hash stringlengths 32 32 | doc_id stringlengths 7 13 | section stringlengths 3 121 | content stringlengths 0 2.2M |
|---|---|---|---|
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.3.2.1.2 Point to Point IP over Ethernet and over ATM | Point to Point IP over Ethernet (PPPoE) and Point to Point IP over ATM (PPPoA) are widely used over the terrestrial ADSL network. In that case the BSM acts as a true bridge for Ethernet and as a "wire" for ATM. All QoS management is provided above IP. The BSM connection is thus transparent to PPPoE/PPPoA and transports... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.3.2.2 Bandwidth on Demand (BoD) | There is a large body of publications on the topic of "Bandwidth on Demand" (BoD). BoD defines a bandwidth sharing mechanism that allocates bandwidth to requesting session based on a number of parameters including priorities and other QoS-related mechanisms. BSM with BoD got started as LAN interconnection via Satellite... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.3.2.3 Link Layer mechanisms | Since the radio spectrum and geostationary orbit slots for new satellites are limited resources it is important to investigate means to improve these resources in order for the BSM to get more bandwidth to allocate to services that provide QoS and hence added revenues. Link layer methods are almost universally used ind... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.3.2.3.1 Power control | Power control uses the variability in transmission conditions (weather in particular) to vary the transmission power of a terminal to maintain the link budget. Its principle is simple: use lower power in good weather, increase power when weather goes bad. Uplink power control is widely used and part of many terminal de... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.3.2.3.2 Use of FEC | In the BSM the channel error correction allows the upper layers to receive what is essentially error-free data. Because of FEC, on the link, the "detected" errors will either lead to correction and delivery or rejection and packet loss/retransmission. The second case results in added delay and potential loss hence a Qo... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.3.2.3.3 Variable information/symbol rate | Another way to improve the link budget is to reduce the rate transmitted by the ST. By doing that the link budget improves by a number of dBs. As was mentioned above, the service can continue either with degraded performance for all services (if the service agreements allow it) or at nominal performance for high QoS cl... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.3.2.3.4 Variable modulation | It is well known that the spectral efficiency of satellite networks can be increased through the use of higher level modulations such as 8-PSK and 16-QAM as compared to traditional QPSK modulation. The new high powered satellite transponders combined with concatenated error correction techniques allow the use of these ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.4 ATM | The Broadband Integrated Service Digital Network (B-ISDN) was developed to become an all purpose network technology to provide multimedia services. While it has failed as a desktop technology it is still widely deployed in the core networks. The Asynchronous Transfer Mode (ATM) was proposed as the B-ISDN transport. Whi... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.4.1 ATM QoS management | ATM provides four main classes of service, Continuous Bit Rate (CBR) or Deterministic Bit Rate (DBR), Variable Bit Rate (VBR), Available Bit Rate (ABR) and Unspecified Bit Rate (UBR). The first three have traffic descriptors that allow allocating resources to requesting calls. CBR traffic is characterized by a Peak Cel... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.4.2 ATM via satellite | For BSMs the most important ATM concepts are described in the User Network Interface (UNI) [4] that has influenced the design of the BSM SDUs and satellite payloads (ATM-like). In particular, the focus of satellite ATM was put on how to provide the ATM classes of service efficiently. In view of the fact that the BSM us... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5 Layer 3 IP QoS management | In the past decade IP networks have moved from providing a single best-effort service to multiple types of services with QoS, generally bounded packet delay and loss. Although the majority of the network traffic presently is still best-effort, emerging networked applications with different QoS constraints found in typi... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.1 Integrated services: the Intserv model | The Integrated Services (Intserv) model was developed to support real time applications and guaranteed resources over the IP protocol. Intserv is flow-based hence deals with a series of datagrams from a single source to a single destination that share a common set of quality of service parameter such as bandwidth, dela... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.1.1 Resource ReSerVation Protocol (RSVP) | RSVP- described in RFCs 2205 [38] and 2210 [39] is a signalling protocol designed to request a specific Quality of Service (QoS) from the network for a data stream. While not strictly at layer 3 (slightly above) it influences the work of routers and other network elements. An RSVP request is propagated node to node thr... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.1.1.1 RSVP QoS parameters | A RSVP reservation defined in RFC 2205 [38] request consists of two parts: a "flowspec" and a "filter spec". Together they form a "flow descriptor". The flowspec specifies the desired QoS. The filter spec, together with a session specification, defines the "flow" (in data packets) to receive the QoS defined by the flow... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.1.1.2 Support to multicast | RSVP already supports multicast but it has not been widely deployed in multicast networks. ETSI ETSI TR 102 157 V1.1.1 (2003-07) 36 |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.1.2 Session Initiation Protocol (SIP) | SIP is defined in RFC 3261 [63]. Originally defined for Voice over IP (as a replacement for the H.323 protocol) it is now extended to multimedia conferencing, messaging etc. SIP initiates multi-media sessions between two or more users and as such is a higher layer signalling protocol. SIP provides [67]: • registration;... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.1.3 Interaction with COPS | The Common Open Policy Service (COPS) is a policy based authorization and provisioning protocol [59]. The COPS protocol is used to transmit policy from a Policy Decision Point (PDP) to a Policy Enforcement Point (PEP). In particular COPS can implement simple client/server architectures for supporting policy control ove... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.1.4 Use of RSVP/Intserv over BSM | For RSVP, the BSM offers an advantage over terrestrial wired networks: because of the coverage of most BSM there will be a lesser number of routers on the reservation path of the BSM than on the equivalent terrestrial network. This helps alleviate some of the scalability issues of RSVP. Intserv (and RSVP) in the BSM wo... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.2 Differentiated services: the Diffserv model | Even with Intserv models, QoS traffic is continuing to stress existing networks, which in turn are unable to provide per- flow QoS because of scalability concerns. An answer to the above challenges is the Differentiated Service (DiffServ) (RFC 2475 [45]) framework, which delivers a coarse level of QoS in a per-node, pe... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.2.1 Codepoints | A DiffServ CodePoint (DSCP) (figure 14), identifies Per-Hop Behaviour (PHB) and is set in each packet header. The DSCP is carried in the DS-field, using six bits of the IP header (see RFC 2474 [44]). The forwarding behaviour related to each code point is defined "locally" within each operator domain. Hence it is the ro... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.2.2 Bandwidth Broker | In DiffServ, the Bandwidth Broker (BB) is an agent responsible for allocating preferred service to users as requested, and for configuring the network routers with the correct forwarding behaviour for the defined service. The idea of a BB was introduced as part of the Differentiated Services architecture to include mor... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.2.3 Support to multicast | In Diffserv the packet markings and their interpretation at the end receiver and in the intermediate queues are independent of the transmission method being used. The only issue would be for users of a multicast session on different domains that could be using different semantics for the DSCP hence where packets would ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.2.4 BSM support for Diffserv | The BSM architecture does not preclude the use of Diffserv at any point in the network. Actually, Diffserv is well suited for small inexpensive devices hence is not a large added cost features for small STs. DSCP can be mapped to internal BSM classes when needed. This is not an easy task as different DS domains may use... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.3 ITU IP transfer capabilities | As defined by the ITU, an IP transfer capability is a set of network capabilities provided by IP based networks to transfer IP packets from source to destination. This relates to the Intserv model in the sense that each IP transfer capability specifies a service model, a traffic descriptor, a traffic conformance (polic... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.4 Intserv and Diffserv co-existence in the BSM | Diffserv and RSVP models can co-exist on the same network, with Intserv deployed in the access network and Diffserv in the core network. The Diffserv/Intserv (and MPLS - see clause 7.5.5.1) combine the possibility for the hosts to request quantifiable resources along end-to-end data paths, provided by the IntServ and M... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.5 QoS routing | Quality of service routing has received considerable attention in the last few years because of the emergence of services requiring QoS and the general belief that the Internet has to offer QoS to become the all-purpose network of the future. Typically the routing decisions in the best effort Internet are at the packet... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.5.1 MPLS | The Internet does not have any intrinsic connection oriented services and usually packets are more of less routed individually. This creates issues for traffic managers who would like to balance loads in the network and, for QoS, as there is not strict guarantee that all packets in a stream will follow the same path. T... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.5.1.1 Routing principles | In a router, packets are sent from input port to output port based on a number of "routing" tables that contains the information mapping an input address to an output port following different rules, shortest path, lower cost etc. MPLS does not have this multiplicity: there is only one algorithm, label swapping. Operati... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.5.1.2 Label definitions | The MPLS header is composed of 4 bytes of which 20 bits are label information, 3 bits are for experimental use and can be used to identify VCs at the output of a labelled path. There is 1 bit to indicate bottom of stack and a byte of Time To Live (TTL) that is used to ensure a misrouted packet gets dropped after a numb... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.5.2 Q-OSPF | Q-OSPF extend the capabilities of OSPF V2 to provide the required level of support for QoS based routing while limiting those additions to the minimum necessary set [49]. All of the existing OSPF mechanisms, data structures, advertisements, and data formats remain in place. The extensions are signalled via the Options ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.5.3 BSM QoS routing requirements | Routing has been addressed in a specific TR 102 155 [13] and routing will not be discussed in details here. In the BSM where most routes are static, QoS routing becomes a queuing and traffic shaping issue. Even when dynamic routing is considered, the constraints for QoS routing over the BSM are not different than on te... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.6 Multicast QoS management | Multicast is the topic of a specific Technical Report, TR 102 156 [14]. In this clause the general aspects of Multicast QoS are highlighted. Multicast in itself is covered under the general RSVP but is rarely used. Diffserv markings are also in principle compatible with multicast if receivers have negotiated how to han... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.6.1 General model | QoS aspects that can be part of a multicast QoS management scheme include: - general reliable multicast as a special form of QoS class. Guaranteed delivery can be considered as specific service parameter supported by the BSM. In this case all receivers get the same "guarantee" and receivers that have not subscribed to ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.5.6.2 MPLS/RSVP approach | In addition recent work in the field of Multicast QoS has defined that combining MPLS (and RSVP-TE) and RSVP to Diffserv markings can lead to efficient deployment of QoS in satellite multicast. This is the solution adopted in the IP ConferEncing with Broadband multimedia over Geostationary Satellites (ICEBERGS) project... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.6 Interlayer QoS functionality | It has been recognized by a number of IETF groups that when IP packets are transported over any wireless sub-network, the access device can have much better information about the behaviour subnetwork than any of the source and destination hosts especially about outages and degradations. This is especially true for BSM ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 7.7 QoS requirements summary | Table 11 presents a summary of BSM services, a preliminary QoS class (to be refined in one proposed TS) and the compliance of a BSM as an appropriate means of delivering the service at the QoS level specified. Table 11: QoS-based services and BSM delivery Service/protocol QoS class QoS Function BSM delivery RTP/UDP (Vo... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8 BSM IP availability and performance | |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.1 Performance management | According to the ETR 309 [68], "Network Performance (NP) is the ability of a network or network portion to provide the functions related to communications between users; it contributes to service accessibility, service retainability and service integrity. Network performance parameter values are usually derived from Qu... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.1.1 Service Level Agreements (SLAs) | A network without predictable performance cannot be profitable. A major feature of the current Internet is the Service Level Agreement or SLA which is closely linked to Diffserv QoS management. It is a set of traffic management rules written as a contract between a subscriber and a service provider. It guarantees that ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.1.2 IP performance | As Internet traffic continues to grow exponentially, it has become essential for both the users and service providers to have a clear understanding on the performance of the network. The "Internet Performance Measurement and Analysis" (IPMA at the University of Michigan and Merit Networks) project studies the performan... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.1.3 BSM performance | The BSM performance management needs to put emphasis on manageable and measurable parameters hence a quantitative vs. qualitative approach and evaluate how and where the QoS parameters metric should be measured. In turn effective performance monitoring should help in establishing and enforcing policy management. In BSM... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2 Performance management functions | The major BSM performance management functions do not differ in essence from those of the terrestrial networks. They are usually at or above the IP layer. As BSM evolve, there will be more performance management in the ST, the satellite and the gateways. For example the application of Diffserv classes allow to improve ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.1 Admission Control | Admission control is for connection oriented traffic classes or traffic classes that have predictable behaviour. Connection Admission Control (CAC) or Call Admission Control is the set of actions taken by the network at a connection set up phase or during the connection re-negotiation phase in order to establish whethe... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.1.1 Admission Control parameters | Admission control parameters vary but will include the typical traffic descriptors (bandwidth, acceptable delay and loss, level of jitter tolerance). It may also include the source and destination of the admitted traffic, a duration for which the admission is valid, the location of the sender and receiver etc. All thes... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.1.2 Adaptive Bandwidth Control | In an aggregate traffic management and control framework, efficiently allocating bandwidth to provide quantitative QoS has been difficult due to unpredictable, unknown statistical characteristics of aggregate traffic. With inaccurate traffic information, static bandwidth allocation results in the network being underuti... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.1.3 Admitting a session in the BSM | For the BSM, admission control crosses the SAP: it has some SIAF and SDAF module. The SI part of the admission control module may also have to communicate with the NCC for global admission policies and the SD part with the satellite (if RSM) to ensure that there is no congestion onboard. Traditional admission control i... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.2 Flow control | Flow control is usually for non admitted classes, those classes that have non predictable behaviour, in essence connectionless best effort traffic classes. Hence it is designed for networks where individual connections do not reserve bandwidth and for the best-effort traffic. It usually uses a round-robin-like queue se... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.3 Traffic shaping | Before entering the network, traffic is queued. Traffic between queues is managed by queuing mechanisms (see next clause). Traffic Shaping imposes a limit to the admitted load in the network within a queue. Traffic Shaping in effect performs a local admission control function. Figure 17 shows how shaping polices traffi... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.3.1 Token Bucket/Leaky Bucket | Traffic shaping algorithms include the "Leaky Bucket" and the "Token Bucket" algorithms. In the leaking bucket algorithm the drip rate of the "bucket" defines the rate at which the queue can empty. In the token bucket it is the generation of tokens that limit the rate of transmission. A packet will leave the queue only... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.3.2 Policy control | Policy control is a function that determines whether a user has administrative permission to reserve network resources at a node. In the BSM it is an NCC function and relates to security and authentication. It will not be addressed here. ETSI ETSI TR 102 157 V1.1.1 (2003-07) 51 |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.3.3 Link between traffic shaping and BOD | In a BSM it is important that the traffic shaping parameters on the SI side of the ST be matched to the transfer capacity defined by the actual allocated bandwidth. Then the ST queue can empty at the right rate to fill all allocated slots and allow the congestion algorithms to be triggered in times where the allocated ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.4 Queuing and congestion avoidance | In a router or switch, shared resources need to be distributed amongst competing customers according to contention resolution policies. This is the role of the queuing policies. To a rough approximation queue management algorithms manage the length of packet queues by dropping packets when necessary or appropriate, whi... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.4.1 Explicit Congestion Notification (ECN) | Explicit Congestion Notification has heritage from the ATM Cell Loss Priority (CLP) bit. It was shown in clause 7.5.2.1 as two bits following the 6 bits of the DSCP. The use of the 4 ECN levels was defined in RFC 3168 [62]. Many protocols (including TCP) use packet drop as an indication of congestion. With the addition... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.4.2 Active Queue Management (AQM) | Queues are used to smooth spikes in incoming packet rates and to allow the router sufficient time for packet transmission. When the incoming packet rate is higher than the router's outgoing packet rate, the queue size will increase, eventually exceeding available buffer space. When the buffer is full, some packets will... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.4.2.1 Early Packet Discard (EPD) | Early packet discard comes from ATM and is implemented in the output ports of a router or switch. Early packet discard keeps track of the passage of frames (ATM cells) on selected flows or virtual circuits. If a new frame begins when the occupancy of the link buffer is above a threshold value, it discards the frame hen... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.4.2.2 RED and WRED | Random Early Discard (RED) is a buffer management scheme designed to prevent tail drop caused by large traffic bursts. RED will randomly selects and drop packets in a queue. This prevents tail drop that can lead to traffic oscillation in the network. RED however has to be used on aggregated flows [19]. Weighted RED (WR... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.4.3 BSM impacts | Congestion control can and will be applied at every queue in the BSM (SI, SAP, OBP, gateway). If congestion control is implemented at the ingress/egress to the BSM (ST or gateway) standardized functions can be used. In an OBP simple methods are to be used. The simplest method will be queue drop, which means that all pa... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.5 Scheduling | Queuing and scheduling mechanisms allow controlling congestion by determining the order in which packets are sent out from a queue to an interface based on priorities assigned to those queues and packets. Scheduling policies are defined by QoS requirements amongst each customer flow and determine the scheduling of each... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.5.1 Round Robin and weighted round Robin | Round Robin is the simplest scheduling algorithm. Resources are allocated to each requesting queue, in turn. After resources are depleted, the next round of allocation starts where the previous one stopped, thus ensuring that all queues will eventually get service. Weighted round Robin is a variant where some queues wi... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.5.2 Weighted fair queuing | The principle between weighed fair queuing is simple. Resources are allocated between requesting queues by first allowing interactive applications at the head of the queue to reduce delay and then allocating the remaining resources to the other queues. |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.2.5.3 BSM scheduling | Scheduling in the BSM is intrinsically linked to the way the system manages bandwidth and other scare resources. At the macro level at the NCC, each ST can be considered a "requesting queue" and transmission opportunities (and bandwidth allocation) will be assigned by a well known algorithm based on the type of ST, whi... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3 TCP performance | The Transport Control Protocol (TCP) is a reliable (acknowledged) protocol widely used over the Internet and its use over different networks including satellites is well known. Because its operation has a great impact on the service that relies on it, TCP operations are described in this clause. While a number of varia... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.1 TCP congestion control | TCP employs four congestion control mechanisms slow start, congestion avoidance, fast retransmit and fast recovery. These algorithms are used to adjust the amount of unacknowledged data that can be injected into the network and to retransmit segments dropped by the network. They are closely linked to the window based f... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.1.1 Slow start and congestion avoidance | TCP is based on the principle that a sender ignores the status of the network. In order to avoid transmitting too much traffic and create un-necessary congestion, TCP uses the "slow start" algorithm. In slow start the cwnd is set to 1 segment (in essence 1 packet) and ssthresh is set to the receiver's advertised window... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.1.2 Fast retransmit and fast recovery | TCP ACKs always acknowledge the highest in-order packet that has arrived. So an ACK for packet N also ACKs all segments sent before N. Also if a segment is received out-of-order the ACK will be for the highest in-order segment not the segment was received. The "fast retransmit" algorithm uses duplicate ACKs to detect l... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.1.3 Selective acknowledgements | Selective Acknowledgements addresses the transmission limitations of TCP. A Selective ACKnowledgment (SACK) mechanism was defined in RFC 2018 [37]. With TCP SACK, the data receiver can inform the sender about all the segments that have arrived successfully, allowing the sender to retransmit only the segments that have ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.1.4 Window scaling | In large delay × bandwidth product networks, the initial fixed windows of early versions of TCP, were preventing the channel to fill to capacity as: TCPwindow = delay × bandwidth RFC 1323 proposes to use window scaling to ensure that the full available bandwidth is used. This is now a feature of all TCP implementations... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.2 TCP operations over BSM | TCP was on of the first Internet protocols that was analysed for transmission over satellites. TCP usage over satellite is standardized in RFC 2488 [46]. The next clauses describe the impacts of satellite transmission over satellite. |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.2.1 Consequences of noise | If TCP ignores why a packet was dropped: congestion or corruption? TCP must assume the drop was due to network congestion; this is by design and to avoid congestion collapse upstream in the network. TCP's default mechanism to detect dropped segments is a timeout (Retransmission Time Out or RTO). If the sender does not ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.2.2 Consequences of delay | In addition to the widow issue highlighted above, large delays have also other impacts on the behaviour of TCP. The slow start and congestion control algorithms can force poor utilization of the available channel bandwidth when using long-delay networks. For example, transmission begins with the transmission of one seg... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.2.3 Recommended mechanisms | Table 13 from RFC 2488 [46] presents the recommendations for using TCP over satellite. Table 13: Mechanisms for TCP over satellite defined in RFC 2488 [46] Mechanisms Location Use Slow Start Sender Required by the protocol Congestion Avoidance Sender Required by the protocol Fast Retransmit Sender Recommended Fast Reco... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.3 Performance Enhancing Proxies | Performance Enhancing Proxies (PEPs) are defined in RFC 3135 [60] and are used to improve the performance of TCP (TCP-PEPs) or other protocols over network paths where typical performance suffers due to characteristics of a link on the path. Different types of PEPs are available depending where they operate in the netw... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.3.1 Split connections and TCP spoofing | The term TCP spoofing is sometimes used synonymously for TCP PEP. The term TCP spoofing more accurately describes the characteristic of intercepting a TCP connection in the at the entry point to the link under consideration and terminating the connection as if this point was the intended destination. This creates in ef... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.3.2 TCP bandwidth snooping | Bandwidth snooping is another transport level performance enhancing proxy. It is a valid mechanism on all networks that use one form or another of bandwidth on demand. BoD requires time between assignments and in reality limits cwnd to the current allocated bandwidth even if the queue is still growing and more packets ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.3.3 TCP Friendly Rate Control (TFRC) | The TCP Friendly Rate Control protocol, defined in RFC 3448 [65] was designed for streaming and audio streams competing with TCP traffic. TFRC is a congestion control mechanism that could be used with RTP. It is one example of a transport level PEP that is not TCP. TFRC allows those flows that compete with TCP to be so... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.3.3.4 Caching | Caching is an application level PEP. It defines the storage of information that is updated infrequently close to the point of usage of that information. Hierarchical organization of caches is commonly used in the World Wide Web for anything from catalogues and company information. Web caching has been recognized as an ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4 Standardized performance metrics | In order to set the performance monitoring specifications performance metrics need to be defined and assigned some agreed on value either based on the actual service requirements or IP specific parameters (based on aggregated service requirements). |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.1 Service performance requirements | The ITU has defined and standardized service requirements which are the basis for all network performance [25]. They are presented in table 14 for audio and video transmission and table 15 for data applications. As can be seen from these tables the BSM performance were not taken into account. While it is true that low ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.2 IP performance metrics | This clause briefly describes the basic performance measurements for Internet traffic, hence for aggregated services. These metrics were defined in the RFC 2330 [42], RFC 2678 [50] and ITU-T Recommendations Y.1540 [26] and Y.1541 [27] and are used to manage traffic in IP networks. They can be summarized in two categori... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.2.1 Link metrics | Link metrics relate to end-to-end and system level capabilities. They include (but are not limited to): • connectivity/reachability: connectivity verifies if there a connection between 2 hosts; the polling all devices in less than 60 sec (for 10 000 devices) is a design criterion used by NANOG; • bulk transport capacit... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.2.2 Per packet and per flow measures | Table 16 lists performance objectives for IP QoS classes based on ITU-T established Diffserv classes presented in table 4. Table 16: QoS class definitions and network performance objectives from ITU-T Recommendation Y.1231 [28] QoS Classes Network Performance Parameter Nature of Network Performance Objective Class 0 Cl... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3 BSM performance metrics | The BSM uses the same metrics as were defined in the previous clause but will apply specific BSM methods to their measure and monitoring. This clause defines the BSM metrics and how monitoring is performed. Annex D lists the IP performance metrics and their characteristics. ETSI ETSI TR 102 157 V1.1.1 (2003-07) 60 |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.1 Link metrics | For the BSM the monitored link level metrics include connectivity, availability, loss and throughput. |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.1.1 Connectivity | The loss of connectivity in the BSM can mean three things: the link failed (or is in fading), the hardware failed or the software failed. In any case for a BSM fault detection has to do with the Satellite Control stations for the space segment and local connectivity managers for the ST. The former has a heritage of ove... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.1.2 Availability | According to the ETR 309 [68], "Availability performance is the ability of an item to be in a state to perform a required function at a given instant of time or at any instant of time within a given time interval, assuming that the external resources, if required, are provided". Alternatively, availability can be loose... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.1.3 Throughput | Throughput defines the effective data transfer rate in bits per second for a particular service user as measured at the egress point of the network. Throughput is influenced by bandwidth sharing policies and overhead due to signalling and added overhead. BSM throughput is usually measured as the number of cells/packets... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.1.4 Goodput | Goodput defines the effective data rate delivered to the application. Goodput is influenced also by overhead and signalling but especially by reliable protocols that rely on acknowledgement. It can happen that while throughput is high goodput could be very low. |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.2 Packet/flow metrics | |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.2.1 Delay | Delay is measured as the time in transit of a packet between the ingress of the network to the egress of the network. For the BSM it could be measured at the SI or the SI/SAP interface. |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.2.2 Delay variation (1 point and 2 points) | Delay variation (jitter) is a measure the differences in delay between the arrival of successive packet from the same flow at the egress of the network (1 point delay variation) or via a transit point (2 point delay variation). It can be controlled for those services that are sensitive to it by use of buffering and oth... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.2.3 IP Loss Ratio | As for availability loss in the BSM can have multiple sources: IPLRBSM = IPLRlink + IPLRshaping + IPLRcongestion, where: - Losslink is the number of packets lost due to channel noise and residual errors; - Lossqueuing is the number packets lost due to traffic shaping; - Losscongestion is the number packets lost due to ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.3.2.4 IP Error Ratio | IPER will be caused mainly by residual errors. Transmission errors will normally be corrected at the link layer (see clause 7.3.2.3.2). Some errors will be detected the packet's CRC and any residual transmission error will normally result in the packet being discarded hence IP packet loss. ETSI ETSI TR 102 157 V1.1.1 (... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.4 End-to-end QoS budgets | When appropriate, end-to-end QoS measures will result in an end-to-end quality budget. To achieve the required end-to-end QoS the quality budgets must be allocated between source and destination. For the BSM it should include contributions from the SIAF but could also be limited to ingress SAP to egress SAP depending o... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.5 Monitoring methods | These define how measurements are obtained using standard methods. |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.5.1 Polling | Polling means namely "ask" if a device is on. Traditional polling will use SNMP "Get" functions or ping functions to find out if a router is on. The traceroute function can also be used to see where a packet transited on the source to destination path. For a BSM, a number of devices could answer to a polling message de... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.5.2 Probing | Probing sends a well-known stream from one host to another and then uses it to measure a number of performance parameters. This could be a powerful tool to ensure that the BSM is indeed fully operational as an IP subnetwork. |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.4.5.3 Traffic sampling | Traffic sampling, which can be done inside a router or a switch, means taking copies of some packets and verify source and destination, TTL etc. While it is not an essential part of performance monitoring it could be important in some security features and to establish statistics about flows that in turn could be used ... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.5 Quality rating | The TIPHON project has defined overall transmission quality Rating (R) as the full acoustic-to-acoustic (mouth to ear) quality, experienced by an average user, for a typical situation using a "standard" telephony handset. While this is only applicable to the voice services the idea of a "quality rating" for BSM service... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.6 Network management | For a BSM the network management can be as complex as in any other broadband network. There is heritage in the FSAN approaches as well as in the current DSL and cable modem deployment. For the BSM itself the management will be located in the NCC but local functions related to usage accounting and billing could be locat... |
ec1da71088ed2b609dd1a945b52a5786 | 102 157 | 8.6.1 SLA negotiations | The SLA negotiation will set the contract for the availability, latency and packet delivery of the offered service. There is one SLA per type of service (Internet, hosting, enterprise connectivity, etc.). |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.