Doppler shift introduced on the NCC-to-Satellite uplink and the Satellite-to-NCC downlink. The residual frequency offset between any two RCSTs includes effects due to Doppler shift on the Satellite-to-RCST downlink and the RCST-to-Satellite uplink. The residual relative frequency offset needs also to be compensated for by the NCC. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.1.1 DVB S2 Receiver implementation aspects | There is a requirement to associate, without any ambiguity, the detected SOF with decoded frames (and hence with NCR fields within the frame), and to make this information available to the NCR synchronization circuit in the RCST. Decoding processing delays are far less critical. An association and interfacing approach that puts almost no functional requirements on the DVB S2 RX chip is outlined in DVB-S2 Specifications [i.2], clause G.5. Decoding delay jitter does not degrade the network clock reference extracted by the terminal. It is important however to verify that SOF detect circuit has low jitter. (Typical requirement, depending on NCR tracking design: SOF detect jitter < 100 ns pp). It is recommended that systematic delays in the SOF detect circuit (for example the delay of the RX filter) are documented by the DVB S2 RX circuit provider. These delays can then be compensated for in the RCST design, if needed. It is also recommended to prevent that data flagged as unreliable enter the NCR tracking circuit. Unreliable data flags can originate from in a DVB S2 receiver from BBHEADER checking, BCH decoding checks, or the CRC-32 associated per BBFRAME (as per clause 5.1.1.2 of ETSI EN 301 545-2 [i.1]). ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 44 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.2 Mobile RCST Synchronization | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.2.0 General | The present clause addresses the impact of mobility (i.e. Doppler) on the physical layer performances on the return link. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.2.1 Doppler shift and time drift | Table 6.1 gives some typical values in Ku-band for Doppler shift and time drift, for different types of mobile terminals (i.e. with various speed and acceleration). The table provides worst case Doppler shifts, assuming terminal motion towards the satellite and minimum elevation angle (leading to a minimum relative angle θ between the vehicle and the satellite θ=0). The Doppler values due to satellite motion are also included for reference. Table 6.1: Doppler shift in Ku-band for different types of mobile terminals Type of mobile terminal (note 1) Speed Acceleration (m/s2) Doppler rate (note 2) Uplink Doppler frequency shift (note 3) (Hz) Downlink Doppler frequency shift (note 4) (Hz) Time drift (ns/s) Uplink frequency drift (Hz/s) Downlink frequency drift (Hz/s) Pedestrian 5 km/h 1 4,6E-09 67 59 4,6 48 43 Maritime 25 km/h 5 2,3E-08 336 295 23,1 242 213 Vehicular 120 km/h 10 1,1E-07 1 611 1 417 111 483 425 Train 350 km/h 5 3,2E-07 4 699 4 132 324 242 213 Aeronautical 330 m/s 17 1,1E-06 15 950 14 025 1 100 822 723 Satellite 3 m/s 0 1,0E-08 145 128 10 4,8 4,3 NOTE 1: Vehicular: bus, car, truck Aeronautical: < speed of sound Satellite: satellite movement (GSO) assuming satellite motion is versus nadir (reference point) NOTE 2: The maximum Doppler values due to satellite motion are typical for geostationary satellite during main mission life. The worst case Doppler values (e.g. when satellite mission is extended using inclined orbit satellite) are not considered here. NOTE 3: Uplink frequency: 14,5 GHz NOTE 4: Downlink frequency: 12,75 GHz Table 6.2 provides typical values for Doppler shift in Ka-band. Table 6.2: Doppler shift in Ka-band for different types of mobile terminals Type of mobile terminal (note 1) Speed Acceleratio n (m/s2) Doppler rate (note 2) Uplink Doppler frequency shift (note 3) (Hz) Downlink Doppler frequency shift (note 4) (Hz) Time drift (ns/s) Uplink frequency drift (Hz/s) Downlink frequency drift (Hz/s) Pedestrian 5 km/h 1 4,6E-09 139 94 4,6 100 67 Maritime 25 km/h 5 2,3E-08 694 468 23,1 500 337 Vehicular 120 km/h 10 1,1E-07 3 333 2 244 111 1 000 673 Train 350 km/h 5 3,2E-07 9 722 6 546 324 500 337 Aeronautical 330 m/s 17 1,1E-06 33 000 22 220 1 100 1 700 1 145 Satellite 3 m/s 0 1,0E-08 300 202 10 10,0 6,7 NOTE 1: Vehicular: bus, car, truck Aeronautical: < speed of sound Satellite: satellite movement (GSO) assuming satellite motion is versus nadir (reference point) NOTE 2: The maximum Doppler values due to satellite motion are typical for geostationary satellite during main mission life. The worst case Doppler values (e.g. when satellite mission is extended using inclined orbit satellite) are not considered here. NOTE 3: Uplink frequency: 30,0 GHz NOTE 4: Downlink frequency: 20,2 GHz ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 45 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.2.2 Frequency accuracy | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.2.2.0 General | The frequency accuracy of the terminal burst is the result of a number of contributors, some of which are independent of the terminal speed (i.e. of the terminal-related Doppler Effect). Typical fixed contribution - i.e. the frequency accuracy typical of a classical system - is provided in clause 10 (Table 10.2). The frequency accuracy provided in Table 10.2 is typical of that obtained in a DVB-RCS(2) satellite system. In the case of a mobile environment, the major additional contribution is due to the terminal motion. As explained before, it induces two types of effects: the first one is related to the induced Doppler on the NCR received reference (which derives in a frequency offset on the terminal transmit frequency). The second one is a frequency offset on the return uplink path (from terminal to satellite). The resulting frequency Doppler shift provided in Tables 6.3 and 6.4 includes these two contributions (one relevant to downlink Doppler and the other to uplink Doppler), but both applying to the uplink frequency. The tolerable burst frequency offset within the gateway modem is dependent on the gateway receiver modem implementation, and as is such not directly specified in the DVB-RCS2 standard. A representative set of values for acceptable burst frequency offset that range from 0,5 % to 3 % of a symbol rate is considered instead, in agreement with DVB-RCS2 system and gateway manufacturers. This allows to derive several combinations of maximum terminal speed and compatible terminal symbol rates. The following analysis assumes that the terminal frequency burst accuracy is the same at initial access of the terminal (logon burst) as for the subsequent traffic bursts. This relies on the assumption that no specific enhancement on the logon burst is performed to facilitate its demodulation and frequency detection. It also means that the traffic burst does not further benefit from frequency correction provided by the network. The discussion is hereafter provided on that assumption, covered by the current DVB-RCS2 standard and allowing to extend the range of standard applicability to mobile services. The analysis is performed both for Ku-band and Ka-band. Table 6.3 summarizes for the Ku-band case the minimum symbol rate requirement in order to be compatible with the aggregated frequency shift generated by the terminal motion and the fixed contribution detailed in Table 10.2. Table 6.3: Minimum symbol rate requirement as a function of terminal speed (Ku-band) Pedestrian Maritime Vehicular (bus car, truck) Train Aeronautical (< speed of sound) Fixed Terminal Speed of the terminal 5 km/h 25 km/h 120 km/h 350 km/h 1 188 km/h 0 km/h Freq. Doppler Shift (U/L and D/L) 134 Hz 671 Hz 3 222 Hz 9 398 Hz 31 900 Hz 0 Hz Fixed contribution 1 590 Hz 1 590 Hz 1 590 Hz 1 590 Hz 1 590 Hz 1 590 Hz Aggregated Frequency Drift 1 724 Hz 2 261 Hz 4 812 Hz 10 988 Hz 33 490 Hz 1 590 Hz Symbol rate frequency accuracy Minimum symbol rate (ksym/s) 0,5 % 345 452 962 2 198 6 698 318 1 % 172 226 481 1 099 3 349 159 2 % 86 113 241 549 1 675 80 3 % 57 75 160 366 1 116 53 Figures 6.1 and 6.2 present the allowed terminal speed as a function of the symbol rate of the terminal for the different acceptable burst frequency accuracy within the gateway modem (expressed as a percentage of the symbol rate). The range of symbol rate is intentionally limited to about 2 Msym/s, in order to remain in representative target rates in terms of service (typically from a few kbits/s to 1 Mbits/s). ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 46 Figure 6.1: Minimum symbol rate compatible with high-speed terminal motion (Ku-band) Minimum Symbol Rate compatible with terminal motion (Ku-band) 0 km/h 20 km/h 40 km/h 60 km/h 80 km/h 100 km/h 120 km/h 140 km/h 0 ksym/s 200 ksym/s 400 ksym/s 600 ksym/s 800 ksym/s 1000 ksym/s 1200 ksym/s Symbol Rate Terminal Speed 0.5% of symbol rate frequency accuracy 1% of symbol rate frequency accuracy 2% of symbol rate frequency accuracy 3% of symbol rate frequency accuracy Vehicular Ship Pedestrian Figure 6.2: Minimum symbol rate compatible with low-speed terminal motion (Ku-band) Table 6.4 summarizes for the Ka-band case the minimum symbol rate requirement in order to be compatible with the aggregated frequency shift generated by the terminal motion and the fixed contribution detailed in Table 10.2. Minimum Symbol Rate compatible with terminal motion (Ku-band) 0 km/h 200 km/h 400 km/h 600 km/h 800 km/h 1000 km/h 1200 km/h 1400 km/h 0 ksym/s 400 ksym/s 800 ksym/s 1200 ksym/s 1600 ksym/s 2000 ksym/s 2400 ksym/s Symbol Rate Terminal Speed 0.5% of symbol rate frequency accuracy 1% of symbol rate frequency accuracy 2% of symbol rate frequency accuracy 3% of symbol rate frequency accuracy Vehicular Aeronautical Train ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 47 Table 6.4: Minimum symbol rate requirement as a function of terminal speed (Ka-band) Pedestrian Maritime Vehicular (bus car, truck) Train Aeronautical (< speed of sound) Fixed Terminal Speed of the terminal 5 km/h 25 km/h 120 km/h 350 km/h 1 188 km/h 0 km/h Freq. Doppler Shift (U/L and D/L) 278 Hz 1 389 Hz 6 667 Hz 19 444 Hz 66 000 Hz 0 Hz Fixed contribution 3 784 Hz 3 784 Hz 3 784 Hz 3 784 Hz 3 784 Hz 3 784 Hz Aggregated Frequency Drift 4 062 Hz 5 173 Hz 10 451 Hz 23 228 Hz 69 784 Hz 3 784 Hz Symbol rate frequency accuracy Minimum symbol rate (ksym/s) 0,5 % 812 1 035 2 090 4 646 13 957 757 1 % 406 517 1 045 2 323 6 978 378 2 % 203 259 523 1 161 3 489 189 3 % 135 172 348 774 2 326 126 Figures 6.3 and 6.4 represent the range of minimum symbol rates for maximum allowable terminal speed. The range of symbol rate is here again intentionally limited to about 2 Msym/s, in order to remain in representative target rates in terms of service (typically from a few kbits/s to 1 Mbits/s). Figure 6.3: Minimum symbol rate compatible with high-speed terminal motion (Ka-band) Minimum Symbol Rate compatible with terminal motion (Ka-band) 0 km/h 200 km/h 400 km/h 600 km/h 800 km/h 1000 km/h 1200 km/h 1400 km/h 0 ksym/s 400 ksym/s 800 ksym/s 1200 ksym/s 1600 ksym/s 2000 ksym/s 2400 ksym/s Symbol Rate Terminal Speed 0.5% of symbol rate frequency accuracy 1% of symbol rate frequency accuracy 2% of symbol rate frequency accuracy 3% of symbol rate frequency accuracy Vehicular Aeronautical Train ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 48 Minimum Symbol Rate compatible with terminal motion (Ka-band) 0 km/h 20 km/h 40 km/h 60 km/h 80 km/h 100 km/h 120 km/h 140 km/h 0 ksym/s 400 ksym/s 800 ksym/s 1200 ksym/s 1600 ksym/s 2000 ksym/s Symbol Rate Terminal Speed 0.5% of symbol rate frequency accuracy 1% of symbol rate frequency accuracy 2% of symbol rate frequency accuracy 3% of symbol rate frequency accuracy Vehicular Ship Pedestrian Figure 6.4: Minimum symbol rate compatible with low-speed terminal motion (Ka-band) Figures 6.1 to 6.4 give some combinations of terminal speed and transmit symbol rates which are feasible within the DVB-RCS2 standard definition, for the defined typical gateway performance. The values obtained rely on the assumption that the maximum defined acceptable frequency offset is applicable for both initial burst (logon) and traffic and control bursts, assuming that no frequency correction is performed. These values could be reduced (and the range of applicability improved) in case tolerance for logon burst frequency offset is improved and frequency correction performed. In particular, the minimum rates for aeronautical applications could be reduced to better reflect application needs (512 kbits/s, 1 024 kbits/s). NOTE: When the speed and the targeted symbol rate of the mobile terminal are not within the defined envelope, operation may be facilitated by introducing some frequency Doppler pre-compensation mechanisms within the terminal (e.g. by using GPS location and by speed and direction information about the vehicle - aircraft for example-, or through frequency offset estimation deduced from forward downlink reception). The pre-compensation, which is not in the present document definition, would allow operating conditions similar to those obtained in the non-mobile environment. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.2.2.1 Frequency and timing drift within the burst | The frequency drift and timing drifts that will occur within the burst impact the burst demodulation performances, and may constrain the burst duration, thus the applicable burst formats and the profiles for some applications. The frequency drift is mainly due to the terminal acceleration. Assuming that the frequency is estimated on the preamble, frequency drift will induce phase rotation within the burst, with a maximum value on the last symbol of the burst. Assuming a typical value of 4° maximum phase rotation for acceptable degradation, a maximum burst duration for each mobile application can be defined. NOTE: The above paragraph is a worst case assumption. Implementations exist where frequency detection is made on the whole burst or where phase tracking can be made over the burst. In that case, the phase rotation on any symbol within the burst can be relaxed significantly. Tables 6.5 and 6.6 provide the worst case maximum burst duration values for both Ku-band and Ka-band, considering the frequency drifts defined in Tables 6.1 and 6.2. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 49 Table 6.5: Example of maximum burst duration for acceptable impact of frequency drift on the return link (Ku-band) Type of mobile Uplink Frequency Drift (Hz/s) Maximum burst duration (ms) Pedestrian 48 21,4 Maritime 242 9,6 Vehicular (bus, car, truck) 483 6,8 Train 242 9,6 Aeronautical (< speed of sound) 822 5,2 NOTE: The condition for acceptable impact of frequency drift on a burst is a 4° maximum phase drift. Table 6.6: Example of maximum burst duration for acceptable impact of frequency drift on the return link (Ka-band) Type of mobile Uplink Frequency Drift (Hz/s) Maximum burst duration (ms) Pedestrian 100 14,9 Maritime 500 6,7 Vehicular (bus, car, truck) 1 000 4,7 Train 500 6,7 Aeronautical (< speed of sound) 1 700 3,6 NOTE: The condition for acceptable impact of frequency drift on a burst is a 4° maximum phase drift. Adequate burst formats should be selected in order to remain within the above constraints. Concerning time drift, it is assumed that the timing drift resulting from both symbol timing inaccuracy and Doppler Effect should not induce a timing error of any symbol within the burst higher than of 0,1 symbol duration. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 50 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3 Physical Layer | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.1 Turbo-Phi Encoder | The Turbo-Phi encoder architecture used for linear modulation in ETSI EN 301 545-2 [i.1] is shown in Figure 6.5. It is based on a parallel concatenation of two double-binary Recursive Systematic Convolutional (RSC) encoders, fed by blocks of K bits (N = K/2 couples). For this encoder, N should not be a multiple of the linear shift register period (i.e. 15 for the 16-state code). N should be a multiple of 4, because of the permutation law Π. NOTE: Redundancy y2 is only used for coding rates less than 1/2. Figure 6.5: Structure of the proposed 16-state double-binary turbo encoder The encoding of a block involves encoding the information sequence in the natural order (switch in position "1"), permuting the data (interleaver Π) and encoding the information sequence in the natural order (switch in position "2"). At the end of the encoding process, the final state is the same as the initial state. Such a code can be represented using a circular trellis. For each block of user bits each component encoder computes a state, denoted the circular state, in such a way that the trellis "tail-bites" itself, making the trellis circular: by initializing the encoder's shift register with this circular state, the shift register terminates in the same state when all user bits have been fed into the encoder. The permutations (i.e. interleavers) between the component convolutional codes is based on simple algebraic laws, avoiding the use of memory-consuming look-up tables for the permutations. The laws are independent of the code rates and have been fine-tuned for each block size to avoid flattening of the error curve for BER above 10-9. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.2 Assessing the SISO decoder complexity | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.2.0 SISO Decoder Structure | Figure 6.6 gives the generic processing engine of an associated turbo decoder. This engine is built around two Soft-In/Soft-Out modules (SISO). The SISO are identical in structure, however, as inputs, one receives data in the natural order and the other one in the interleaved order. The outputs of one SISO, after proper scaling and after reordering, are used by its dual SISO in the next step. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 51 Figure 6.6: The principle of the turbo decoding |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.2.1 Max-Log-MAP decoding of an RSC code | In practical hardware turbo decoders, the so-called Max-Log-MAP algorithm [i.48] is adopted to implement the SISO decoders. The Max-Log-MAP algorithm is derived from a simplified version of the symbol-by-symbol Maximum A Posteriori (MAP) algorithm, also known as the Bahl, Cock, Jelinek and Raviv (BCJR) algorithm [i.41] in the logarithmic domain. The BCJR algorithm uses the trellis of the code to computes, for each data symbol, an A Posteriori Probability (APP) by evaluating the probabilities of all possible paths from the initial state to the final state in the trellis. Metrics computed in practice in this algorithm are proportional to the probabilities computed in the original MAP algorithm: 2 log met p σ = − (1) Moreover, the so-called Max-Log approximation is adopted: ln(exp( ) exp( )) max( , ) a b a b + ≈ (2) the max operator becoming a min one in practice due to the minus sign in (1). Consider a Recursive Systematic Convolutional (RSC) code with the following parameters: • ν is the memory length of the code; • m is the size of the input symbols at the encoder input (the code is said to be m-binary); • n is the number of coded bits provided by the encoder at each trellis stage, when no puncturing is performed. The kth m-binary information symbol ,1 , ( ) k k k m d d = d L at the encoder input is usually represented by the scalar quantity 1 , 1 2 m j k k j j d δ − = = , taking values between 0 and 2 1 m −, and writing k kδ ≡ d . The whole information sequence 0 1 ( ) N − d d L is denoted by d. Denoting by 0 1 ( ) N − = u u u L the sequence of corresponding encoded and modulated symbols ( ,1 , ( ) k k k n u u = u L with , 1 k l u = ± ) and by 0 1 ( ) N − = v v v L the sequence observed at the channel output or equivalently at the decoder input ( ,1 , ( ) k k k n v v = v L ). Here, sequence v is a concatenation of the sequences Ls and Lp. Symbols , k l u and , k l v correspond to systematic bits for l m ≤ and to parity bits for l > m. La (i,1) SISO 2 Π-1 Π Π Π-1 SISO 1 La (i,2) Le (i,2) Le (i,1) L (i,2) L (i,1) Lp (1) Lp (2) Ls (1) Ls (2) Hard decision La (i+1,1) ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 52 The Max-Log-MAP algorithm purpose is the computation of the a posteriori log-likelihoods ( ) k L δ defined as: 2 ( ) logPr( ) 2 k k L σ δ δ = − ≡ d v , 0, ,2 1 m δ = − L (3) The decoding algorithm involves: • The computation of branch metrics ( , ) kc s s ′ at each time step k is written as: , , 1 ( , ) 2 ( ) . ( , ) m a e k k k l k l k l c s s L v u c s s δ = ′ ′ = − + (4) • The computation of forward and backward state metrics for each trellis state s at each time step k: The state metrics are computed recursively using the following relations: Forward recursion: 1 1 ( ) min( ( ) ( , )) k k k s a s a s c s s − − ′ ′ ′ = + for 1, , k N = L (5) Backward recursion: 1 ( ) min( ( ) ( , )) k k k s b s b s c s s + ′ ′ ′ = + for 1, ,0 k N = −L (6) • The computation of the a posteriori log-likelihood ( ) k L δ at each time step k: Denote by ( ) kλ δ the soft information defined as: ( ) 1 ( , ) ( ) min ( ) ( , ) ( ) k k k k s s a s c s s b s λ δ + ′ ′ ′ = + + (7) the a posteriori log-likelihood related to data vector k d is computed as: ( ) 1 ( ) ( ) min ( ) 2 k k k L δ δ λ δ λ δ ′ ′ = − (8) • The computation of the hard decision at each time step k: - Actually, the hard decision provided by the decoder correspond to the binary representation of δ that minimizes ( ) kλ δ and makes ( ) k L δ equal to zero. ( ) ( ) ˆ arg min ( ) arg min ( ) k k L δ δ δ δ λ δ = = (9) • The computation of the extrinsic information ( ) e k L δ at each time step k: - The extrinsic information computation is similar to the computation of the a posteriori log-likelihood ( ) k L δ , using the extrinsic branch metrics. The extrinsic soft information ( ) e k λ δ defined as: ( ) 1 ( , ) ( ) min ( ) ( , ) ( ) e e k k k k s s a s c s s b s λ δ + ′ ′ ′ = + + (10) is computed, - and then the extrinsic log-likelihood ( ) e k L δ as follows: ( ) 1 ˆ ( ) ( ) ( ) 2 e e k k k L δ λ δ λ δ = − (11) - The term subtracted to ( ) e k λ δ is the extrinsic value corresponding to the hard decision ˆδ . ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 53 Due to the application of the Max-Log approximation (2), there is a systematic overestimation of all metrics and this algorithm turns to be sub-optimal. In order to cope with this problem, a scaling operation of the extrinsic information is performed in the turbo decoder: the value 0,7 is used except for the last iteration where extrinsic information is not scaled. Note that with the definitions of the metrics given in (1), the Max-Log-MAP algorithm does not need the estimation of the noise variance 2 σ . |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.2.2 Computational complexity of the Max-Log-MAP algorithm | This clause lists basic operations involved in the algorithm, such as additions, comparisons, etc. Branch metrics computation The trellis of an m-binary RSC code with memory ν is composed of 2ν with 2m transitions starting from and arriving to each state. At each step k, there are 2n different branch metrics to compute, corresponding to the 2n possible values of each vector k u . In (4), term , , 1 . n k l k l l v u = can be actually written as , 1 n k l l v = ± . The computation of all combinations of 1 2 , , k l k l v v ± ± requires 4 additions: 1 2 , , k l k l v v + , 1 2 , , k l k l v v − , ( ) 1 2 , , k l k l v v − + and ( ) 1 2 , , k l k l v v − − (the inversion of a number is similar to an addition). Hence, all branch metrics of (5) can be computed with an ( 1) n − -stage addition tree, resulting in a total of ( ) 1 4 2 1 n−− additions. The addition of the a priori term requires 2m extra additions and 2m multiplications by 2. The computation of the branch metrics can be performed twice, once for the forward recursion and once for the backward recursion or can be performed once and the metrics have then to be stored. State metrics computation The update of one forward state metric according to (5) involves the comparison and selection of 2m concurrent paths that can be implemented using 2m additions and 2 1 m − comparison-selection operations, implementing a tree structure. The update of backward state metrics requires the same number of operations. Since the trellis has 2ν states, one recursion step (forward or backward) requires 2 m ν + additions and ( ) 2 2 1 m ν − comparison-selection operations. A posteriori log-likelihood and hard decision computation The computation of the a posteriori log-likelihoods according to (7) and (8) requires the computation of the 2m values of ( ) kλ δ , 0, ,2 1 m δ = − L . Relation (10) involves two additions for each transition in the trellis. This complexity can be reduced to one addition by observing that partial terms ( ) ( , ) k k a s c s s ′ ′ + or 1 ( , ) ( ) k k c s s b s + ′ + are already available through the forward or backward recursion. For each value of δ, the minimum value of the 2ν terms 1 ( ) ( , ) ( ) k k k a s c s s b s + ′ ′ + + has to be found, resulting in 2 1 ν − compare-select operations, using a tree structure. Consequently, the computation of the 2m values for ( ) kλ δ requires 2m ν + additions and ( ) 2 2 1 m ν − comparisons and selections. The computation of the 2m a posteriori log-likelihoods requires a compare and select tree to compute the min term in (11), that is 2 1 m − compare and select operations, 2m subtractions and 2m divisions by 2. Actually, the subtraction in the case of ˆ( ) kλ δ can be avoided, since ˆ ( ) 0 kL δ = and the number of subtractions can be reduced to 2 1 m −. The hard decision ˆδ can be inferred from the compare and select tree allowing the minimum value of ( ) kλ δ to be computed. Extrinsic information computation For each symbol δ, the extrinsic information is computed with (10). Assuming that the terms , , / 1 1 ( ) .. 2 k m a k k l k l l L v u δ δ ≡ = − d have already been made available during the branch metric computation step using the decomposition proposed in (7), each piece of extrinsic information is obtained with one addition and one subtraction. The total extrinsic information computation is then performed using 2m additions and 2m subtractions. Note that, in practice, the decoder does not ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 54 need to provide the complete a posteriori log-likelihoods, since only extrinsic pieces of information are exchanged between the component decoders and only the hardware decision needs to be known at the end of the iterative process. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.2.3 Summary of Turbo Decoder complexity | Table 6.7 summarizes the resulting complexity for the process of a trellis stage, or equivalently of an information symbol. The complexity of the multiplications or divisions by the factor 2 can be neglected, since they can be implemented by a simple shift to the left or to the right. Adding the respective complexities of the computation of the branch metrics, the forward and backward recursions, the soft and binary outputs, and the extrinsic log-likelihoods, leads to a complexity of ( ) ( ) 1 1 2 2 4 8 2 1 1 m n ν + − + + − − additions and 1 2 2 2 2 m m ν ν + + + − − compare-select operations. For the DVB-RCS code, the code parameters are 3 ν = (8-state code), 2 m= (double-binary code), 4 n= (RSC rate ≥ 1/3). The computational complexity amounts to 171 additions and 79 compare-select operations. For the DVB-RCS2 code, the code parameters are ( 4 ν= , 2 m= , 4 n= ), 267 additions and 159 compare-select operations are required. Table 6.7: Computational complexity of the Max-Log-MAP algorithm Add (or subtract) Compare-select Mul or div by 2 Branch metrics (forward or backward) ( ) 1 4 2 1 2 n m −− + 2m One step of recursion (forward or backward) 2 m ν + ( ) 2 2 1 m ν − Computation of ( ) kλ δ 2 m ν + ( ) 2 2 1 m ν − A posteriori Log-likelihoods and hard decision 2 1 m − 2 1 m − 2m Extrinsic Log-likelihoods 1 2m+ Total computational requirement per information symbol (or for m information bits) ( ) 1 3 2 5 2 8 2 1 1 m m n ν + − × + × + − − ( ) 2 3 2 2 1 m ν × − − 1 2m+ |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.3 CPM complexity | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.3.0 Introduction | This clause provides an overview of algorithmic complexity of implementing CPM scheme at the transmitter and the receiver. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.3.1 CC-CPM Modulator | Two binary, non-systematic, non-recursive convolutional codes have been selected. One can easily shift from 5/7 to 15/17 encoders as shown in Figure 6.7. The multiplexer can be controlled with the PrecoderFlag signal. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 55 Figure 6.7: CC encoder structure |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.3.2 CPM receiver | Figure 6.8 shows the architectural diagram of a simplified receiver. ADC + Digital Channelizer Burst Detector (UW Detector) Freq. Recovery Fine Timing Recovery Matched Filter Branch Metric Unit BCJR + Phase Tracker SISO Decoder Figure 6.8: CPM receiver architecture Channelization Block This channelization block represents digital down conversion from low Intermediate Frequency (IF) band and separate the desired signal from adjacent signal through Low Pass Filtering (LPF). The LPF is a significant factor of the system because it influences the performance when adjacent carrier interference is considered. In particular, the passband and stop frequencies have been selected according to the Power Spectral Density (PSD) of the signal, which depends on the modulation parameters such as CPM encoder memory length (L), Alphabet Size (M) and Modulation index (h). When the configurations with high spectral efficiencies are considered, the cut off frequency of the filter should be chosen in order to find a good trade-off between the amount of interference coming from the adjacent channels and the amount of signal power of the useful carrier after the filtering. Synchronization Block The synchronization block is composed with Unique Word (UW) detection, carrier frequency offset recovery and fine timing recovery. By using two distributed pilot symbol like preamble and midamble, it enables to obtain good performance through special correlator. It can be designed to estimate the timing offset and carrier frequency offset through Fast Fourier Transform (FFT) computation, jointly. Among various possible approaches, the use of Data-Aided (DA) frequency estimator based on the Rife and Boorstyn (R&B) algorithm [i.42] can be considered. The received signal in AWGN channel can be expressed as: [ ] ) ( ) ,( ) ( ) ( 2 t n e t s t r t t j + = +θ πν α PrecoderFlag =0 5/7 FEC encoder 15/17 FEC encoder PrecoderFlag=1 b1 (polynomial 2) ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 56 where ν is the unknown frequency offset, and θ(t) corresponds to the carrier phase-noise. The frequency estimation technique consists of the following steps: a) Correlation functions Over the two known fields (i.e. the preamble and the midamble), the following correlation sequences are calculated: + + + + + + − = = − = = T T D N n T D N n mid mid n T nT nT pre pre n pre pre N n dt t s t r z N n dt t s t r z ) ( ) ( * * 1 ,... 1,0 ) , ( ) ( 1 ,... 1,0 ) , ( ) ( α α where the parameter D takes into account for Ndist and the number of "normalization sequence" symbols. b) FFT computation The FFT is applied on both sequences as follows: − + + = + + − − + + = + + − = = 1 ) ( 0 ) ( 2 1 ) ( 0 ) ( 2 mid pre mid pre mid pre mid pre N D N n N D N kn j mid n mid k N D N n N D N kn j pre n pre k e z Z e z Z ρ ρ π ρ ρ π where ρ is the R&B pruning factor. The incoming sequences are zero-padded, so that they have the same length, ρ(Npre+D+Nmid). c) Sequence combination The two frequency-domain sequences are combined in order to obtain the following decision vector: 1 ) ,...( 1,0 ) ( ) ( 2 − + + = + = + + + − mid pre N D N D N k j mid k pre k k N D N k e Z Z Z mid pre pre ρ π d) Search for the maximum value Finally, the value kˆ of k corresponding to the maximum value of 2 k Z is selected. Then, the carrier frequency estimate is computed as: T N D N k mid pre ) ( ˆ ˆ + + = ρ ν Matched Filter Block The most convenient approach for deriving low complexity algorithms for CPM digital demodulation consists of resorting to proper approximations of the original CPM waveform. In particular, the adopted technique is based on the Laurent decomposition [i.52]. The complex envelope of a CPM signal could be expressed as: − = − = 1 0 , ) ( ) , ( F k n k n k nT t p t s β α where F=(M−1) × 2(L−1)logM is the number of linearly modulated pulses pk(t), and βk,n are the so-called pseudo symbols. The exact expressions of pulses pk(t) and those of symbols βk,n as a function of the CPM parameters and of the information symbols can be found in [i.52]. For reducing the demodulation complexity, only S<F conditional terms are considered in the CPM front-end linear filtering. The first M-1 modulated pulses are called principal components, and should be sufficient to collect almost all the transmitted energy for L<3. It is assumed S=(M-1)ML-1. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 57 The Laruent filters can be implemented as Finite Impulse Response (FIR) filters. The complexity is determined by the number of selected filter among the number of complete filters. When the impulse responses of two components are very close to one another, the two components can be implemented as a single filter. Computation of branch metrics and BCJR algorithm The branch metrics computation is a demanding task in the CPM receiver. It requires ) ( filter selected of number the Nc ℎ complex multiplications per branch over CPM detection trellis with the forward and backward computations for each branch. One trellis section has p size Alphbet M × ) ( ℎ (Modulation index) branches if only principal components are selected, and p M Memory L × ) ( if more components are considered. Overall, it requires Nc p M Memory L × × ) ( complex multiplication per trellis section during one iteration. One trellis section has p M × branches when principal components only are selected and p M × 2 if more components are taken into account. Phase Tracking In the presence of phase-noise, the phase synchronization can be embedded in the BCJR algorithm [i.41] by using the Bayesian approach, and consists of assuming a probabilistic model for the phase noise process. In practice, the algorithm is obtained by discretizing the channel phase process in only R possible phase values: { } 1 / 2 − = R o i R iπ . The proposed value of R depends on the value of p, i.e. the denominator of the CPM modulation index h. The recommended values of R (as determined based on software simulation) are reported below. p R 5 6p 4 8p 3 8p 2 12p Any other value of p 4p More details on this algorithm and the derivation of the discretized phase estimation technique can be found in [i.53]. However, the discretized phase algorithm is computationally demanding. To reduce the complexity, the simplified algorithm can be considered, as described in [i.54]. However, the complexity reduction will cause performance degradation (1 dB performance degradation at PER = 10-3 has been reported based on software simulations). |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.3.4 Trellis Termination in CPM | Trellis termination is applied at the CPM modulator in order to force the modulator to a known state prior to inserting the Unique Word (UW) symbols. As an example, Figure 6.9 shows the location of the trellis termination symbols within a burst of symbols. Figure 6.9: CPM burst structure ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 58 Trellis termination ensures that the CPM waveforms generated using the UW symbols are data invariant, due to which they can be conveniently inferred/ reproduced at the receiver. The CPM signal phase can be completely specified using a Continuous Phase Encoder(CPE) followed by the Memoryless Modulator (MM) as shown in Figure 6.10. Figure 6.10: CPM Modulator as a continuous phase encoder followed by a memoryless modulator It should be noted that: • The CPE is a linear-time invariant sequential circuit. • The memory in the modulation is captured in the CPE which the MM uses to generate the CPM phase. • The number of states is hp M × . Also, the finite-state machine is time-invariant. • The phase trajectories in any two symbol intervals are time translates of one another [i.39]. Trellis termination involves driving the CPE to a known state, typically the all-zero state, the additional symbols required to do so are called the "tail-symbols". Due to its recursive nature, it is not possible to terminate the CPM trellis by transmitting L '0' tail-symbols. The tail-symbols will depend on the state of the encoder, after, the N data symbols are encoded by the CPE. Since the CPM state depends on the M, L and ph, the tail symbols will also depend on the choice of these parameters. Clause 7.3.7.2.3 of ETSI EN 301 545-2 [i.1], specifies the tail-symbols required for different selections of M, L and ph. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4 Return Link Encapsulation | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.1 RLE Principles | The Return Link Encapsulation (RLE) is designed to convey higher layer packets, such as IP datagrams, Ethernet frames, MPEG units or signalling packets, into a return link burst. As a design feature, the RLE protocol adopted for DVB-RCS2, is a modification of GSE (used on the forward link) but with lower overhead and specially tailored to the return-link properties. As such its implementation complexity is comparable if not slightly less than the overhead and complexity of GSE. Clause 6.4.1 and its subclauses describe the position of the protocol in the overall system and describe the protocol. Figure 6.11 illustrates an encapsulation example for the return link. IP datagrams (or other network layer protocol units) are fragmented according to the payload size of the bursts allocated by the resource management and according to scheduling decisions which may be driven by QoS requirements. Each fragment is put into an RLE packet which starts with an RLE header followed by data and one or more of the resulting RLE packets are packed into a return link burst and transmitted. Figure 6.11 shows several features of the protocol. The first, long, IP packet is followed by two short ones. It is assumed that the long one has lower priority and the second and third a higher. As can be seen the fragmentation of the first one can be suspended and the second and third will be sent before the first packet is resumed. Also the physical layer bursts have different sizes (in terms of payload bits but not symbols), because the module and coding changed between the first burst and the second one. n u 1 n u − n V [ , ] n n n X u S = mod h p ( ( 1) ) nT t n T ψ < ≤ + CPE MM n S D D ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 59 Figure 6.11: Return Link Encapsulation Example |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.2 RLE Interfaces | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.2.0 Overview | This clause provides informative material on conceptual RLE transmitter interfaces. The interfaces are conceptual because the description does not provide an implementation guideline. Rather logical interfaces that correspond to information flows between RLE and other modules of the terminal are described. These interfaces may or may not map to actual, physical interfaces in an implementation. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.2.1 RLE transmitter external interfaces | The RLE transmitter is located in the terminal. Its conceptual interfaces are shown in Figure 6.12. It is assumed that RLE is integrated with the scheduler and the scheduler queues for optimum performance. The functions of the interfaces are the following: • L2-IH: This is the input interface from the higher layers. RLE receives higher layer packets (mainly IP packets) with some attached information: - The protocol type. This will be IPv4 (0x0800) or IPv6 (0x86dd), but other types are also possible (see Table 5-1 in ETSI EN 301 545-2 [i.1]). - The packet label. This is a one-byte value containing the upper byte of the MAC24 to send the packet from. - A list of extension headers. This list may be empty. - The higher layer packet. For some types of extension headers this may be empty. - A QoS tag. This optional tag is used by the scheduler and may contain traffic classes, priorities. The RLE module encapsulates the packet and puts the result into one of the scheduler queues. • L2-IS: This is the interface for L2 M&C signalling packets that have to be sent to the hub. The protocol type of these packets is fixed to 0x0082, there may be no extension headers and no label. The RLE module puts these packets into one of its scheduling queues. • L2-MQ: This is the interface for an optional congestion control module. The module may monitor the queue fill states, fill rates and drain rates, derive congestion signals and trigger packet dropping or traffic shaping in the queues. • L2-MR: This is the interface to the request manager. The request manager monitors the fill states, fill rates and drain rates of the scheduler queues and produces resource requests to be sent to the hub. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 60 • L2-MB: On this interface the RLE module receives burst descriptions for the bursts it needs to produce. These descriptions contain the burst size for use by the burst packing sub-layer and may contain additional information to be passed to the PHY (timing, frequency, modulation, coding rate, etc.). For random access bursts the burst label contents are also required. • The information comes on one hand directly from the decoded tables in the forward link (DAMA slots) and from the random access management which uses also these tables and other signalling information to produce descriptions for random access bursts. • L2-MN: Interface to monitoring and fault management. On this interface the management module retrieves statistical information and error signals. • L2-CF: Configuration interface. This is used by the terminal infrastructure to configure the RLE module and the scheduler. • L2-OB: Burst output. This interface delivers burst payloads to the physical layer. Figure 6.12: RLE transmitter conceptual interfaces |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.2.2 RLE transmitter internal interfaces | Figure 6.13 illustrates a conceptual RLE transmitter structure. This structure employs an integrated approach doing joint scheduling; encapsulation and queue management. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 61 The higher layer packets (including signalling packets) are encapsulated and then put into scheduler queues according to their QoS tag. These queues are monitored by the request manager via the L2-MR interfaces and managed by the active queue management which is controlled from congestion control on the L2-MQ interface. For each possible fragment Id there is one fragmentation context. Each fragmentation context includes at least a fragmentation buffer and the next_fragment_id sequence number. The scheduler in each step selects either one of the busy fragmentation contexts or a packet from one of the input queues to produce the next RLE packet. If a packet from a queue is selected and it does not fit completely into an RLE packet, it is put into an idle fragmentation context. The scheduler is triggered by the burst packing function when it fills a burst. The burst packing function initializes each burst with a signalling byte (if configured to do so) and a burst label (for random access bursts). It then loops around triggering the scheduler until either the scheduler cannot return any data or the burst is full. Then the burst payload is handed over to the L2-OB interface. The burst packing function is triggered over the L2-MB interface which carries burst descriptions with possibly attached information for the physical layer. Figure 6.13: Conceptual RLE transmitter structure |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.2.3 Internal structure of the receiver | The conceptual structure of the receiver is given in Figure 6.14. The received burst payloads are first split into RLE packets. This can be done by accessing just the Fragment_Length fields of the packets. Unfragmented packets (start and end indicator are both set) are then sent directly to the decapsulation, while fragmented packets are sent to the defragmentation engine. An optional filtering on the fragment label can also be applied here in some meshed scenarios (transparent mesh, fragment switching). ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 62 Figure 6.14: Conceptual structure of L2 (receiver) The defragmentation engine needs to manage fragmentation buffers. Each packet that is currently defragmented requires a buffer which consists of the actual data buffer and some data like the expected sequence number, the CRC flag. The buffers are keyed by the combination of the sender address (which is received together with the burst from the physical layer) and the fragment id - each pair of <sender, receiver> has its own id space. Given that after filtering the receiver should be equal to the terminal; this gives a fragment id space for each sender. For this reason it may be necessary to support dynamic management of these buffers. Once the packet is defragmented it is handed over to the decapsulator which chops off the protocol type, packet labels and extended headers, possibly inserting default values or decompressing values and dispatches the resulting L3 packet to either the L3 stack or the signalling stack. The conceptual interfaces of the RLE block are as follows: • L2-IB: On this interface the RLE block receives burst payloads from the physical layer. • L2-OIP: Reassembled and decapsulated L3 packets are sent upstream from the RLE block to the higher layers. • L2-OS: Reassembled and decapsulated signalling packets are sent upstream from the RLE block to the signalling modules. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.3 RLE Implementation Guidelines | The Return Link Encapsulation protocol (RLE) can be sub-divided into three distinct sub-layers as shown in Figure 6.15. The upper layer (encapsulation) takes higher layer packets together with information about extension headers and the associated protocol type (Ethertype) of the packet and produces encapsulated packets. The fragmentation layer takes these and produces RLE packets that contain parts of the encapsulated packets or entire encapsulated packets. The burst packing layer finally produces physical layer burst payloads by taking one or more RLE packets optionally prepending a label and a signalling byte and optionally appending padding and a CRC. These burst payloads are then handed over to the Physical Layer (PHY) for further processing. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 63 On the reception side the processing is in opposite order: the signalling byte, the burst label, the CRC and padding (if any of them is present) are stripped from the burst payloads by the burst unpacking layer. It then splits off the RLE packets. The fragmentation process reassembles encapsulated packets from the RLE packets and hands them over to the decapsulation where the protocol type, the extension headers and the Layer 3 payload are reconstructed. The result of the decapsulation process is given to the upper layers. The following terminology is slightly different from the terminology in the RCS2 LLS for easier understanding. Table 6.8 provides the mapping between the terms used in ETSI EN 301 545-2 [i.1] and the terms used in the present document. Figure 6.15: RLE sublayers Table 6.8: Terminology mapping for RLE LL normative text LL Section ([i.1]) Guidelines Comments SDU 7.1.1 higher layer packet usually an IP or a signalling packet; the SDU also may include extension headers ALPDU 7.2.1 encapsulated higher layer packet higher layer packet with encoded extension headers, packet label and protocol type attached (all of them may be empty) ALPDU label packet label label attached to the higher layer packet (MAC address or SVN number) PPDU 7.2.2 RLE packet a packet containing a fragment of an encapsulated higher layer packet or a complete encapsulated packet PPDU label fragment label a label attached to an RLE packet; not used in RCS2 LL; zero length Frame PDU 7.2.3 burst payload the payload contents of a physical layer burst Payload_label burst label a label attached to a physical layer burst; can contain addresses or CRDSA information Payload header map signalling byte the optional first byte of the burst payload signalling the length of the burst and fragment labels Figure 6.16 provides the general format of the RLE packets as specified in ETSI EN 301 545-2 [i.1]. The green fields constitute the original higher layer packet (an IP packet for example). The red fields (except for the sequence number and the CRC) together with the green fields are the encapsulated higher layer packet and the green, the red and the blue fields together constitute RLE packets. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 64 In the RLE burst payload (see Figure 6.17) zero or more RLE packets (green) are concatenated. They optionally may be prefixed with a signalling byte and a burst label. The remaining space in the burst that is not used by RLE packets is filled with padding and the last four bytes may optionally be occupied by a burst CRC (in ETSI EN 301 545-2 [i.1]) this CRC is not used because there is another CRC below the spreading layer). Bit padding at the end occurs if the payload length required by the physical layer is not a multiple of 8 bit. This has been avoided in RCS2 by making all burst definitions have multiple of 8 bits. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 65 S 1 E 0 RLE_Packet_Length ID 1b 1b 11b 3b Protocol_Type (opt) Ext headers (opt) 13b 16b variable Packet_Label (opt) 0..15B S 1 E 1 RLE_Packet_Length 1b 1b 11b Packet_Label (opt) 0..15B Protocol_Type (opt) Ext headers (opt) 16b variable S 0 E 0 RLE_Packet_Length 1b 1b 11b S 0 E 1 RLE_Packet_Length 1b 1b 11b Data (opt) SeqNo 8b LT 2b ID 3b ID 3b Data Data (opt) Data (opt) variable variable variable Fragment_Label (opt) Fragment_Label (opt) Fragment_Label (opt) Fragment_Label (opt) 1b T 0..15B LT 2b 1b T 0..15B Total length 0..15B 0..15B variable S 0 E 1 RLE_Packet_Length 1b 1b 11b Data (opt) CRC32 32b ID 3b Fragment_Label (opt) 0..15B variable Figure 6.16: RLE packet formats Figure 6.17: RLE burst payload format ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 66 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4 RLE Profile Specification | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.0 Overview | The RLE protocol has a number of features that are configured and used differently for different usage scenarios and different instances in a DVB-RCS2 system. For each of the uses cases the fixed parameters and options as well as the configurable options are collected into an RLE profile. This clause provides the RLE profile for a transparent star configuration. It additionally provides some hints and guidelines on designing profiles for transparent mesh and regenerative systems. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1 Transparent star profile | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.1 Higher layer packets | Higher layer packet is completely mapped into the data field of the encapsulated packet. Other means of transporting higher layer packets like the PDU concatenation extension header are not required by the LLS and should be used only when the transmitter knows that the receiver supports this extension header. Examples of higher layer packets are IPv4, IPv6 or signalling packets. The total length of the higher layer packet, extension headers, the packet label and the (maybe compressed) protocol type cannot exceed 4 095 bytes. The upper layer MTU should be set to a value that takes into account the maximum of 2 byte for the protocol type, the maximum packet label and the maximum length of extension headers that is supported. The maximum packet label length depends on the transmission context (see Table 7-10 in ETSI EN 301 545-2 [i.1]). NOTE: If during encapsulation of a packet more than 4 095 bytes are produced the packet should be dropped. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.2 Extension headers | Extension headers are specified in IETF RFC 4326 [i.71], IETF RFC 5163 [i.72] and in ETSI EN 301 790 (V1.5.1) [i.46]. The list of allocated header fields is defined in the Internet Assigned Numbers Agency (IANA) Next-Header Registry, located at: http://www.iana.org/assignments/ule-next-headers/ule-next-headers.xml. The list of values recommended for DVB-RCS2 is provided in ETSI EN 301 545-2 [i.1]. Extension headers are encoded in the same way as in GSE and are fragmentable. If the length of the (possibly compressed or defaulted) protocol type, the packet label, the extension headers and the higher layer packet exceed 4 095 bytes, the complete packet should be dropped by the encapsulator. Care should be taken when using extension headers, especially mandatory headers. The term mandatory in this case does not require that the extension header is supported, but that if the receiver finds such a header and it does not support it, the entire encapsulated packet should be dropped. The LLS does not require support for specific extension headers so other means should be used to determine whether to send extension headers or not. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.3 Packet label (ALPDU label) | The RLE protocol generically allows configuring the length of the packet label for each of the four label types to have a length from 0 to 15 bytes. For RCS2 three of the label lengths are fixed and the fourth one should be specified by the hub to be 1 byte (Table 6.9). Table 6.9: RLE label types Label type Packet label size Default protocol type Comment 0 1 implicit_protocol_type of Frame payload format descriptor type_0_alpdu_label_size of the Frame payload format descriptor 1 3 implicit_protocol_type of Frame payload format descriptor 2 0 implicit_protocol_type of Frame payload format descriptor 3 0 0x0082/0x42 - Internal M&C signalling (L2S) default protocol type different from label type 2 ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 67 Label type 3 should be used by the terminal to send signalling information. The default protocol type for label type 3 is the protocol type used for signalling (0x0082 uncompressed/0x42 compressed) and thus can be omitted. The default label type for the other three label types is indicated by the hub in the implicit_protocol_type field of the frame_payload_format_descriptor. For normal RCS2 operation label type 0 is used for higher layer traffic on the non-default SVN. In this case the one byte SVN tag needs to be inserted into the packet label. For a higher layer packet on the default SVN the tag may be omitted by using label type 2. Use of label type 1 is for compatibility with on-the-fly translation to GSE and is not mandated by ETSI EN 301 545-2 [i.1]. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.4 Protocol type | The protocol type specifies the type of higher layer packet (for example IPv4) or the type of the first extension header. In order to save bandwidth the protocol type may be defaulted or compressed. The default protocol types are defined as follows: • The implicit compressed default protocol type for label type 3 is 0x42 which is the RCS2 internal L2 signalling. This is not changeable. • The implicit compressed default protocol type for the other label types is indicated by the NCC in the frame_payload_format_descriptor field implicit_protocol_type. The protocol type encoding is done based on the values signalled by the NCC as follows (as shown in Figure 6.15): • The allow_ptype_omission flag allows using the default protocol type. If a higher layer packet is to be transmitted and there is a label type which has a default protocol type equal to the protocol type of the packet and a packet label length equal to the packet label of the packet, then that label type should be used for the encapsulation of the packet. In this case the protocol_type field of the encapsulated packet (ALPDU) should be omitted and the protocol_type_suppressed field in the first RLE packet (PPDU) containing parts or the entire encapsulated packet is set to 1. • If allow_ptype_omission is not set or there is no suitable label type that allows the omission of the protocol type then, if the use_compressed_ptype flag in the frame_payload_format_descriptor is cleared the two byte protocol type should be inserted into the encapsulated packet (high byte first) and the protocol_type_suppressed field in the first RLE packet (PPDU) containing parts or the entire encapsulated packet is cleared. • Otherwise if use_compressed_ptype is set and the protocol type table contains a compressed value for the protocol type, then the 8-bit compressed protocol type is inserted into the encapsulated packet and the protocol_type_suppressed flag in the first RLE packet (PPDU) containing parts or the entire encapsulated packet is cleared. • Otherwise if the use_compressed_ptype is set and the protocol type table does not contain a compressed value of the protocol type the extension mechanism should be used: a value of 0xff is inserted as compressed protocol type and the actual two byte protocol type is inserted (high byte first) directly behind the packet label. The protocol_type_suppressed in the first RLE packet (PPDU) containing parts or the entire encapsulated packet is cleared. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 68 Figure 6.18: Protocol type encoding |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.5 Total length field | The total length field is the sum of the length of the protocol type (this may be 0 for a suppressed protocol type, 1 for a compressed one, 2 for an uncompressed one or 3 for a compressed using the extension mechanism), the length of the packet label, the length of all extension headers and the length of higher layer packet. The value of the total length may not exceed 4 095. This field is used only if the transmission requires fragmentation of the packet. If the packet is transmitted in a COMPLETE RLE packet, no total length field is used because the packet length can be deduced from the RLE packet length. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.6 Protocol_type_suppressed field | This bit should be set depending on the protocol type processing (see clause 6.4.4.1.4). |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.7 Label type | This field is set depending on the packet label and protocol type processing (see clauses 6.4.4.1.3 and 6.4.4.1.4). |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.8 Fragment label (PPDU label) | This field is not used and has a zero length for the transparent scenario. This length (0) is signalled via the implicit_ppdu_label_size field of the frame_payload_format_descriptor. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.9 Fragment Id | This field is used to mark all RLE packets that contain fragments of the same higher layer packet with the same id so that the packet can be reassembled. The sender should use a round-robin algorithm to use the 8 available fragment ids: when fragmentation is required the search for the next fragmentation id is started from the fragmentation id from the last transmitted START packet plus 1 (in modulo 8 arithmetic). This ensures good spacing of the fragment IDs. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 69 No two currently fragmented higher layer packets should use the same fragment id. This means that a maximum of 8 higher layer packets can be fragmented in parallel at a given transmitter for the same receiver (there is only one return link receiver in a transparent star). |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.10 Sequence number | The use of the sequence number in contrast of the CRC for reassembly error checking is controlled by the NCC in the frame_payload_format_descriptor. If the allow_alpdu_sequence_number bit is set the terminal should use the sequence number. If the bit is not set, the allow_alpdu_crc should be set by the NCC and the terminal should use the CRC field instead. The NCC should set exactly one of these bits. For each receiver/fragment id pair the transmitter should maintain a variable next_sequence_number. For the transparent star case, where there is only one receiver, the transmitter should maintain 8 of these variables (one for each fragment id). These variables are initialized to 0. Whenever the transmitter produces an END RLE packet it inserts the current value of the next_sequence_number for the corresponding fragment id into the sequence number field of the packet and increments the next_sequence_number modulo 256. The use_alpdu_crc field of the corresponding START RLE packet should be cleared. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.11 CRC field | If the CRC field is to be used for reassembly error checking (as described in the previous clause), the CRC should be computed as follows: • initialize the 32-bit CRC with the value 0xffffffff • feed the following fields into the CRC algorithm (MSByte first LSBit first): - the original two-byte protocol type - the packet label - all extension headers - the higher layer packet • invert the computed value and insert the resulting 4 bytes MSByte first into the CRC field. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.1.12 Start_indicator, End_indicator and RLE packet length | The following values are set as required by the fragmentation algorithm in the same way as in GSE: Signalling byte (payload map) The use of the signalling byte is controlled by the NCC via the use_explicit_payload_map of the frame_payload_format_descriptor. The use of this field is not strictly necessary for RCS2, because the fragment label has always the length zero and the burst label has the same length for all burst transmitted from within a given context. Therefore the value of the signalling byte will be the same for all burst in that context. Burst label The burst label can have a size ranging from 0 to 15 bytes in RLE. As per ETSI EN 301 545-2 [i.1] however, the sizes are fixed for the DAMA, the slotted aloha and the CRDSA context (see Table 7-10 in ETSI EN 301 545-2 [i.1]). For dedicated access the only method to recover the terminal (source) address of a received burst is implicit recovery by matching the reception time of the burst to the burst-time allocation table. The RCST should produce only the label lengths specified in this table for strict standard conformance. The implicit_payload_label_size allows the specification of the label sizes only for traffic bursts (last row of Table 7-10 in ETSI EN 301 545-2 [i.1]); label sizes for the other burst types is fixed and is not signalled. An alternate method to identify the source terminal which simplifies the hub, but requires corresponding support in the RCST is to explicitly signal the source terminal in DAMA slots. In this case the same label lengths and contents as for the slotted aloha access should be used. Non-standard signalling is required in this case. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 70 Burst CRC ETSI EN 301 545-2 [i.1] does not use a burst CRC on the RLE burst packing layer, because a CRC is inserted below the spreading (see Figure 7-8 and clause 7.3.4 of ETSI EN 301 545-2 [i.1]). |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.4.2 Mesh and regenerative scenarios | This clause provides some general guidelines on how to select profile options for MESH and regenerative systems. The main problem is to select the right format for the burst, the fragment and the packet labels depending on the scenario and the on-board processing technology. Both meshed and regenerative scenarios require additional labels for (i) addressing and (ii) reassembly. In a transparent star scenario there is only one receiver on the return link by definition. Therefore, no destination addressing is required at the lower layers. The only kind of destination addressing used is the SVN tag which is (indirectly) transported in the higher layer packet label (ALPDU label). In a transparent star scenario since there are multiple transmitters sending to the single receiver, the receive addressing is required for reassembly. Since all transmitters use the same fragment ID space, the receiver needs to deduce this information, for example, an RLE packet with fragment ID 1 from terminal A should not be reassembled with an RLE packet with fragment ID 1 from terminal B (see Figure 6.19). For DAMA bursts this is done indirectly by matching the receive time of the burst with the actual transmission plan. This directly leads to the terminal that did send the burst. For random access bursts, the source address is contained in the burst label as specified by Table 7-10 of ETSI EN 301 545-2 [i.1]. Figure 6.19: Fragment demultiplexing at the RLE receiver In a mesh system the receiver needs to do the same kind of demultiplexing as the receiver in the transparent star. Source identification can be done in the following ways: ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 71 • Implicit by timing. As in the transparent star case the receiver knows the transmission plan and can correlate the receive time of a burst to the plan to find the sender. • Explicit in the burst label. This may be a MAC address, a logon id or a connection identifier depending on the system. In a regenerative system the necessity of source identification depends on the switching layer. If the OBP performs the reassembly and fragmentation then there will be only one transmitter received by a given terminal and in this case no source identification is necessary. If the OBP performs RLE to GSE translation or just uses RLE in the downlink source identification is necessary because the downlink will contain RLE packets from more than one sender. The source identification can be done in one of the following ways: • For GSE in the downlink the only option is to overload the semantics of the fragment id field. The OBP should translate the same RLE fragment id from different terminals to different GSE fragment ids. This maybe done either by using a connection identifier which will limit the number of connections in a downlink to 32 or by dynamically allocating fragment ids and relying on receiver filtering based on the packet label (then a maximum of 256 packets can be switched at the same time into a given down link). • For RLE in the downlink the fragment label can be used to carry the sender identification in form of a connection identifier or some address. In a regenerative case a label is also required for switching. This switching can be done on the burst label (in this case all RLE packets in the same burst will go to the same downlink or even to the same destination terminal), on the fragment label or on the packet label. In the latter case the OBP should either do reassembly and fragmentation to switch on encapsulated or IP packets, or it should implement label caching, i.e. read the label from the first RLE packet of a given higher layer packet, enter it together with the fragment id and the source id into a cache and use the cached value to switch all the subsequence RLE packets of this higher layer packet. The different options have different implications in terms of efficiency, overhead, required processing power, required signalling and limits on the number of terminals and connections. In the case of on-the-fly translation of RLE packets to GSE packets in an OBP the following RLE configuration should be used: • The label lengths of label type 0 should be set to 6. • The CRC should be used for reassembly checking. • The sender should not fragment the protocol type, the packet label (ALPDU label) or the CRC. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.5 RLE Reassembly Error Checking Guidelines | |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.5.0 Overview | The RLE protocol allows parallel fragmentation and reassembly of several higher layer packets from the same sender. This is achieved by combining the information in the Fragment_ID header field with either the Sequence_Number field (if the Use_Packet_CRC header field is cleared), or with the CRC32 field (if the Use_Packet_CRC header field is set to 1). The option to use a 1 byte sequence number instead of a 4 byte CRC32 is provided to reduce the packet overhead from 4 bytes to 1 byte in the cases that the environment allows for that. CRC32 should be used in environments with very large drop-outs (more than 255 bursts in a row, mobile environments, for example) or with OBPs that do conversion to GSE. It is recommended not to mix both options at a sender. On the one hand, the suitability of one or the other method is a trade-off between propagation impairments, system design (for example in the case of OBPs that do conversion to GSE) and overhead reduction; this allows selecting one method or the other for the sender once. On the other side, letting the sender switch from one to the other option includes higher complexity for scheduling RLE packets in a burst, as the minimum length of the End packet varies for different setting of the Use_Packet_CRC header field. The methods and rules to apply the sequence number and the CRC32 for reassembly error checking are described below. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 72 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.5.1 Reassembly error check algorithm with sequence number | If the sequence number is applied for reassembly error check, the following rules should be applied: • The sender has a variable next_sequence_number for every Fragment_ID from 0 to 7. When emitting an RLE END packet (PPDU) the fragmentation process appends the value of next_sequence_number to the encapsulated higher layer packet and increments the variable. When the value reaches 255, the next increment cycles back to 0. • The receiver has a variable next_sequence_number for every Fragment_ID. This variable contains the Sequence_Number expected in the next RLE END packet (PPDU). When the receiver receives such a fragment it does the following: - It compares the Sequence_Number from the fragment to the next_sequence_number for the given Fragment_ID. If they differ, the reassembled packet is thrown away. If they are equal the packet is passed to the higher layer. - It sets the next_sequence_number for the logical link to the value of Sequence_Number + 1 independently of whether the packet was thrown away in the previous step or not. If the new value is 256 it cycles back to 0. Both the fragmentation process and the defragmentation process initialize their next_sequence_number values for all Fragment_IDs to 0 at initialization time. If the sender sends a packet with a CRC32 the next_sequence_number is not incremented. When the receiver receives a packet with a CRC32, its next_sequence_number is not incremented. This is done so that in the case of the loss of a packet with a CRC32 the next packet with a sequence number is not lost too (a packet with a CRC32 cannot resynchronize the next expected sequence number in the receiver). Figure 6.20 illustrates how the reassembly error detection mechanism works: Figure 6.20: Reassembly error detection with sequence number In Figure 6.20 the fragmentation and reassembly of several packets on the same Fragment_ID is shown. The upper data flow shows the process without transmission errors and the lower one with two fragments lost (as denoted by the red crosses). The red (lower) numbers show the changes of the next_sequence_number in the fragmentation process, the green (upper) numbers the changes in the defragmentation process. In the error-less case the first fragment shown is the RLE END packet of a higher layer packet. The current next_sequence_number in the sender is 11 so this value is inserted into the packet trailer and the variable is incremented to 12. The receiver compares the trailer value (11) to its next_sequence_number (11). It finds them to be equal and delivers the packet to the upper layer. Then it sets its next_sequence_number to the trailer value plus 1 (12). The sender then fragments the next packet: an RLE START packet, an RLE intermediate packet and an RLE END packet. It inserts its current next_sequence_number (12) into the trailer and increments the variable to 13. The receiver finds the trailer (12) to equal the next_sequence_number value (12), delivers the packet and sets its next_sequence_number to 13. Then the sender processes the next packet the same way resulting in the next_sequence_numbers on both sides to become 14. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 73 In the error case an RLE START packet and an RLE END packet are lost. It is assumed, that the sum of the sizes of the other fragments happen to sum up to the correct Total_Length value (otherwise the error is detected already by the length check). When the sender sends the second RLE END packet it inserts 12 into the trailer and sets its next_sequence_number to 13. This fragment is lost so the receiver still has value 12. Now the sender sends the fragments for the next higher layer PDU. The trailer now contains 13 and the next_sequence_number is 14. The RLE START packet is also lost (otherwise the error would be detected earlier). When the RLE END packet finally arrives at the receiver it finds a Sequence_Number of 13 which is different from its next_sequence_number (12). Therefore it throws out the reassembled packet and sets the next_sequence_number to the trailer value plus 1 (14). This mechanism replaces the packet CRC of GSE. The benefit is a reduction in trailer size from 4 bytes to 1 byte. This is only one reassembly error detection mechanism. Other rules for reassembly error detection are: • If an RLE packet arrives which has the Start_Indicator 0, but no higher layer packet is currently reassembled for the given Fragment_ID, then an RLE START packet has been lost and the fragment is thrown away. If the RLE packet has the End_Indicator equal to 1, the next_sequence_number is updated with the trailer value plus 1. • If an RLE packet arrives which has the Start_Indicator 1, and a higher layer packet is currently reassembled on the same Fragment_ID, an RLE END packet has been lost. The currently reassembled higher layer packet is thrown away, the next_sequence_number is incremented and the RLE START packet is processed normally. • If the sum of the length of the received RLE packet and the already reassembled length is larger than the advertised Total_Length then at least an RLE START and an RLE END packet have been lost. Both the currently reassembled higher layer packet and the new RLE packet are thrown away. If the packet has the End_Indicator set to 1 and the Start_Indicator is 0, the next_sequence_number is updated with the trailer value plus 1. • If an RLE END packet is received and the sum of the Packet_Length and the length of the already buffered packet part does not match the expected Total_Length, the buffered higher layer packet and the new RLE packet are thrown away. The next_sequence_number is set to the trailer value plus one. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.5.2 Reassembly error check algorithm with CRC32 | If the CRC32 is applied for reassembly error check, the packet CRC uses the same algorithm as the burst CRC. It is, computed over the 16-bit protocol type (even if suppressed or compressed), the label, the extension headers and the higher layer data of the packet. The CRC uses the well known CCITT polynomial 0x104c11db7 with an initial value of 0xffffffff and a final negation of the result. The bit ordering for CRC calculation is as follows: bytes are taken most significant first. Within each byte, bits are taken least significant first. The resulting CRC is appended to the data field of the END packet with the most significant byte first and the highest order bit of the reminder in the most significant bit of the byte. The following rules should be applied for assembly error detection: • When emitting an RLE END fragment corresponding to a Fragment_ID, the fragmentation process appends the CRC32 (calculated as stated above) to the higher layer packet data. • When the receiver receives the RLE END packet of the currently assembled Fragment_ID it does the following: - It performs an integrity check by independently calculating the same CRC value over the fields indicated above of RLE packets with the same Fragment_ID as the received RLE END packet. - It compares the calculated CRC with the received value in the RLE END packet trailer. If they are not identical (integrity check fails), the reassembled packet with the same Fragment_ID as the RLE END packet is discarded. Otherwise, if the integrity check is successful, the reassembled higher layer packet is handed over to the decapsulator. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 74 Figure 6.21 illustrates how the reassembly error detection mechanism works: Figure 6.21: Reassembly error detection with CRC32 In Figure 6.21 the fragmentation and reassembly of several higher layer packets on the same Fragment_ID is shown. The upper data flow shows the process without transmission errors and the lower one with two fragments lost (as denoted by the red crosses). In the error-less case the first fragment shown is the RLE END packet of a higher layer packet (it is assumed that the previous RLE START and RLE intermediate packets are received correctly). The receiver will calculate the CRC and compare it successfully with the received one in the RLE END packet. The same will happen with the two following pairs of RLE START and RLE END packets, so that the reassembled higher layer packets will be handed over to the decapsulator. In the error case, the RLE END packet of the second higher layer packet and the RLE START packet of the third higher layer packet are lost. This means, that eventually the reassembly function could wrongly interpret that the consecutively received RLE START and RLE END packets belong to the same higher layer packet. If the sizes of the correctly received RLE START and RLE END packets do not match the Total_Length field in the RLE START packet, the error would be automatically detected. Otherwise, the CRC mechanism should detect the error by comparing the locally calculated CRC over the reassembled encapsulated packet with the CRC trailer contained in the RLE END packet. If they are not identical, the reassembly function at the receiver will discard both, the RLE START and the RLE END packet. Other general rules for reassembly error detection (previous to applying the CRC): • If an RLE packet arrives which has the Start_Indicator 0, but no higher layer packet is currently reassembled for the given Fragment_ID, then an RLE START packet has been lost and the new RLE packet is discarded. • If an RLE packet arrives which has the Start_Indicator 1, and a higher layer packet is currently reassembled on the same Fragment_ID, an RLE END packet has been lost. The currently reassembled higher layer packet is discarded and the new RLE packet is processed normally. • If the sum of the length of the received RLE packet and the already reassembled length is larger than the advertised Total_Length then at least an RLE START and an RLE END packet have been lost. Both the currently reassembled packet and the new RLE packet are discarded. • If an RLE END packet is received and the sum of the Packet_Length and the length of the already buffered higher layer packet part does not match the expected Total_Length, the buffered packet higher layer packet and the new RLE packet are discarded. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 75 |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.4.5.3 Reassembly process | This clause is an illustration of the reassembly process. It is provided to simplify the understanding of the protocol. In a transparent system with a star configuration the layer 2 processing in the gateway needs to handle burst streams from all terminals in the system. So there need to be distinct data structures for all the terminals in the system. After demodulation and decoding the resulting burst payload is first process by the upper sub-layer of the physical layer. If this is successful it is handed over to the layer 2. The processes go approximately as follows (implementation details abstracted, star system with no fragment labels assumed): • (PHY) If the burst is a DAMA burst get the source information from combining the TBTP information with the receive time of the burst and tag the source identifier (Terminal_Id) to the burst. • (L2 burst unpacking/PHY) If there is burst label, separate it from the data and process it (for example for random access). If the burst is a random access burst gets the source Terminal_Id from the label and tag it to the burst. • (L2 burst unpacking) Set a pointer to the start of the packet (after the label) and loop until the pointer points beyond the burst or at the last byte: If only one byte is left this byte is a padding, therefore the burst is already processed: - (L2 burst unpacking) From the next two bytes decode the Fragment_Length field. If this is zero, this is either the start of padding (with Start_Indicator and End_Indicator both zero) or an error. In any case this burst is done. - (L2 burst unpacking) Add 2 and the Fragment_Length to the current pointer. This new pointer will be the start of the next RLE packet used for the previous step in the next loop. - (L2 burst unpacking) Decode the Start_Indicator and the End_Indicator: (L2 burst unpacking) If both indicators are 1 this is a full RLE packet so hand over the entire payload of the RLE packet, the last four bits of the header (the Label_Type and the PType_Suppressed flag) and the Terminal_Id to the decapsulator. (L2 burst unpacking) Hand over the packet payload, the last four bits of the header (the Fragment_ID), the Start_Indicator, the End_Indicator and the source information (Terminal_Id) to the reassembler for the given Fragment_Id and terminal id. Each combination of Fragment_Id and terminal id defines one reassembly context storing all the data for a single packet reassembly (next_sequence_number variable, reassembly buffer and others). - (L2 burst unpacking) Set the current burst pointer to the pointer computed above and repeat the steps for the next RLE packet. The reassembler is basically a switch statement over the Start_Indicator and End_Indicator values: • Locate the reassembly context for the given combination of Terminal_ID (source information) and Fragment_ID. This context contains among other fields the reassembly buffer and the next_sequence_number field and is responsible to handle packets for one Fragmentation_ID of a single sender terminal. • If Start_Indicator=0 and End_Indicator=0 then (intermediate packet): - If Use_Packet_CRC is set to 0: Do the checks described in clause 4.1 and if they are ok append the fragment payload to the reassembly buffer. - If Use_Packet_CRC is set to 1: Do the checks described in clause 4.2 and if they are ok append the fragment payload to the reassembly buffer. ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 76 • If Start_Indicator=1 and End_Indicator=0 then (START packet): - Ensure that the RLE packet length is at least 2 and decode the Total_Length, the Label_Type, the Type_Suppressed flag and the Use_Packet_CRC flag. - If Use_Packet_CRC is set to 0: Do the checks from clause 4.1 and if they are ok allocate the reassembly buffer (if the implementation strategy requires this). Put the fragment payload into the reassembly buffer and remember the Label_Type, Type_Suppressed, Use_Packet_CRC flags and the Total_Length. - If Use_Packet_CRC is set to 1: Do the checks from clause 4.2 and if they are ok allocate the reassembly buffer (if the implementation strategy requires this). Put the fragment payload into the reassembly buffer and remember the Label_Type, Type_Suppressed, Use_Packet_CRC flags and the Total_Length. • If Start_Indicator=0 and End_Indicator=1 then (END packet): - If Use_Packet_CRC is set to 0: Do the checks from clause 4.1 and if they are ok append the fragment data to the reassembly buffer. Chop of the Sequence_Number from the buffer end and compare it with the next_sequence_number. If they are equal hand over the payload, the Label_Type field, the PType_Suppressed flag and the Terminal_Id to the decapsulator. Set the next_sequence_number to the value from the packet plus 1. - If Use_Packet_CRC is set to 1: Do the checks from clause 4.2 and if they are ok append the fragment data to the reassembly buffer. Calculate locally the CRC over the fields indicated in clause 4.2 of the RLE packets in the reassembly buffer. Compare the result with the CRC trailer in the RLE End packet. If they are equal hand over the payload, the Label_Type field, the PType_Suppressed flag and the Terminal_Id to the decapsulator. • If Start_Indicator=1 and End_Indicator=1 then (full packet): - This should not happen because this was processed at the burst unpacking layer The decapsulator receives the payload, the Label_Type field, the PType_Suppressed field and the Terminal_Id. Depending on the two flag values it parses the payload into the label, the protocol type (if necessary taking the default protocol type) and the extension headers and hands over everything together with the Terminal_Id to the higher layer functions. Conceptually there are three sublayers in this process: the lowest layer handles the packing and unpacking of RLE packet into and from bursts (including burst label and signalling byte if necessary), the middle layer handles just reassembly and needs to look only at the Terminal_ID, the two bytes from the fixed header, the first two bytes of the fragment payload in RLE START packets (the total packet length) and at the trailer in the RLE END packets. All other data is just appended to the buffer transparently. Once the packet is reassembled it is passed to the upper sublayer which does higher layer packet processing: getting the protocol type field, handling extended headers and the optional label. This mechanism also allows the header fragmentation, because the only variable header field actually needed during the processing of the reassembly sublayer is the total packet length. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.5 Demand Assignment in DVB-RCS2 | An RCST using dedicated access should send capacity requests to the NCC for the traffic to be sent on the DA-AC. These requests are made by the RCST using a Capacity Request (CR) message sent to the NC. The CR messages are generated by a Request Class (RC). ETSI ETSI TR 101 545-4 V1.2.1 (2026-01) 77 A RC describes a method to provide allocations for a DA-AC. An RCST may use more than one RC, each identified by an RC identifier (RC_INDEX). This RC_INDEX is included in the CR message sent to the NCC. On reception of a CR, the NCC makes the corresponding allocations, taking into consideration any information associated with the RC (such as relative priority of the CR or limits on resource usage). A set of corresponding allocations is made by the NCC in the TBTP2. These allocations are associated with a specific DA-AC, identified by mapping the assignment_id. The RC may be accessed from the HLS Service through an LL Service. One way to realize the required interaction is to feed information about the HLS PDU queue (BA) from the HL Service to the LL Service. This information could include the queue size and average BA arrival rate. The LL Service could then use this information to generate a CR for the BA, or a set of BAs that correspond to the same RC. An LL Service is expected to use a RC that is compatible with the QoS expectations of the HL Services that they support. This requires configuration of a set of RCs and a mapping between the BA (or its PHB) and the set of available RCs. The behaviour of an RC is defined by its usage of the set of capacity categories: CRA, RBDC, A/VBDC, FCA (defined below). Each RC can support any mix of capacity categories. One of three different sets of request options can be authorized for each RC. The mapping of the capacity category to a RC is configured for the RCST. The allocation process can support the following six allocation methods: • Constant Rate Assignment (CRA): Rate capacity, which is provided in full for each allocation while required. This capacity is not requested using a CR but may be associated with an RC. • Rate Based Dynamic Capacity (RBDC): Rate capacity, requested dynamically by the RCST using an RBDC CR under a specified RC. • Volume Based Dynamic Capacity (VBDC): Volume capacity, requested dynamically by the RCST using an VBDC CR under a specified RC. These requests are cumulative (i.e. each request adds to all previous requests from an RCST). • Absolute Volume Based Dynamic Capacity (AVBDC): Volume capacity requested dynamically by the RCST. The VBDC capacity is provided by the NCC in response to explicit CRs from an RCST. These requests are absolute (i.e. replace any previous VBDC CR). • Random Access Capacity (RA). Capacity assigned to a set of RCSTs for shared access. This capacity is not requested using a CR. • Free Capacity Assignment (FCA): Volume capacity assigned to an RCST from the free capacity that would be otherwise unused. This capacity is not requested using a CR. The behaviour of each RC in terms of the type and amount of capacity requested is configurable and may use for any desired combination of (A)VBDC, RBDC and CRA. The default behaviour of the default RC class is to use VBDC for any amount that exceeds the CRA allocation. If the NCC authorizes use of the volume-based capacity category, it will offer AVBDC and VBDC in combination, to an RCST or an RC. The combination of AVBDC and VBDC is seen as a single Capacity Category, denoted A/VBDC. CRA and FCA can be combined with any of the above. |
cb3dadab4f22493b63142dd0ebb828dd | 101 545-4 | 6.6 Examples of QoS configuration |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.