hash stringlengths 32 32 | doc_id stringlengths 7 13 | section stringlengths 3 121 | content stringlengths 0 2.2M |
|---|---|---|---|
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.2.6 Signal Phase and Timing Message (SPATEM) | SPATEMs are semi-static messages which means that their size can slightly change. The timing of the SPATEM depends on the changes in the traffic-light behaviour. Today's dynamically SPATEM assigning systems could update the sequence about once every 0,1 second. So, updates of 10 Hz are not an exception. The packages however are not large and both size as update rates although dynamic are still quite predictable. With regards to channel use, only Roadside Units (RSUs) will transmit SPATEMs and therefore it can be expected that in a range of 200 m to 400 m only a very few RSU will transmit. After analysing the SPATEM, it can be recognized that the SPATEM can use the same parameter structure as the CAM and only will have different values. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.2.7 MAP Message (MAPEM) | The MAPEM messages are static messages which provide an overview over the road topology with all lane descriptions and stop lines etc. The size always stays the same and it is disseminated with a fixed lower frequency such as 1 Hz to 2 Hz by RSUs. Like for SPATEM, the same parameter structure as for CAM plus additional parameters taking their Geobroadcast nature into account. can be used. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.2.8 In Vehicle Information Message (IVIM) | The IVIM is a message which is intended to represent for instance a sign. A sign is expected to be static, like a speed limit sign. However, road operators do change the prelimits depending on the situation on the road. Signs can therefore be static or semi-static (changing ones in a while, in intervals of minimal 30 seconds). IVIMs are awareness messages which can only be disseminated by authorities which could be RSUs but also special vehicles. Like SPATEM and DENM they are generated by authority controlled ITS-Ss. The number of ITS-Ss within a certain vicinity will be limited. It can be assumed not to be more then 4. The impact on the channel use can be assumed minor. The size is known and static while the rate may be of a few Hz and it can slightly change depending on the road it is active. The dissemination is timely predictable to allow a resource management functionality to fulfil its task. The same parameter structure as for CAM plus additional parameters taking their Geobroadcast nature into account. can be used. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.2.9 Service Announcement Message (SAM) | At present from a message dissemination perspective, the SAM can be seen similar to the IVIM. The SAM is not yet considered to be used for the announcement of dynamic safety use cases, which could result in some additional dissemination requirements (at present this is not foreseen). SAM dissemination can be predicted sufficiently to allow resource management to fulfil its task. The same parameter structure as for CAM plus additional parameters taking their Geobroadcast nature into account can be used. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 46 |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.2.10 Vulnerable road user Awareness Message (VAM) | The VAM is an awareness message similar as the CAM its size is smaller and rate more predictable and lower. Considering road scenarios, the number of present VRUs in an area could however be much more then Vehicles in the same area. The dynamic behaviour of VRUs is much slower than that of Vehicles and therefore the channel use behaviour is likely to be more predictable for the RM than that of the dissemination of CAMs. The same parameter structure as for CAM plus additional parameters taking their Geobroadcast nature into account. can be used. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.3 Types of message services from RM perspective | The envisioned RM solution will provide indications to the message services so that they can dynamically adapt the messages they generate in real time. To this aim, the message services could reduce the number of messages and/or their size to follow the indications provided by the RM, while at the same time send additional messages to be offloaded to alternative channels. In this context, the following types of message services are identified: • Size adaptation message services that adapt the message size (dynamic inclusion of objects in CPMs or vehicles in MIM, or optional elements in any other Message service) but keep the message interval fixed. • Interval adaptation message services that adapt the interval of the message generation instead of adapting their size. • Size and interval adaptation message services that have the flexibility and intelligence to adapt both the message interval and size following the indications of RM. According to how the message services generate their messages, the following types are also envisioned: • Predefined rules message services that have their own message generation rules that trigger the generation of new messages. When enough resources are available, they generate the necessary messages following these rules. If the available resources are higher than the resources needed by these message services, they simply follow they predefined message generation rules. Examples are the CA and DEN services (see Figure 12a). • Adaptive rules message services that adapt the messages they generate to the resources available. The more resources available the more messages they can generate, up to a certain limit that can be high. One example is the CP service, that could adapt the number of perceived objects and regions following the instructions of the RM (see Figure 12b). (a) CAM (b) CAM and CPM Figure 12: Illustration of resources consumed by CAM and CPM ETSI ETSI TR 104 073 V2.2.1 (2026-02) 47 |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.4 Heterogeneous resource requirements | |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.4.1 Same services but different resource needs | One important aspect to consider in the design of the RM is the fact that two (or more) ITS-S could be running exactly the same message services, but the radio resources each one consumes can differ widely. Some examples are described below: • Cooperative Perception Service (CPS). An ITS-S on a vehicle typically detects less objects than an ITS-S on an RSU, so that the amount of information they need to generate is different. Also, two nearby vehicles can also detect a very different number of objects based on their specific location and sensors. The quality of the sensors is also an important factor, since the quality of the detections has an impact on the amount of information that needs to be transmitted. • Automated Vehicle Marshalling (AVM). Vehicles participating in an AVM system transmit Marshalling Vehicle Messages (MVMs). The infrastructure controlling these vehicles generates Marshalling Infrastructure Messages (MIMs) that are larger in size and have to be more frequently transmitted (see for more information the publication "Automated Vehicle Marshalling" [i.29]). Even though both participate in AVM, their resource needs are intrinsically asymmetric. • Cooperative Awareness (CA). Two implementations may have different resource requirements depending on the optional elements implemented, and the specific driving conditions. As an example, a vehicle could be in a traffic jam in one driving direction and a nearby vehicle could experience free flow conditions in the other direction. The CA service of the vehicle stopped would generate CAMs at 1 Hz, while the other could require the transmission of CAMs at 10 Hz depending mainly on its speed. The design of the RM should be able to handle this heterogeneity of resource needs to optimize the bandwidth efficiency and system performance. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.4.2 Different stations with different needs | The resources needed by different ITS-S also depend on the number of Message services they implement or their Release, since they are expected to implement different message services in each release. Some examples are shown below: • A Release 1 ITS-S generates essentially CAMs and only occasional DENMs. • A Release 2 ITS-S also implements collective perception, so it generates CPMs in addition to CAMs and sporadic DENMs. • A Release 3 ITS-S could also implement manoeuvre coordination, generating extended Manoeuvre Coordination Messages (MCMs) in addition to CPMs, CAMs and DENMs. On any given road different ITS-S implementing different Releases can be intermixed, each with its own resource needs-ranging from a lightweight CAM-only sender to a fully-featured node that generates CAM, CPM, MCM and DENM concurrently. The design of the RM should ideally take into account these aspects. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.5 Indications to message services | Message services will dynamically adapt the messages they generate (size and/or interval) following the indications of RM. Different options are possible with different levels of abstractions for the RM to inform the ITS-S about the messages they can generate: • Bits/s. The RM could limit the amount of bits/s that each ITS-S can generate in the default channel (or alternative channels). With this approach the message services could adapt e.g. the message interval for a given size. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 48 • Bits. The RM could assign a certain number of bits to the message services to indicate them that they are allowed to generate one or more messages so that the total amount of bits does not exceed the assigned one, irrespective of the time. In the next update, the RM could take into account if the message services consumed all the bits assigned or not, and re-assign them accordingly. This approach is similar to the previous one but avoids the complexity of time management in the message services, since they only have to check the number of assigned bits left to generate new messages. • Ton/Toff. This approach was used in Release 1 as detailed in clause 5.2. Radio access technologies like LTE-V2X and 5G NR-V2X would not support this approach because they have a fixed Ton, except in those cases where one facility-layer message has to be segmented in multiple packets at the access layer. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 6.6 Conclusions | At present, there are two types of Message Services (MSs) identified. The first type corresponds to those that directly use the sensors information to generate and disseminate messages, such as CAS, and therefore should autonomously register to an RM service when present. The second type corresponds to those where the generation and dissemination are triggered by applications. In this case, in principles, the applications should have knowledge about the dissemination needs; however, present known applications do not have direct control on the disseminated messages but only provide basic data element requirements and trigger or cancel commands to an MS, which implies that it is not needed that these applications register to the RM. Therefore, at present, both types can behave with the MS communicating with the RM. In case RM functionality is present in an ITS-S, MSs should register to the RM service and inform the RM about changes in their dissemination needs. In return, MSs should receive back from the RM the limitations at any time that the limitations change. This information exchange is realized through a control interface between the RM and all active MSs. In case RM functionality is present in an ITS-S, an MS provides initiated disseminations to the RM, allowing the RM to control the real disseminations. The RM could in case of issues inform the MS about misbehaviour or unknown changed restrictions. With regards to the dissemination requirements of MSs, an MS should at least provide these requirements through the control interface, but it may also be relevant to allow this information being part of the FL payload in the data path. MSs should provide at least the following minimum set of parameters: • Principle dissemination path requirements (ALIgroup or ALIgroups + possible technology specific) parameters. • Requested bandwidth with related expected time period. • Expected packet rate in the same time period. MSs could provide additional parameters such as: • Secondary (alternative) dissemination path requirements (ALIgroup or ALIgroups + possible technology specific) parameters. • Requested bandwidth with related expected time period. • Expected packet rate in the same time period. The RM should communicate with the message services using appropriate metrics. The main options considered to define the requested bandwidth are in terms of bits per second (bits/s) or in bits per update period. This latter approach offers an advantage by simplifying time management for the message services, as they only need to check the remaining allocated bits before generating a new message, ensuring that the dynamic adaptation of size and/or interval remains within the assigned resource limits. The design of the RM solution should recognize that message services have the capability to dynamically adapt their generated messages by modifying either the message size, the generation interval, or a combination of both, according to RM indications. The solution designed will have to provide sufficient flexibility to the message services to adapt the messages they generate to their needs. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 49 Another important aspect is the intrinsic heterogeneity in the resource requirements of ITS stations. On any given road, stations with very different capabilities can coexist, ranging from older ones (Release 1) that primarily generate CAMs and sporadic DENMs, to newer ones that also include CPMs and MCMs. This mix of lightweight and fully-featured nodes increases the complexity of the RM operation and underscores the need for the RM to intelligently manage different demands. In addition, even when running the exact same service, such as Cooperative Perception (CP) or Cooperative Awareness (CA), different stations may consume different resources. Factors such as specific location, driving conditions (e.g. congestion vs. free flow), sensor quality, or the station's role (e.g. in an Automated Vehicle Marshalling or AVM system) lead to substantial asymmetry in the quantity and frequency of data that needs to be transmitted. All these aspects will need to be taken into account in the design of the RM solution. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7 Resource management concepts | |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.1 Introduction | Resource management plays a pivotal role in ensuring efficient operation and Quality of Service (QoS) in communication systems. The FL is particularly well-suited for this task, as it can integrate the capabilities of lower layers with the dynamic functional requirements of message services. As highlighted in the Multi-Channel Operation (MCO) concept study in ETSI TR 103 439 [i.1], ITS-S-MS at this layer are unaware of the communication needs of other ITS-S-MS. This necessitates a robust Resource Management (RM) functionality to ensure a consistent and efficient use of available channels. The RM ITS-S functionality should be seen as an improvement of the original MCO functionalities as specified by the set of MCO specifications. The RM operates by harmonizing dataflows between message handling services to realize improved trustworthiness and QoS while making an efficient use of the channels. It leverages the capabilities of lower layers while supporting technology agnostic operation of message-disseminating message services. This layered approach ensures that resource allocation and communication management remain dynamic, adaptive, and aligned with real-time network conditions. The following clauses outline key concepts and mechanisms for resource management, drawing analogies to wired network practices while addressing the unique requirements of vehicular networks. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.2 Analogy with wired networks | Resource management in vehicular networks shares conceptual similarities with wired networks, particularly in the context of resource reservation and QoS. In wired networks, clients request resources for specific flows, defining the flow parameters and QoS requirements. The network evaluates these requests and accepts them only if all nodes along the path can support the resource demands. Nodes include edge and inner nodes ensuring seamless communication. Figure 13 illustrates resource reservation for QoS flows in wired networks. Figure 13: Resource Reservation in Wired Networks In vehicular networks, this analogy holds with certain adjustments. Here, applications or message services act as clients, requesting resources for message dissemination based on their QoS needs. The RM evaluates these requests and grants or denies them based on channel availability and service priority. Unlike wired networks, vehicular networks lack intermediate nodes for resource validation. Instead, the system uses channel load measurements for resource management. This model supports dynamic and adaptive resource allocation tailored to the real-time demands of vehicular communication systems. Figure 14 shows how the concept of resource reservation can be adapted to vehicular networks. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 50 Figure 14: Resource Reservation Analogy in Vehicular Networks The trustworthiness and QoS of the vehicular network are not guaranteed by RM. RM is only able to manage and supervise the dissemination of the messages. In wired vehicular networks the trustworthiness and QoS are fixed by the system design, which is a closed box approach in which it is known what network can be expected at least in terms of congestion. In the ITS, this should also be managed. This management is handled by an agreement between the stakeholders on what use cases and message services are allowed to exchange their message in the specific available radio channel or channels. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.3 Resource management architecture and its mechanisms | |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.3.1 Overview | This clause describes mechanisms for resource management that can be part of the RM functionality, in particular mechanisms that can be used in Bandwidth Management Component (BMC) and Message Handling Component (MHC), as defined in the MCO architecture. For the RM, the BMC can include functionalities such as admission control and bandwidth management. These functionalities require the BMC to collect application requirements, monitor channel conditions, and configure radio interfaces: • The admission control functionality is critical in regulating resource usage when a message service is activated. It evaluates resource availability and prevents the activation of message services that exceed current bandwidth or violate predefined priorities or regulatory constraints. See clause 7.3.2 for details on different admission control techniques. • The Bandwidth Management functionality ensures the adaptive allocation of radio resources to meet varying traffic demands. This mechanism is applied exclusively to message services that have been admitted through admission control, ensuring that only authorized message services utilize the available bandwidth. Additionally, the priorities of the messages are a key factor in bandwidth management. Higher-priority messages, such as safety-critical notifications (e.g. DENMs), are allocated bandwidth preferentially to ensure timely delivery. By dynamically adjusting resource allocation in real-time based on priorities and current conditions, this approach optimizes performance while adhering to predefined admission policies. This approach adjusts bandwidth allocation in real-time based on service requirements and current conditions. See clause 7.3.2 for examples of bandwidth management techniques. The MHC can implement traffic shaping and traffic policing policies to manage the transmission of messages effectively, ensuring compliance with the configuration limits set by the BMC: • The traffic shaping policies can be used to smooth and regulate the traffic generated by each message service, buffering or delaying excess traffic to ensure long-term compliance with traffic limits. The MHC can be responsible for performing this task to ensure that each message service adheres to its allocated bandwidth and complies with the resource management policies defined by the BMC. For common algorithms see clause 7.3.4. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 51 • Traffic policing ensures strict traffic limits by marking or discarding excess traffic without buffering. This enforcement guarantees that the aggregation of all the messages generated by the message services and sent down to the lower layers operate within their permitted limits, maintaining overall network stability and fairness. See for more details clause 7.3.5. These entities work collaboratively to optimize resource allocation and maintain QoS in vehicular communication systems. Figure 15 illustrates how these mechanisms could be integrated into the BMC and MHC of RM. Figure 15: Architecture and mechanisms for Resource Management (RM) at the facilities layer The RM can be implemented in a synchronous or in an asynchronous way. Synchronously by synchronizing to a system heartbeat which could be the maximum CAM repetition rate (or twice) of any other relevant heartbeat. It can also be realized asynchronously but just waiting for changes in the MSs dissemination requirements, the available radio resources or issues detected in the message dissemination flow (see clause 7.4 for details). Overall, it is expected that an asynchronous implementation allows a more efficient implementation however this is not proven. In the following clauses these aspects are detailed and highlighted. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.3.2 Admission control techniques | Peak Resource-Based Admission Control: Limits resource allocation based on the peak resource demands of a message service. It avoids allocating resources that might exceed network capacity during peak usage periods. A practical example is restricting the activation of multiple message service with a high peak message generation rate during a traffic jam. Average Resources Admission Control: Ensures that the cumulative resource usage remains within bounds by summing up the average resource demands of all active message services. This typically involves adding average resource demands, rather than peak values that could be rarely produced. For instance, this method might deactivate a message service when the sum of average resources is higher than the available resources. Equivalent Bandwidth Admission Control: Allocates resources based on the equivalent bandwidth required for a message service, considering both average and peak demands. The equivalent bandwidth is calculated by combining the statistical distribution of traffic loads with the desired QoS parameters, such as delay, jitter, and packet loss rate. For example, if a message service exhibits bursty traffic patterns, the equivalent bandwidth will factor in both the average rate and a margin to accommodate bursts, ensuring reliable operation. This approach is well-suited for message services where peak demands need to be balanced against average usage to optimize resource allocation. This approach provides a more realistic estimation of resource requirements compared to simple peak or average calculations. Statistical Admission Control: Utilizes probabilistic models to estimate resource requirements and allocate them based on expected traffic patterns. Unlike equivalent bandwidth control, statistical admission control focuses on the likelihood of multiple message services requiring peak resources simultaneously. This approach uses historical data and probability distributions to anticipate resource needs, ensuring that resources are not over-allocated based on rare peak scenarios. In practice, it might prioritize safety-critical messages during periods of high traffic density, while reserving capacity for unexpected spikes in demand. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 52 Policy-Based Admission Control: Implements resource constraints based on pre-defined policies, such as user priority levels or regulatory requirements, ensuring that critical message services receive necessary resources. This method can be combined with other admission control mechanisms. For instance, policy-based rules can complement Statistical Admission Control by defining thresholds for resource allocation under specific conditions, such as prioritizing certain messages during congestion. Similarly, it can enhance Equivalent Bandwidth Admission Control by incorporating policies that adjust the equivalent bandwidth calculations based on application-specific or regulatory priorities, ensuring more granular and adaptive resource allocation. For example, a policy might specify that safety-critical messages always have precedence over awareness messages, regardless of statistical or bandwidth estimates. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.3.3 Dynamic bandwidth management techniques | Proportional Fairness: Balances resource distribution equitably among applications and message services by ensuring that each of them receives a fair share relative to its needs, promoting overall network efficiency. Message priorities play a critical role in this mechanism. Messages are grouped by their priority levels, and resources are allocated proportionally within each group. For multiple priority levels, higher-priority groups are allocated resources first, and any remaining bandwidth is distributed among lower-priority groups using proportional fairness. For instance, DENMs (high-priority messages) are served first to ensure safety-critical operations, while CAMs and CPMs share the remaining bandwidth in proportion to their demands (assuming that they have the same priority). Max-Min Fairness: Prioritizes applications and message services with the least resources by maximizing their resource allocation without significantly impacting others, ensuring minimum fairness for all users. Message priorities are also important in this context. High-priority messages are allocated resources first, ensuring their timely delivery. Once higher-priority demands are satisfied, the remaining resources are distributed among lower-priority messages in a way that maximizes the minimum allocation, ensuring no message service is entirely starved of resources. Message Generation Scheduling: This advanced technique involves the BME directly controlling the scheduling of message generation by message services. The BME considers the available bandwidth, current channel conditions, and message priorities to decide when each message service can generate a new message. For instance, higher-priority messages might be scheduled for immediate generation, while lower-priority messages are deferred to avoid congestion. This approach ensures an optimal balance between resource utilization and the timely delivery of high-priority messages. Machine Learning-Based Management: Applies predictive analytics to forecast traffic patterns and optimize bandwidth allocation proactively. Machine learning algorithms can also account for message priorities by learning from historical traffic patterns and adapting allocation strategies. For example, the system might predict an increase in high-priority messages and pre-emptively allocate bandwidth to accommodate this demand, ensuring timely message delivery across all priority levels. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.3.4 Traffic shaping policies | Token Bucket Algorithm: Controls the flow of traffic by allowing bursts within a limit, regulated by token generation rates. The algorithm internally uses tokens, which are generated at a constant rate, to authorize the sending of packets. Each packet consumes a token, and traffic exceeding the token rate is delayed until tokens are available again. This mechanism ensures compliance with average traffic rates while permitting flexibility. For instance, it can handle sudden bursts of messages without disrupting other message services. Leaky Bucket Algorithm: Smooths traffic by enforcing a constant output rate, discarding excess data beyond the bucket's capacity. The algorithm uses a fixed-size bucket where packets are added at any rate but are released at a steady, predetermined rate. If the bucket overflows, excess packets are dropped. This ensures a steady flow of traffic and can be used to regulate bursts of messages to avoid network saturation. Dual Token Bucket Algorithm: Combines two token buckets to manage multiple traffic priorities. The primary bucket regulates high-priority traffic, while the secondary bucket manages lower-priority traffic. High-priority traffic consumes tokens from the primary bucket, ensuring prompt transmission, while lower-priority traffic waits until both buckets have sufficient tokens. Virtual Scheduling Algorithm: Simulates scheduling in a virtual timeline, ensuring fair distribution of resources by assigning each packet a virtual departure time. Packets are sent in the order of their virtual departure times, maintaining compliance with predefined traffic limits. This could prioritize transmission of event-driven messages over general status updates, ensuring timely delivery of essential information. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 53 Generic Cell Rate Algorithm (GCRA): Monitors and regulates traffic based on cell arrival times to ensure adherence to the specified rate and burst tolerance. GCRA uses a virtual scheduling mechanism to check if incoming packets comply with the configured traffic profile. Non-compliant packets are either dropped or marked. Sliding Window Algorithm: Controls the rate of data transmission by monitoring traffic over a moving time window. It calculates the volume of data sent during the window and ensures it does not exceed a predefined threshold. If traffic exceeds the limit, packets are delayed or dropped. This technique can be employed to manage the steady flow of messages from vehicles during a convoy, maintaining a balance between throughput and compliance with traffic limits. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.3.5 Traffic policing policies | Unlike traffic shaping, which smooths traffic over time by buffering excess data, traffic policing focuses on immediate compliance by dropping or marking traffic that exceeds the defined limits. This ensures that the system adheres strictly to resource policies without introducing delays caused by buffering. Traffic policing mechanisms often include techniques similar to those used in traffic shaping, such as the Token Bucket and Leaky Bucket algorithms. However, in traffic policing, these mechanisms enforce strict limits by discarding non-compliant packets instead of buffering them. Figure 16 illustrates the differences between traffic policing and shaping when they are applied to the same message flow. In the figure, the vertical axes represent the message rate, and the horizontal axes represent the time. Figure 16: Effect of traffic Policing and Shaping on a message flow |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.4 Resource Management operation | The operation of RM is closely linked to the behaviour of lower layers, since it relies on channel load measurements to assess dissemination possibilities. As these measurements take time and are updated frequently to ensure stable resource management, the timing of RM should be aligned with the optimal channel load measurement period. There should be a direct relation between the channel load measurement period and the RM period. For instance, a new RM cycle could start every n channel load measurement periods (with n ∈ {1,2,3,…}). Additionally, filtering mechanisms might be required to smooth fluctuations in channel load measurements, particularly if shorter measurement periods are used. Finally, the potential synchronization of channel load measurements across ITS-Ss should also be evaluated. RM can also be triggered by message services or applications. In this case, dissemination requirements are still provided by the message services, but the triggering is message service-driven rather than bound to a fixed schedule. This enables message service-triggered RM interactions, for instance when a safety-critical sequence of messages requests immediate transmission. Such an approach reduces the latency for specific messages, though at the cost of increased complexity. It also implies that message services do not need to operate synchronously; instead, they may function independently according to their own pace. Since each application or message service may have its own update rate, defined either statically or dynamically, it is difficult to prescribe a single RM operation model. Instead, RM should be flexible enough to adapt to the timing needs of the message services it supports. Two main approaches can be envisaged. First, a periodic approach where RM operates according to channel load measurement cycles and checks for updated message service requirements during those cycles (Figure 17). Second, an interrupt-based approach where RM is triggered whenever an message service registers or updates its requirements, making decisions based on the most recent channel load values (preferred in many cases, though ultimately an implementation choice). A hybrid solution combining both approaches may also be envisaged (Figure 18). ETSI ETSI TR 104 073 V2.2.1 (2026-02) 54 Figure 17: RM workflow with periodic operation driven by channel load measurements Figure 18: RM workflow with hybrid operation driven by channel load measurements and message services A more detailed operation of the bandwidth management performed by RM is illustrated in Figure 19 for the periodic approach where RM operates according to channel load measurement cycles. As can be observed, during the first RM interval, t = [0, ∆T), there is no information yet available about the current channel load, so all message services operate in an open-loop mode - each one generating message according to its default behaviour or immediate needs. At the end of this first interval, t = ∆T, the RM module obtains a new channel load measurement and collects the individual resource needs Ri from each active message service i. Based on the aggregate demand = ∑ , and the channel load, RM computes the available resources for the station and for each message service, referred to as δi for message service i. In the second RM interval, t = [∆T, 2∆T), message services adjust their message generation rates according to the assigned resources, thereby entering a closed-loop control phase. At the end of this second interval, the process is repeated: a new measurement of the channel load is taken, updated message service demands are reported, and RM recomputes the available resources for the station and for each message service. This approach that computes the available resources for the station at the FL opens the door to handle scenarios with heterogeneous resource needs (see clause 6.4). Congestion control mechanisms such as achieving weighted-fairness [i.30] could be applied to allow that different ITS-S experiencing the same channel load have a different number of available resources depending on their message service needs. Time Resources Resources for service A Resources for service B Services send new or updated required resources to RM Resources for service A Resources for service B Resources for service C RM informs the services about the updated resources Resources for service A Resources for service B Resources for service C Services send new or updated required resources to RM RM informs the services about the updated resources RM receives new channel load measurements Time Resources RM receives new channel load measurements Resources for service A Resources for service B Resources for service A Resources for service B Resources for service C A new service C asks for resources Resources for service A Resources for service B Resources for service C RM informs the services about the updated resources RM informs the services about the updated resources Resources for service A Resources for service B Resources for service C Resources for service A Resources for service B Resources for service C Service B requests more resources RM informs the services about the updated resources RM informs the services about the updated resources ETSI ETSI TR 104 073 V2.2.1 (2026-02) 55 Figure 19: Interaction between RM and message services for bandwidth management with periodic operation |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5 Resource Management approach | |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.0 Overview | While RM can become highly complex when all functionalities are implemented in detail, a basic yet effective approach is achievable by focusing on essential requirements. The key idea is to implement a simple RM mechanism that still fulfils necessary operational and regulatory needs without the introduction of too much complexity. At the core of a basic solution is the division of functions between two primary entities: Admission Control (AC) and Bandwidth Management. The AC serves as the initial gatekeeper by applying straightforward policies to regulate message service activation. For example, it may restrict the activation of certain message services in some regions, or the simultaneous activation of multiple message services in scenarios where their combined transmission rates might exceed system limits. This regulation ensures that certain message services are either disallowed in specific regions or managed to avoid extreme cases, such as preventing three message services disseminating at high frequencies when the overall system can only support a lower cumulative rate. Once message services pass through AC, the Bandwidth Management is in charge of computing the resources available for the ITS-S and distributing them among the message services. In this approach, RM calculates (at FL, see clause 5.5) the resources available for the ITS-S according to the current channel load and distributes it among the active message services according to their priority and requirements. Each message service receives an allocation expressed as e.g. bits per second, which is derived from translating the computed resources and the ALI with the table discussed in clause 5.5. While a fixed ALI might be sufficient for basic implementations, more advanced setups could dynamically adjust the ALI based on factors like available bandwidth, radio access technology, or modulation and coding schemes. In addition to Admission Control and Bandwidth Management, the RM process may include a message handling entity. In the basic solution, the system does not actively enforce traffic shaping or policing. Instead, it assumes that message services will adhere to their resource assignments and that any temporary excesses will not significantly disrupt overall performance. Continuous monitoring of resource consumption - for instance, measuring the number of bits generated by each message service in the last second - provides feedback to adjust future resource distributions. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.1 RM design and implementation | The proposed RM implements a Bandwidth Management mechanism, and assumes that Admission Control simply admits all the message services implemented in the ITS-S. The Bandwidth Management mechanism operates periodically. The process is triggered when a new CBR measurement is received from the lower layers, and is as follows: • RM request: RM sends a message to all registered message services asking for their requirements. • Service replies: Each message service responds with the resources it needs, defined in bits/s. Each message service estimates its requirements over the last second. Lower layers RM Service 1 Service N Lower layers Channel load RM Service 1 Service N δ N δ1 Lower layers RM Service 1 Service N RN R1 Lower layers Channel load RM Service 1 Service N t = [0, ∆T) t = ∆T t = (∆T, 2∆T) t = 2∆T δN δ1 RN R1 Control Data ETSI ETSI TR 104 073 V2.2.1 (2026-02) 56 • Computation: RM computes the maximum allowed resources for each message service based on the requirements and measured CBR. RM computes the maximum allowed resources as a proportion (δ). Algorithms inside the Bandwidth Management mechanism include: • Computation of available resources at the station: based on the Adaptive DCC approach discussed and analysed in clause 5. Two different solutions are evaluated: - Adaptive DCC. - Adaptive DCC with adaptive beta. • Distribution of resources: the available resources are distributed among the message services based on the concept of proportional fairness. Under this concept, message services with the same priority level receive resources in proportion to their requirements. Resource distribution follows a tiered process, beginning with the highest priority message services. If any resources remain, they are assigned to lower priority message services. When the available resources can only cover a certain percentage of the overall demand for message services of a given priority, each message service is allocated the same percentage of its requested data rate. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.2 Simulation scenario and settings | The scenario considered is a static 1-hop scenario with 60 vehicles. Three different generic message services are considered following 3GPP guidelines: • Message service 1 corresponds to a low-load periodic traffic model. Packets are generated every 100 ms, following a predefined pattern of packet sizes: {300 bytes, 190 bytes, 190 bytes, 190 bytes, 190 bytes}. Each vehicle starts at a random point of this sequence, which ensures variability across vehicles. The resulting average data rate is low, around 17 kbps. This message service represents lightweight periodic traffic, typical of applications that transmit small amounts of information regularly. • Message service 2 is based on a high-load periodic traffic model. The inter-packet arrival time is much shorter (10 ms), which significantly increases the data rate compared to message service 1. The packet size varies between 1 200 bytes (with probability 0,2) and 800 bytes (with probability 0,8), yielding an average throughput of approximately 272 kbps. This message service emulates applications with steady and more demanding traffic generation, where both packet size and frequency are larger than in message service 1. • Message service 3 represents an aperiodic medium-load traffic model, introducing variability in the message generation process. The inter-packet arrival time is defined as 50 ms plus an exponentially distributed random variable with a mean of 50 ms, which results in irregular message intervals. Packet sizes are uniformly distributed between 200 and 2 000 bytes, with a quantization step of 200 bytes. The average data rate is about 56 kbps. This message service mimics traffic generated by applications with event-driven or perception-related dynamics, where both packet size and message interval may fluctuate. All the message services have the same priority. Each message service adapts messages as instructed by RM: • Message service 1 adapts message interval. • Message service 2 adapts message size. • Message service 3 adapts both interval and size. Three vehicle types are considered in the scenario: • Vehicles of Type 1 implement message service 1. • Vehicles of Type 2 implement message services 1 and 2. • Vehicles of Type 3 implement message services 1, 2 and 3. In the scenario considered, the vehicle types are evenly distributed, with one third of the vehicles belonging to each type. Within each vehicle type, the message services generate packets asynchronously. The CBR is measured by each vehicle asynchronously and periodically every 100 ms as the sum of the proportion of time that the channel is busy (received signal higher than a threshold) and the proportion of time used for transmission. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 57 All vehicles run RM at the FL with ITS-G5 at the lower layers with a default data rate of 6 Mbps. Simulations were conducted using INET Framework with Veins over OMNeT++. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.3 Results | |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.3.0 Overview | This study starts with the analysis of the time evolution of the requirements of each message service (Figure 20(a)) and each vehicle type (Figure 20(b)). The requirements are computed by each vehicle per message service based on the last second and normalized by the default data rate. The measurements of required resources revealed the differences among the three vehicle types considered. Vehicles of Type 1, implementing only message service 1, demanded the lowest amount of resources, while vehicles of Type 2, running message services 1 and 2, exhibited intermediate requirements. Finally, vehicles of Type 3, implementing all three message services, were characterized by the highest demands. These disparities were consistent with the design of the message services, where message service 1 generated a low traffic load, message service 2 a medium one, and message service 3 added further aperiodic traffic. (a) (b) Figure 20: Average time evolution of the message service and vehicle requirements |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.3.1 Adaptive DCC | This clause presents the results obtained considering the default configuration of Adaptive DCC for the computation of available resources at the station. Figure 21 illustrates the time evolution of the CBR measured by the vehicles in the considered scenario. The figure differentiates the median, as well as the 5th, 25th, 75th and 95th percentiles. The results show that the RM mechanism achieved the expected behaviour of stability and convergence, without significant difference among vehicles in the scenario. Figure 21: Time evolution of the CBR (median and different percentiles) The evaluation of the maximum allowed resources per ITS-S, expressed through the delta (δ), showed that all vehicle types ended up being assigned the same value, despite their very different requirements (shown in Figure 20(b)). In practice, this means that each vehicle was assigned the same share of resources, regardless of whether it implemented one, two, or three message services. This uniform distribution is a direct consequence of the use of the Adaptive DCC solution that computes delta only based on the CBR, i.e. without considering the message service requirements. 300 320 340 360 380 400 Time (s) 0 0.01 0.02 0.03 0.04 0.05 0.06 Service 1 Service 2 Service 3 300 320 340 360 380 400 Time (s) 0 0.01 0.02 0.03 0.04 0.05 0.06 Vehicle type 1 Vehicle type 2 Vehicle type 3 300 320 340 360 380 400 Time (s) 0.5 0.55 0.6 0.65 0.7 0.75 0.8 P5 P25 P50 P75 P95 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 58 Figure 22: Time evolution of the average delta computed per vehicle type As a result, the actual resource usage reflected these limitations, as illustrated in Figure 23. Type 1 vehicles, having modest requirements, were able to satisfy their message service demands fully with the allocated share. In contrast, Type 2 vehicles had to proportionally reduce the resources devoted to message services 1 and 2, while Type 3 vehicles faced even stronger reductions across all three message services. This behaviour confirmed that, although the RM ensured fairness at the station level, it did not account for the heterogeneous requirements introduced by different message service sets. (a) message service 1 (b) message service 2 (c) message service 3 Figure 23: Time evolution of the average resources assigned per message service and vehicle type In summary, the results of the one-hop scenario highlighted the limitations of the baseline RM design: even though it guaranteed stability and convergence, it allocated identical resources to all vehicles, leading to disproportionate reductions for those with higher demands. This finding motivates the need for re-designs that explicitly take into account the heterogeneity of message service requirements across stations, which are presented in clause 7.5.3.2. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.3.2 Adaptive DCC with dynamic beta | In order to overcome the limitations observed in the previous clause, where all vehicles were assigned the same number of resources regardless of their actual demands, an adaptive beta mechanism is described and evaluated in this clause. This approach is based on the weighted LIMERIC scheme [i.30], that ensures that vehicles running the Adaptive DCC mechanism with different beta values converge to delta values proportional to those betas. In other words, if one vehicle has a beta that is twice as large as another, its steady-state delta will also be twice as large, and the same principle holds for any proportionality factor. In this clause, RM configures its beta value proportionally to the resources it requires. In practice, this means that vehicles with higher demands dynamically increase their beta, while vehicles with lower requirements maintain smaller values. In the proposed design, the RM computes the beta parameter at every RM interval, and all vehicles should follow the same computation method to ensure consistent behaviour across the network. Backwards compatibility is guaranteed, since Release 1 stations continue to operate with a constant beta value. 300 320 340 360 380 400 Time (s) 0.005 0.01 0.015 0.02 0.025 0.03 Vehicle type 1 Vehicle type 2 Vehicle type 3 328.2 328.3 9.78 9.8 9.82 9.84 10-3 300 320 340 360 380 400 Time (s) 0 0.5 1 1.5 2 2.5 3 10-3 Vehicle type 1 Vehicle type 2 Vehicle type 3 V2X service 1 300 320 340 360 380 400 Time (s) 0.005 0.01 0.015 0.02 0.025 0.03 Vehicle type 2 Vehicle type 3 V2X service 2 300 320 340 360 380 400 Time (s) 1.5 2 2.5 3 3.5 4 4.5 5 10-3 Vehicle type 3 V2X service 3 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 59 A linear adaptive beta solution is introduced, where beta is computed proportionally to the resources required by each station. Formally: where: • BETA_BASE is the beta value used by Release 1 vehicles using Adaptive DCC. • DATA_RATE_BASE corresponds to the average resources required by Release 1 vehicles using Adaptive DCC. • RequiredResources is the sum of the resources demanded by all message services implemented in the station. This formulation ensures that vehicles with larger requirements obtain proportionally larger beta values. Consequently, the algorithm is expected to converge to delta values that scale with beta. This property enables the RM to allocate resources fairly among heterogeneous stations, while preserving compatibility with legacy vehicles. The results of the considered scenario with adaptive beta showed a clear improvement in the alignment between allocated resources and actual message service requirements. The computation of beta values across vehicles confirmed that the adaptation worked as intended: vehicles of the same type exhibited similar beta values, with small deviations due to the stochastic nature of their traffic generation. These differences, however, did not compromise the overall stability of the system. The measurement of the CBR further validated the proposed solution. Stability and convergence were preserved, as in the baseline case, but the system now converged closer to the target CBR. This improvement was associated with the larger beta values assigned to vehicles with greater needs, which allowed a more efficient utilization of the channel. Figure 24: Time evolution of the CBR (median and different percentiles) The maximum allowed resources per vehicle type, expressed as delta (δ), also reflected the expected behaviour, as observed in Figure 25. Unlike the previous case where all vehicles were assigned the same δ, here vehicles with higher resource requirements obtained proportionally more resources. Within each vehicle type, the δ values were nearly identical, with only small deviations, while across types, the differences were consistent with their respective Message service demands. Vehicles of Type 1, which had the lowest requirements, showed slower convergence in their δ values, an effect attributed to their reduced traffic generation and the initialization of delta to δmax= 0,03. Figure 25: Time evolution and boxplots of the delta computed per vehicle type 300 320 340 360 380 400 Time (s) 0.5 0.55 0.6 0.65 0.7 0.75 0.8 P5 P25 P50 P75 P95 300 320 340 360 380 400 Time (s) 0 0.005 0.01 0.015 0.02 0.025 0.03 Vehicle type 1 Vehicle type 2 Vehicle type 3 0 10 20 30 40 50 60 Vehicle ID 0 0.005 0.01 0.015 0.02 0.025 0.03 Vehicle type 1 Vehicle type 3 Vehicle type 2 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 60 When analysing the actual usage of resources, the benefits of the adaptive beta approach can be observed in Figure 26. Vehicles of Type 2 and Type 3 adapted rapidly, reducing or maintaining their service loads in proportion to the resources allocated. Vehicles of Type 1, however, exhibited a slower adaptation process, again linked to their smaller demands and lower traffic intensity. Despite this, the system as a whole achieved a balanced allocation, i.e. all message services converge to the same delta, that better reflected the heterogeneity of the scenario. (a) message service 1 (b) message service 2 (c) message service 3 Figure 26: Time evolution of the average resources assigned per message service and vehicle type Finally, the evaluation of the message service satisfaction ratio - defined as the ratio between the resources assigned and the resources required - showed that all message services across all vehicle types converged towards the same value, approximately 0,29 (see Figure 27). This result demonstrates that the adaptive beta mechanism not only ensures proportional fairness across heterogeneous vehicles but also equalizes the level of satisfaction among message services, regardless of the vehicle type implementing them. (a) message service 1 (b) message service 2 (c) message service 3 Figure 27: Time evolution of the message service satisfaction ratio per message service and vehicle type In summary, the one-hop scenario with adaptive beta highlighted the effectiveness of the proposed re-design. By dynamically adjusting the beta values according to the actual requirements of each vehicle, the Bandwidth Management mechanism of RM was able to provide more resources to vehicles with greater needs, maintain system stability, and bring the operation closer to the CBR target. At the same time, it preserved fairness among message services by equalizing their satisfaction levels, thereby overcoming the limitations of the baseline configuration. |
fbcda0cf1f08aabcdc0778bd9c7a2a39 | 104 073 | 7.5.4 Conclusion Resource Management approach | The results demonstrate that an RM approach running at the FL based on Adaptive DCC converges properly and provides stable operation, as also shown in clause 5 and Annex C. However, when applied to scenarios with heterogeneous message services of equal priority, the approach exhibits clear limitations, since it does not fully account for the diversity of message service requirements across stations. A promising solution would be to allow the RM to dynamically adapt the configuration of Adaptive DCC according to the resources demanded by each vehicle, thereby improving fairness and efficiency in heterogeneous deployments. Nevertheless, further investigation is required to assess the behaviour of this solution in scenarios where message services are assigned different priorities, which remains an open line of analysis. 300 320 340 360 380 400 Time (s) 0.5 1 1.5 2 2.5 3 10-3 Vehicle type 1 Vehicle type 2 Vehicle type 3 V2X Service 1 300 320 340 360 380 400 Time (s) 0.01 0.015 0.02 0.025 0.03 Vehicle type 2 Vehicle type 3 V2X Service 2 300 320 340 360 380 400 Time (s) 2 2.5 3 3.5 4 4.5 5 10-3 Vehicle type 3 V2X Service 3 300 320 340 360 380 400 Time (s) 0 0.2 0.4 0.6 0.8 1 1.2 Vehicle type 1 Vehicle type 2 Vehicle type 3 V2X service 1 300 320 340 360 380 400 Time (s) 0 0.2 0.4 0.6 0.8 1 1.2 Vehicle type 2 Vehicle type 3 V2X service 2 300 320 340 360 380 400 Time (s) 0 0.2 0.4 0.6 0.8 1 1.2 Vehicle type 3 V2X service 3 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 61 Annex A: Rationale for the limits imposed by congestion control The rationale for equation (2) in clause 5.2.4, which is derived from ETSI EN 302 571 [i.16], is described in ETSI TS 103 175 [i.15] and hereafter elaborated. The starting point is the derivation of the maximum channel occupation of the generic station, hereafter called , as a function of the overall channel load, which is measured by the . The maximum channel occupation can also be written as a function of the minimum time between two consecutive transmissions where the transceiver is not allowed to generate a signal, denoted as and the duration of the last transmission as: = (A.1) Since the maximum channel occupation of the generic station needs to depend on the number of stations, but this information is not explicitly available, an assumption is made on the CBR. In particular, it is assumed that the maximum CBR, denoted as , is a function of the number of stations concurring to access the same channel in the same area, denoted as . This assumption is introduced to allow an implicit derivation of the number of stations directly from the measured . This assumption can also be written as: = × + (A.2) where and are two constants, and thus: = (A.3) Given the maximum CBR and the number of stations, the maximum channel occupation of the generic station can be calculated as: = = = × (A.4) Using equations (A.1) and (A.4), by first writing as a function of and then substituting with its expression as a function of , resulting in: = × = × −1 = × −1 (A.5) The final step is that the measured CBR, indicated as , is assumed equal to , which brings to: = × −1 (A.6) Which corresponds to equation (2) in clause 5.2.4 (and thus the limit indicated by ETSI EN 302 571 [i.16]) when = 1/4 000 and = 0,62 are used. These two parameters were empirically determined to limit the packet collision ratio and to provide a robust feedback for deriving from the measured CBR. It can be observed that the assumption that the measured CBR is equal to the limit implies that each station tends to consider the channel always congested, which may not be true. However, if the channel is not congested the effect is that the station under observation underestimates the number of contending stations and overestimates the portion of resources it can use. Given that the channel is not really congested, this may only cause an increase of the CBR until it actually reaches the level of congestion. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 62 Annex B: Methods to characterize congestion control algorithms B.1 Modelling of a rate control loop B.1.1 Definition of the channel busy ratio limit Several independent transmitters that want to share a radio channel without coordination by a central station need a mechanism that avoids data packet collisions and that takes care of the available channel resources. For simplification of the derived model of the congestion control it is assumed that data packet collisions are effectively avoided by the CSMA/CA algorithm when the radio channel is not overloaded and that the load is only controlled by the rate and duration of the data packets. This simplifies the description and makes an analytic evaluation possible. Anyhow, the congestion control is designed to avoid packet collisions by keeping the channel load reasonably low. Obviously, all transmitters should not try to put more packets on the channel than the channel capacity allows. It is even so that the full channel capacity can only be reached at the cost of massive packet collisions (packets of different transmitters overlap in time). Most of these overlapping packets cannot be decoded in the receivers and are therefore waste of channel resources. Hence, the optimum channel load is well below the channel capacity. A congestion control should allow a system of multiple transmitters to use the channel resources up to, or close to, the optimum channel load, and avoid an operation above this load limit. The channel resources R can be seen as percentage of the total available transmit time for all transmitting nodes (0 < R < 1). Whereas the channel utilization u is the percentage of the total available time each node can transmit. Assuming that all nodes N share the available resources equally, u can be found by dividing R by N (equation (B.1)): = (B.1) When the channel load is the only input parameter to the control algorithm, the transmitters do not know the number of other transmitters contributing to the channel load, and the distribution of the channel load cannot be done by just dividing the available resources by the number of nodes. The control algorithm can only inherently divide the resources equally from only knowing the channel load. This is possible when the channel load has a one-to-one relation to the number of nodes. ETSI specifies in ETSI TS 103 175 [i.15] a linear relation between the channel load limit and the number of contributing nodes (see also Annex A and equation (B.2)): = × + (B.2) Where CBRLimit is the upper channel busy ratio limit that should not be exceeded by the measured channel load CL and a and b are parameters. An example for a = 1/4 000 and b = 0,62 based on ETSI EN 303 797 [i.32] is shown in Figure B.1. Figure B.1: Example of a CBR limit according to equation (B.2) The measured channel load results from the sum of the channel utilizations u of all nodes (equation (B.3)): = ∑ (B.3) 0,0 0,2 0,4 0,6 0,8 1,0 0 100 200 300 400 500 600 700 CBR Number of ITS-S ETSI ETSI TR 104 073 V2.2.1 (2026-02) 63 Figure B.2 shows the contribution of each rate controller to the channel load. Other types of congestion control are not considered in the present document since they are less effective and more complicated to implement. Figure B.2: Contribution of each rate control to the channel load When all rate controllers are the same, all channel utilizations uk are equal to u and the resulting channel load CL is given by equation (B.4): (B.4) When these equal rate controllers are working properly and equilibrium is reached, all uk are equal and the resulting steady state channel load is given by equation (B.5): (B.5) B.1.2 Function of the rate controller The channel utilization u of each node is controlled by adjusting the time Toff in between two transmitted packages according to the given duration Ton of the package (equation B.6), so that the resulting utilization u does not exceed the limit Umax given by (equation (B.7) and equation (B.8)): (B.6) (B.7) (B.8) Hence, the rate controller is working in discrete time steps with a variable length of Toff + Ton. The number of nodes N is unknown to each node, but an upper bound for the number of nodes Nest can be estimated from the measured channel load CL when assuming that CL ≤ CBRLimit and by substituting CBRLimit by CL in equation (B.2). Since CL can be smaller than b, the minimum number of estimated nodes Nest is fixed to at least one in equation (B.9): 1; (B.9) From this a lower bound of Umax can be calculated by use of equation (B.7) and equation (B.2) when substituting N with Nest (see equation (B.10)): ∙ (B.10) The distributed rate controllers are not synchronized and therefore the measurement period τ for the channel load determination should be long enough. Either 100 ms or a duration of Toff +Ton are considered. Also, a dissemination of the CL values between the nodes can be foreseen to increase the robustness of the measurement. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 64 B.1.3 Structure of a feedback controller Figure B.3 shows a common basic structure how a discrete time rate control in each node can be implemented. The channel load CL(t-τ) measured in the previous time step at t-τ and the previous dynamic upper bound of the channel utilization umax(t-τ) is used by the control function cfn and the filter function α × ffn to determine the next dynamic channel utilization limit umax(t). NOTE: The filter function ffn might not be used, and even the factor α is set to one in most implementations. Figure B.3: Block diagram of the rate control Equation (B.11) relates to the rate controller shown in Figure B.3 and results in the dynamic channel utilization limit umax(t): ffn cfn 1 ffn (B.11) When no filtering to the input signal CL(t-τ) is used in the controller, the filter function is constant and equal to one (α × ffn = 1). In this case the dynamic channel utilization umax(t) is just given by the control function cfn(CL(t-τ)) = umax(t). This type of controller implementation is the simplest one, but it is prone to instabilities, since the stability and the control equilibrium are both given by the control function. Using a filtering function that differs from one (α × ffn ≠ 1), offers the possibility to decouple the control equilibrium from the controller stability as will be shown in clause B.2. B.2 Control equilibrium The control equilibrium for a control function ffn is the steady state channel load CL(t-τ) = CL(t) = CL produced by a given number of nodes N that are all utilizing the channel with the same constant umax = u1 = u2 =…= uN = Umax. For the steady state equilibrium equation (B.11) simplifies to equation (B.12): ffn cfn 1 ffn (B.12) From equation B.12 the channel utilization limit Umax as function of the channel load CL can be determined: cfn (B.13) Equation (B.13) shows that the filter function α×ffn has no influence on the control equilibrium CL and on the steady state channel utilization Umax. Only the control function cfn determines the steady state characteristics of the controller shown in Figure B.3. Clause B.3 and clause B.4 will show that the controller stability and the controller dynamics are not only influenced by the control function cfn but also by filter function α × ffn. Finally, when substituting Umax=u into equation (B.4), the steady state control equilibrium can be calculated from the control function cfn for a given number of nodes N when solving equation (B.14) for : cfn (B.14) ETSI ETSI TR 104 073 V2.2.1 (2026-02) 65 B.3 Stability B.3.1 Different types of stability Congestion control algorithms are implemented as discrete-time controllers. That means that they will change the control value (channel utilization u) not at any arbitrary point in time, but only after discrete time intervals τ. This implies that a stability evaluation cannot be done by inspecting the continuous time differential equation of the control loop alone. In addition, discrete time oscillations, caused by too long intervals τ, need to be studied. Controller stability is given when within a given range of node numbers N the control loop converges independently of the initial channel load CL(t0) to a channel load CL(t) that is bounded within a defined small range (bounded stability). Such a bounded stability criteria is necessary to account for quantization steps in time and channel utilization. B.3.2 Description of the controller by a differential equation To analyse the dynamic behaviour of the distributed controllers the discrete time control equation (B.11) is converted into the differential equation (B.17). This is done in a first step by dividing equation (B.11) by the discrete time step size τ and rewriting it in such way that the difference equation (B.15) is obtained. The left side of equation (B.15) is the ratio between the utilization difference Δumax = umax(t)-umax(t-τ) and the time difference Δt = τ between umax(t) and umax(t-τ). When additionally substituting t-τ by t-Δt on the right side of equation (B.15), equation (B.16) is obtained. Under the assumption that the control function cfn and the filtering function ffn are both analytic, the difference equation (B.16) can be converted by a limiting process into the differential equation (B.17). Where u(t) is the continuous time function corresponding to the discrete time function umax(t): = ×ffn × cfn − − − (B.15) ∆ ∆ = ×ffn ∆ × cfn −∆ − −∆ lim ∆ → (B.16) = = ×ffn × cfn − () (B.17) For N identical controllers, the time behaviour of the channel utilization u(t) as function of the number of nodes N results from solving differential equation (B.18): = ×ffn× × cfn × − (B.18) By use of equation (B.4) also the time behaviour of the channel load CL(t) can be obtained from equation (B.18). B.3.3 Convergence Even when equation (B.14) has a solution for the equilibrium channel load that lies within 0 and 1 as given by the definition of the channel load, the controller might not converge to this equilibrium. This can be the case when the slope of the control characteristic is not monotonic towards the point of equilibrium. This could be caused by an ill formed control function with ripples in the slope, or when equation (B.14) has more than one solution for . For convergence following criteria based on equation (B.16) should be met: cfn × − < 0 for 1 > > and cfn × − > 0 for 0 < < (B.19) Where is the controller equilibrium as calculated from equation (B.5) an equation (B.14). ETSI ETSI TR 104 073 V2.2.1 (2026-02) 66 B.3.4 Continuous-time stability The convergence criteria given by equation (B.19) are not sufficient to guarantee a stable control loop. Only when in addition the solution of differential equation (B.18) shows at least a decaying oscillation, or even no oscillation, for the channel utilization u(t) within a given range of node numbers N and for all starting values 0<u(t0)<1, the continuous- time control loop is stable for this range of N. B.3.5 Discrete-time Stability Since the congestion control is implemented as controller that measures the channel load CL and changes the channel utilization umax repeatedly after discrete-time intervals τ, the continuous-time stability as given in clause B.3.4 is a necessary, but not sufficient criteria for the stability of the control loop. When the continuous time solution converges for all stations to the same stable value (see equation (B.5)), the discrete-time controller can be unstable when u(t) changes in one time step within an interval of τ by at least twice the distance to (overshoot). The u(t) change over one time step with a duration of τ equals × . Where is the slope of the u(t) function at the time t. From this, the discrete time realization for a controller that overshoots is stable when equation (B.20) is fulfilled for all t ≥ 0 within a given range of node numbers N and for all starting values 0 < u(t0) < 1: × | ′| < 2 × | − | (B.20) This implies that a discrete time controller is also stable when the channel utilization does not overshoot the controller equilibrium as expressed by equation B.21. When equation B.21 is met, the controller shows a favourable behaviour without discrete time oscillations: × | ′| ≤| − | (B.21) Since τ is always positive, equation (B.20) and equation (B.18) can be combined to form the discrete-time stability criterium given in equation (B.22): × ffn × × cfn × − < 2 × | − | (B.22) This criterium is a function of the channel utilization and the node number N. It should be fulfilled within the whole range of 0 < u(t) <1 and within the given range of node numbers N where the controller should be stable. B.4 Control dynamics The convergence speed of the controller can be characterized with a channel load step function. Since the equilibrium might not be exactly reached because of quantization effects and measurement noise, a definition of the convergence speed by the time until a certain percentage of the channel load equilibrium is reached can be used. In addition, it makes sense to characterize the decay time of the controller oscillations, since it can be much longer that the time the channel utilization reaches a certain percentage of the equilibrium. Another criterium to characterize the control dynamics is the overshoot and undershoot relative to the equilibrium channel utilization when a channel load step function is applied. Also, the step size of the channel utilization quantization is an important parameter when defining the bounded stability criteria. To get a first impression of the convergence speed and the decay time of the controller oscillations an inspection of the result of the continuous time differential equation (B.18) is helpful. But this result is just a guide value for the discrete time controller as given by equation (B.11). The discrete time implementation can converge even faster than the continuous time controller when × | ′| = | − |, or much slower when only equation (B.22) is fulfilled. The control dynamics of the discrete time controller can be calculated iteratively from equation (B.11). Only for very simple control functions cfn and filtering functions ffn exact closed form analytic solutions are possible for the dynamics of a discrete time controller. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 67 Annex C: Characterization of congestion control algorithms C.1 Reactive congestion control C.1.1 Control equilibrium of the reactive congestion control The reactive congestion control algorithm is based on a control function for Toff (equation (C.1)): () = ( −) × + (C.1) Where Ta and Tb are parameters and the channel busy ratio CBR is the measured channel load. This implies that the equilibrium channel load CL is not only depending on the number of nodes N, but also on the duration of the transmissions Ton. This follows from equation (B.4) and equation (B.6) when assuming that CL is equal CBR (CL = CBR): = × (C.2) When combining equation (C.1) with equation (C.2) the Toff control function can be rewritten to: () = × ( ) ( ) ( ) × + (C.3) To simplify the calculation, it is assumed that all nodes use the same time independent Ton = const. With this assumption and based on equation (C.3) the equilibrium can be calculated from equation (C.4): = × × + (C.4) What leads to a quadratic equation for : + × − − × + × = 0 (C.5) Equation (C.5) has following two solutions for : = − ± + + 4 × × × (C.6) The channel load equilibrium results from equation (C.2) when substituting Toff by obtained from equation (C.6). = ×× ± × ×× (C.7) As a result of the quadratic equation (C.5) there are two channel load equilibrium solutions given by equation (C.7). For the control stability this requires that only one solution lies within the channel load boundaries of 0 < CL < 1 and an analysis of the convergence within this channel load range is necessary. As an example, from Table A.1 and Table A.2 in ETSI TS 102 687 [i.14] the parameters Ta and Tb that fit best to the table entries can be calculated by a linear regression as shown in Figure C.1 and Figure C.2. The results are listed in Table C.1. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 68 Figure C.1: Linear regression of the DCC function given in Table A.1 of ETSI TS 102 687 [i.14] Figure C.2: Linear regression of the DCC function given in Table A.2 of ETSI TS 102 687 [i.14] Table C.1: Parameters Ta and Tb derived from ETSI TS 102 687 [i.14] Parameter Ton < 0,5 ms 0,5 ms ≤ Ton ≤ 1 ms Ta 598 ms 1 295 ms Tb -127 ms -285 ms Figure C.3 and Figure C.4 show the positive channel load equilibria calculated from equation (C.7) as function of the node number N and the parameters given in Table C.1. The dashed lines show the results outside the Ton range to show that these results either over utilize or underutilize the channel for large node numbers. The dotted line is the CBR limit for the example based on ETSI EN 303 797 [i.32] shown in Figure C.3. y = 1294,5x - 284,93 0 200 400 600 800 1000 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 Toff/ m s channel load CL Reactive DCC characteristic for Ton ≤1 ms y = 598,13x - 126,64 0 200 400 600 800 1000 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 Toff/ m s channel load CL Reactive DCC characteristic for Ton < 0,5 ms ETSI ETSI TR 104 073 V2.2.1 (2026-02) 69 Figure C.3: Equilibrium channel load of reactive congestion control for Ta and Tb from Table C.1 for 0,5 ms <Ton ≤ 1 ms Figure C.4: Equilibrium channel load of reactive congestion control for Ta and Tb from Table C.1 for Ton ≤ 0,5 ms C.1.2 Convergence of the reactive congestion control To analyse the convergence, the channel load control function cfn of the reactive controller is be determined by first substituting equation (C.1) in equation (C.2) under the assumption that CL=CBR: (C.8) With the help of equation (B.14) the channel load control function cfn can be found from equation (C.8): cfn (C.9) To check the convergence criteria given by equation (B.19), the channel load control function cfn (equation (C.9)) is used: 0 for 1 | and 0 for 0 | (C.10) ETSI ETSI TR 104 073 V2.2.1 (2026-02) 70 Where | represents the solution of the equilibrium channel utilization that is within the allowed range of 0 1 obtained from equation (B.5) and equation (C.7). In case equation (C.7) has two solutions within this range, convergence is only given when the range of u is limited in such a way that only one solution is within this range and equation (C.10) is fulfilled there. Or in terms of CL equation (C.10) can be written as equation (C.11): cfn 0 for 1 and cfn 0 for 0 (C.11) When applying the convergence criteria to the examples based on ETSI TS 102 687 [i.14] given in in Table C.1 it can be seen form Figure C.5 and Figure C.6 that the reactive congestion control will converge to the negative equilibrium when CL is somewhere below 0,2. The correct convergence area above a CL of around 0,2 to the positive CL equilibrium is highlighted in green, the wrong convergence to the negative CL equilibrium is marked in red, and the negative (impossible) CL area is marked in white. This is the reason why this control algorithm needs to be implemented by a table that has no entries below a CL of 0,3 to avoid a convergence to the impossible negative CL equilibrium. The convergence direction is given by cfn which has a pole when gets zero. Since Ta and Ton are always positive, a pole for a positive CL can only happen when Tb is negative. For the example given in in Table C.1 and when assuming that the Ton values are below 2 ms the pole is at a 285 !" 1 295 !" $ 0,22 ⁄ or at a 127 !" 598 !" $ 0,21 ⁄ as shown in Figure C.5 and Figure C.6. Figure C.5: Convergence criteria function for N = 100, Ton = 1 ms, and the parameters given in Table C.1 Figure C.6: Convergence criteria function for N = 100, Ton = 0,5 ms, and the parameters given in Table C.1 C.1.3 Stability of the reactive congestion control C.1.3.1 Stability of the reactive congestion control without filtering according to ETSI TS 102 687 C.1.3.1.1 Continuous time stability According to clause B.1.3 the control equation without filtering for a reactive congestion control is: ( (C.12) ETSI ETSI TR 104 073 V2.2.1 (2026-02) 71 Equation (C.12) leads with equation (B.17) and equation (B.4) to differential equation (C.13) hat can be solved by separation of the variables: (C.13) ) * + * 1 +( (C.14) After some calculations this leads to a solution for t in relation to the step size ) given by equation (C.15): log . on log × (C.15) Where T0 is an integration constant to adjust the time offset and A and B are constants given by: / 2 4 1 and . (C.16) The solution of the differential equation shows no oscillating term. Together with the convergence criteria and the initial value, this defines a parameter range and boundary conditions where a stable convergence to the control equilibrium of the continuous time reactive congestion control algorithm can be granted analytically. Figure C.7 shows in red the solution given by equation (C.15) for N = 100, Ton = 1 ms, and the parameters given in Table C.1. Where the offset T0 was adjusted so that the time t is zero for a channel load CL equal to one. The blue crosses show a numerical solution starting at a CL of one. The step size for this numeric solution is ) 50 ⁄ . With this small step size, the difference to the analytic solution is neglectable. Furthermore, the analytic solution shows for a CL below 0,22 that the CL converges with increasing time t to decreasing channel load CL values and does not reach the positive equilibrium at CL = 0,405 as also shown in Figure C.5. Figure C.7: Analytic and numeric solution of the continuous time reactive congestion control for N = 100, Ton = 1 ms, and the parameters given in Table C.1 C.1.3.1.2 Discrete Toff stability and convergence speed Figure C.7 shows that a discrete time numeric solution of the reactive congestion control can reproduce the analytic solution when the time step size is small enough. This numeric solution uses 100 steps to converge to a value close to the equilibrium. Assuming a time step size of 100 ms, such an implementation could need around10 seconds to converge to a CL close to the control equilibrium, what is much longer than the time constants to be expected in mobile radio channels and road traffic scenarios. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 72 The numeric solution shown in Figure C.7 is calculated iteratively using equation (C.17) and the parameters given in Table C.1 when setting = 0,02: = − + × × × − − (C.17) Equation (C.17) follows from equation (B.11) when using cfn from equation (C.9), setting ffn to one, and using equation (B.4) to calculate CL. α corresponds to the ratio . Where is a fixed time step size used for the numeric solution. α defines how many iteration steps from zero to the time τ of the analytic solution are calculated. Hence, = 0,02 used for Figure C.7 corresponds to 50 iteration steps per τ. α is needed to scale the numeric solution in the same way as the analytic solution. It is obvious that the step size of the numeric solution needs to be small to resemble the analytic solution. For a discrete-time congestion control α together with ffn is also essential to fulfil the stability criterium given in equation (B.22). For a control algorithm it is not essential to resemble the analytic solution of the control function, it should be stable and converge quickly to the equilibrium. Therefore, in ETSI TS 102 687 [i.14] a state machine approach was proposed that discretises the CL and the Toff time into fife value pairs as shown in Figure C.1 and Figure C.2. To limit the Toff step size for each time step, the state machine will only transit from one state to the consecutive one. It is not allowed to step over a state, even when the measured CL is not in the range defined for the consecutive state. To determine the direction of the state transition, the CL range of the current state is compared with the measured CL. If the measured CL is within the CL range of the current state, then the state and the Toff time are not changed. Otherwise, depending on whether the measured CL exceeds or is below the CL range of the current state, the state transits to the next or the previous state with a longer or shorter Toff time. Figure C.8 shows as solid-coloured lines the CL that results from the Toff values given in Table A.1 of ETSI TS 102 687 [i.14] for Ton = 1 ms. These are straight lines, what follows from equation (C.2). Each line corresponds to one of the states ( to ) of the state machine. The dashed horizontal lines are the CL limits for a state transition that correspond to the state with the same colour. The red circles show where in the diagram the table entries for Ton = 1 ms are located. For comparison, the control equilibrium for continuous CL values according to equation (C.7) with Ta and Tb from Table C.1 is shown as black line. NOTE: For other values of Ton the number of nodes where the state transition limit is reached would be different, as follows from equation (C.2). Figure C.8: CL resulting for the Toff values taken from the reactive DCC Table A.1 in ETSI TS 102 687 [i.14] for Ton = 1 ms Figure C.9 takes a closer look on the control equilibrium of the table based DCC proposed in ETSI TS 102 687 [i.14]. The equilibrium is shown as bold black line. It can only be reached as average over time when the nodes are not synchronized and when only state transitions to the consecutive states are allowed. 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 0 100 200 300 400 500 600 700 800 900 1000 channel load CL number of nodes N Table Limit Equilibrium 100 200 400 500 1000 Toff / ms ETSI ETSI TR 104 073 V2.2.1 (2026-02) 73 The shaded areas (blue, orange, grey) are regions where not all nodes are in the same state so that in average the CL results to the respective state transition limit. Hence, in such a region the ratio between the number of nodes in one state and the number of nodes in another state depends on the total number of nodes N. EXAMPLE: For node numbers below the blue shaded area in Figure C.9 all nodes will be in state . For a number of nodes within the blue shaded area the limit of CL = 0,3 would be exceeded when all nodes are in state . Since the nodes are assumed to be not synchronized, they measure the CL at random times within a 100 ms interval. Assuming that equilibrium is reached, all nodes measuring a CL of more or equal 0,3 will apply state all others will apply state . Since the nodes are not synchronized and are measuring the CL not at the same time, they will do the measurements one after the other with a random time difference. This allows them to get different measurements results and hence to transit to different states, so that in average in this example within the blue shaded area the CL will converge to the black line shown in Figure C.9. Annex D shows how the CL measurement influences this convergence equilibrium and the convergence speed. Figure C.9: CL resulting from the Toff values taken from the reactive DCC Table A.1 in ETSI TS 102 687 [i.14] for Ton = 1 ms and the bounded stability regions up to N = 300 From Figure C.9 it also gets clear why the nodes should not be synchronized and only transitions between consecutive states are allowed. For the orange shaded area in Figure C.9 it can be seen that when all nodes are synchronized and in state the CL will be below the limit of 0,4 and all nodes would synchronously transit to state . As a result, the CL will jump to a much higher value in the next measurement period, and all nodes will transit back to state . This results in a strong CL oscillation that can even overload the channel. For the same scenario it can be seen that when all nodes are synchronized and in state the CL can be below 0,3. When it would be allowed to transit directly to state the channel would be heavily overloaded, and the controller gets immediately instable. For not synchronized nodes this issue is not so obvious, but it still exists and can lead to serious oscillations and channel overload. C.1.3.2 Stability of the reactive congestion control with filtering according to C2C-CC In C2C-CC: "Vehicle C-ITS station profile #2037", RS_BSP_240 [i.34] a sliding average filter for the controller input signal is specified. This makes an analysis of the nonlinear control algorithm more complicated. Therefore, such an input filtering is not foreseen in clause B.1.3. To model this input filtering, the channel load at −2 × is needed. With equation C.8 this results in the dynamic channel load equation (C.18): = × × × (C.18) 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0 50 100 150 200 250 300 channel load CL number of nodes N ETSI ETSI TR 104 073 V2.2.1 (2026-02) 74 This input filtering does not change the equilibrium channel load. Because it uses the average of two previous channel loads as input, and since in the steady state case all channel loads are considered to be equal to the equilibrium channel load, the average will also be the equilibrium channel load as given in equation (C.7). Furthermore, when is close to zero, the average × will be − × : lim → × = − × (C.19) From equation (C.19) follows that differential equation (C.13), its solution in equation (C.15), and all conclusions drawn for the continuous time behaviour of the reactive approach according to ETSI TS 102 687 [i.14] also apply for the DCC as specified in C2C-CC: "Vehicle C-ITS station profile #2037", RS_BSP_240 [i.34] with a time scaling factor of two. The difference lies in the behaviour of the discrete time / discrete Toff implementation of the channel load measurement that is further detailed in Annex D. C.2 Adaptive congestion control C.2.1 Control equilibrium of the adaptive congestion control The adaptive congestion control concept is based on the idea that the radio channel can support a given maximum total target message rate rg. A Proportional Integral (PI) message rate control algorithm is used to adjust the transmission rate rj of each network node according to equation (C.20): = 1−∝ × −τ + × − −τ (C.20) − is the total rate contribution of all nodes N in the previous time step and results from equation (C.21). The time step duration is given by τ: − = × − (C.21) The target message rate rg that can be supported by the radio channel depends on the message duration, therefore this concept works only fine when the message duration is (almost) fixed and known. Otherwise, a maximum message duration Ton max needs to be estimated or considered. rg can then be set according to equation (C.22) so that the channel load is limited to e.g. CLmax = 70 %, which is a good default value: = (C.22) Due to the shortcoming that rg depends on Ton, equation (C.20) was reformulated in ETSI TS 102 687 [i.14] so that the estimated constant reciprocal message duration was put into the factor and instead of the total rate contribution r the channel busy ratio CBR which is the measured channel load, was used. This results in equation (C.23): = 1−∝ × −τ + × − −τ (C.23) This notation has the advantage that it directly uses the measured CBR and that the channel load is limited to CBRtarget independent of Ton. Since the message rate is controlled, the channel utilization of each node still depends on Ton: = × (C.24) Consequently, from equation (C.24) follows in equation (C.25) with equation (B.4) the channel load: = × × (C.25) Assuming that the CL is equal to the measured CBR and j = = − the equilibrium rate j can be calculated from equation (C.26): j = 1−∝ × j + × − j × × (C.26) j = × × × (C.27) ETSI ETSI TR 104 073 V2.2.1 (2026-02) 75 With equation (C.25) the channel load equilibrium as function of the number of nodes N follows from equation (C.27): = × × × × × (C.28) The target channel load CBRtarget is reached asymptotically for an infinite number of nodes N as follows from equation (C.28). As an example, Figure C.10 shows as function of the node number N and the Ton time as coloured solid lines for the parameter values given in Table 3 of ETSI TS 102 687 [i.14]. does not exceed the CBR DCC limit specified in ETSI EN 303 797 [i.32] (dot ted line), but for a small number of nodes and short Ton times the channel is underutilized. Figure C.10: Equilibrium channel load of adaptive congestion control for the parameter values given in Table 3 of ETSI TS 102 687 [i.14] C.2.2 Convergence of the adaptive congestion control C.2.2.1 Continuous time behaviour of the adaptive congestion control From equation (C.25) the packet rate as function of the channel load , the number of nodes , and the packet duration can be derived: = × (C.29) Substituting equation (C.29) into equation (C.23) leads to equation (C.30): × = 1−∝ × × + × − −τ (C.30) For the assumption that the packet duration is time invariant and equal for all nodes = − = equation (C.30) can be rewritten to: = × × × × − −τ × ∝+ × × . (C.31) Equation (C.31) can be written as differential equation for small τ when →0: = × × × × − × ∝+ × × (C.32) When substituting = × × × and =∝+ × × into equation (C.32) it can be rewritten after separation of the variables into: × = (C.33) 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 0 100 200 300 400 500 600 700 800 900 1000 channel load CL number of nodes N 0,25 0,5 1 1,5 2 Ton / ms ETSI ETSI TR 104 073 V2.2.1 (2026-02) 76 When integrating equation (C.33) it results to: | | (C.34) When undoing the substitution for A and B in equation (C.34) the continuous time solution for the adaptive congestion control results to: ∝ ∝ (C.35) Figure C.11 shows the analytic result calculated with equation (C.35) in red colour and a numeric result as blue crosses for the for N = 100, Ton = 1 ms, and the parameters given in in Table 3 of ETSI TS 102 687 [i.14]. The numeric result is obtained with a step size of t/τ = 1 as specified in of ETSI TS 102 687 [i.14]. In this example, for τ = 100 ms the channel load is close to the equilibrium after around 2 seconds, and in practice reaches it after 4 seconds. The convergence time of the discrete time adaptive congestion control is strongly depending on the term ∝ as will be shown in clause C.2.2.2. Figure C.11: Analytic and numeric solution of the continuous time adaptive congestion control for N = 100, Ton = 1 ms, and the parameters given in in Table 3 of ETSI TS 102 687 [i.14] There are two analytic solutions for the continuous time behaviour of the adaptive congestion control depending on whether the initial channel load is above or below the equilibrium channel load . When then results from equation (C.36): (C.36) ∝ ∝ (C.37) With from equation (C.37) results for from equation (C.38): ∝ ∝ (C.38) When then results from equation (C.39): (C.39) ∝ ∝ (C.40) ETSI ETSI TR 104 073 V2.2.1 (2026-02) 77 With from equation (C.40) results for < from equation (C.41): = × = ×× × × ∝ ×× ∝ ×× (C.41) C.2.2.2 Stability of the discrete time adaptive congestion control For calculating the stability criterium given in equation (B.22) the control function cfn, the filtering function ffn, and the equilibrium channel utilization of the adaptive congestion control is needed. From equation (C.23) and equation (C.24) the control equation of the adaptive congestion control as function of the channel utilization ua can be calculated: = 1−∝ × −τ + × × − −τ (C.42) From equation (B.11) and equation (C.42) it follows that the filter function of the adaptive congestion control ffn is constant and equal to one: ffn − = 1 (C.43) With equation (C.43) equation (B.11) simplifies to: = × cfn − + 1 − × − (C.44) From equation (C.42) and equation (C.44) the control function for the adaptive congestion control cfn can be calculated: cfn = × × − (C.45) Substituting equation (B.4), equation (C.28), equation (C.43), and equation (C.45) in equation (B.22) leads to equation (C.46): × 1 × × × − × − < 2 × × × ×× − (C.46) When factorizing equation (C.46) to equation (C.47) it can be simplified to equation (C.48) which is in line with the result given in [i.33] that was found via a series expansion approach: + × × × × × − × × × − × < < 2 × × × − × × × − × (C.47) ∝+ × × < 2 (C.48) This result means that the discrete time adaptive algorithm converges to the value from equation (C.28) when the stability criterion of equation (C.48) is fulfilled. The fastest convergence of the discrete time control is reached for ∝+ × × = 1 where the equilibrium is reached within one time step. For ∝+ × × > 1 the channel load overshoots the equilibrium and converges with an exponentially decaying oscillation. When ∝+ × × ≪1 the discrete time control converges approximately like the analytic continuous time solution given by equation (C.35). For the parameter values given in Table 3 of ETSI TS 102 687 [i.14] the stability criterion is given by × < 1 653,3 . ETSI ETSI TR 104 073 V2.2.1 (2026-02) 78 Since this criterion cannot be guaranteed for large node numbers and long packet durations, the algorithm was enhanced in ETSI TS 102 687 [i.14] by a control loop gain saturation that limits the factor of in the second term of equation (C.26) to a constant value and by an input filtering function. While the filtering function has no influence on the control equilibrium function, a saturation of the control loop gain by setting the second term of equation (C.26) to a constant slightly reduces the channel utilization and has an influence on the convergence speed (see clause C.2.2.3). In ETSI TS 102 687 [i.14] different values for are defined when is larger or smaller than −τ . For the control equilibrium only is relevant since the resulting new channel load equilibrium is below and the gain saturation is realized by replacing the second term of the control equation (C.26) by . From this follows the control equilibrium sj in equation (C.49) for the gain saturated case: sj = 1−∝ × sj + (C.49) What results to a fixed equilibrium packet rate sj independent of the channel utilization: sj = ∝. (C.50) From equation (C.50) follows with equation (C.25) a linear relation between × and the equilibrium channel load as shown by equation (C.51) for cannel loads within the lower gain saturation area (see Figure C.12): = ∝ × × (C.51) The gain saturation areas result from the gain saturation values and that are replacing the second term of equation (C.26). Based on this the gain saturation areas are given by an upper area with a lower bound − as shown in equation (C.52) and a lower area with an upper bound + as shown in equation (C.53): = − (C.52) = − (C.53) Figure C.12 shows these areas and the impact of the gain saturation on the equilibrium channel load for the parameter values given in Table 3 of ETSI TS 102 687 [i.14]. It can be seen that in the gain saturation area below + the equilibrium channel load follows the linear relation given by equation (C.51), while above + it follows the non-linear equation (C.28). The upper gain saturation area is only reached in dynamic channel load scenarios. It has no influence on the equilibrium channel load. Figure C.12: Detail of equilibrium channel load of adaptive congestion control for the parameter values given in Table 3 of ETSI TS 102 687 [i.14] showing the effect of the gain saturation The filtering as proposed in ETSI TS 102 687 [i.14] increases the stability range for × by a factor of two. Hence, the control loop can converge to an equilibrium for twice as many nodes or for doubled message sizes. Details about this need to be investigated. 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 0 10 20 30 40 50 channel load CL number of nodes N 0,25 0,5 1 1,5 2 Ton / ms Upper gain saturation area Lower gain saturation area − ETSI ETSI TR 104 073 V2.2.1 (2026-02) 79 C.2.2.3 Convergence speed of the discrete time adaptive congestion control In [i.33] a series expansion approach was used to find the channel load values after each time step of the discrete time adaptive congestion control. From there, with = × × × and = ∝+ × × the channel load after n time steps can be calculated according to equation (C.54) for the discrete time adaptive congestion control under the assumption that all nodes are synchronized. The more realistic case for unsynchronized nodes is discussed in Annex D: × = ×× (C.54) From equation (C.54) the convergence speed of the discrete time adaptive congestion control can be calculated. Where the convergence speed is defined by the time where the channel load crosses the boundaries of ±5 % deviation from the equilibrium value and afterwards stays within these boundaries when applying a channel load step from 0 % to 100 % or from 100 % to 0 %. Since the convergence times depend on the initial conditions at the time t = 0 s (see Figure C.11) these conditions of the network nodes need to be defined. An initial transmission rate = 0 and a resulting 0 = 0 can be used to test an implementation of the control algorithm, while the example with 0 = 1 is more to demonstrate what happens when the channel is fully loaded and = ×, what is a bit of an arbitrary assumption just to show the difference to = 0. For 1 − > 0 the power function 1 − is monotonically increasing (see examples in Figure C.13), for 1 − < 0 it oscillates between positive results for even time steps values n and negative results for odd values of n. Since the factor 0 × − is negative for 0 = 0 this leads to an overshoot of the equilibrium (equation (C.28)) as can be seen as an example in Figure C.14. Figure C.13: Channel load of adaptive congestion control for different values of N x Ton resulting from equation (C.54) for the parameter values given in Table 3 of ETSI TS 102 687 [i.14] 0.8 0.6 0.4 0.2 0.0 Channel load CLa 50 40 30 20 10 0 time steps n N x Ton = 10 ms N x Ton = 100 ms N x Ton = 500 ms ETSI ETSI TR 104 073 V2.2.1 (2026-02) 80 Figure C.14: Channel load of adaptive congestion control for N x Ton = 1 200 ms resulting from equation (C.54) for the parameter values given in Table 3 of ETSI TS 102 687 [i.14] Thus, for < 1 the number of steps to reach convergence results from equation (C.54) when setting to 95 % or 105 % of the equilibrium (equation (C.28)). Which value to take depends on the initial channel load 0 as shown in equation (C.55) and equation (C.56): For < 1 and 0 = 0 follows × 0,95 = × (C.55) For < 1 and 0 = 1 follows × 1,05 = × (C.56) The time steps to reach 95 % of the channel load equilibrium for an initial channel load of 0 result from equation (C.57). For an initial message rate = × the time steps to reach 105 % of the channel load equilibrium result from equation (C.58). Figure C.15 compares the results for and from equation (C.57) and equation (C.58) for the initial cannel load 0 equal to 0 or respectively equal to 1: For < 1 and 0 = 0 follows = , = , ∝ ×× (C.57) For < 1 and 0 = 1 follows = × , = ! ××× × , ∝ ××× " ∝ ×× (C.58) 1.0 0.8 0.6 0.4 0.2 0.0 Channel load CLa 10 9 8 7 6 5 4 3 2 1 0 time steps n N x Ton = 1200 ms ETSI ETSI TR 104 073 V2.2.1 (2026-02) 81 Figure C.15: Number of time steps to reach ±5 % deviation from the equilibrium channel load for the adaptive congestion control with parameter values as given in Table 3 of ETSI TS 102 687 [i.14] For > 1 the channel load will overshoot the equilibrium as has been shown in Figure C.14. This overshoot can overload the channel, and the convergence time consists of the channel overload recovery time plus the exponential channel load decay time. The linear channel load model will not hold for the channel load saturation at high loads. But it can give at least a rough estimation of the decay time as long as that parameter is below a channel load of 0,8. The envelope of the channel load decay is given by equation (C.59): For > 1 follows # $ = − 0 − × × (C.59) With a gain saturation the convergence time increases, since below the channel load follows equation (C.60) which is less steep in n compared to equation (C.54): For < follows × = %×&×% ××' ×× % (C.60) Depending on whether the channel load equilibrium is below or above , the convergence speed can either be determined from equation (C.60) or by piecewise use of equation (C.60) up to and from there by use of equation (C.54) as shown in the beginning of the present clause. ETSI TS 102 687 [i.14] foresees despite the gain saturation that was described already in clause C.2.2.2 a filtering based on the last three channel load measurements. When × is constant over time this filtering is given by equation (C.61) and has an impact on the convergence behaviour: ( = 1−∝ × ( − + × × × − ) * + *×) +×) , (C.61) Figure C.16 and Figure C.17 show based on some examples the impact of the gain saturation and the input filtering as specified in ETSI TS 102 687 [i.14] with the parameter values given in Table 3 of ETSI TS 102 687 [i.14]. The given input filtering does not improve the stability it even increases the overshoots for × > 826,7 as shown in Figure C.17 and can even lead to an overshoot for × < 826,7 as shown in Figure C.16. An analytic evaluation of this behaviour is too complex to be shown in the present document. 1 10 100 1000 1 10 100 1000 nc N x Ton / ms initial channel load = 0 initial channel load = 1 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 82 Detailed evaluation of the gain compression independently from the filtering show that the gain compression effectively reduces the overshoot while only slightly prolonging the convergence time. This evaluation also shows that an output filtering can improve the stability, since it does not add additional delay in the control loop. But such an output filtering also increases the convergence time and therefore the right balance between stability and convergence time needs to be found. Figure C.16: Comparison of the difference between the channel load given by equation (C.54) for a basic adaptive DCC and the channel load of a DCC including filtering and gain saturation Figure C.17: Comparison of the difference between the channel load given by equation (C.54) for a basic adaptive DCC and the channel load of a DCC including filtering and gain saturation ETSI ETSI TR 104 073 V2.2.1 (2026-02) 83 C.2.2.4 Time behaviour of a discrete time adaptive congestion control implementation For the evaluation in the previous clauses of Annex C, it was assumed that the channel load can be determined and influenced according to equation (C.2) instantaneously. Both does not hold for a DCC that controls the packet rate. A change of the channel utilization is done by adjusting the time Toff in between two transmissions according to the packet duration Ton that is given by the MCS and the message size. This means that changing the channel utilization needs at least a time of Ton+Toff to take effect, what follows from equation (B.6). The same applies to the channel load measurement, the correct channel load can only be determined when the measurement time is long enough to account for all network nodes in range. Annex D will show the impact of the channel load measurement on the control behaviour. In this clause the time scaling of the convergence process is evaluated. Clause C.2.2.3 only dealt with time steps to simplify the evaluation of the discrete time adaptive congestion control. To get the convergence function over time of a real implementation, the minimum time step duration Δtj(k) for node j and step k can be assumed to be Ton j(k) +Toff j(k). The adaptive congestion control algorithm determines the packet rate rj for the next time step of node j with equation (C.23). The minimum time step duration Δtj is the reciprocal of this packet rate rj as shown in equation (C.62): ∆ = = (C.62) The time tnj after n time steps of node j is then the sum of the time step durations Δtkj, where k is the time step index. Δt0j is the initial time step duration resulting from CLa(0). This value can be infinite when CLa(0)=0 and is therefore not included in the sum: = ∑ ∆ (C.63) When assuming that all nodes j behave similar, equation (C.63)can be rewritten with equation (C.54) and equation (C.25) to equation (C.64). Where A and B are given in clause C.2.2.3. = × × ∑ × × (C.64) Assuming that Ton is constant over time and equal for all nodes, then equation (C.64) can be rewritten to equation (C.65) when using the substitution = × and = 1 −. = ×× × × ∑ (C.65) The sum over has no simple analytic solution, but it can be converted into an integral that results for B<1 in equation (C.66): = + − ( ) () (C.66) The offset coefficient koffs in equation (C.70) can be found after some calculation when subtracting the first two terms of the sum in equation (C.65) from each other (equation (C.67) and equation (C.68)): = + − ( ) () (C.67) + = + − ( ) () (C.68) = (× × () ) ( × () ) ×()() () (C.69) The offset coefficient toffs then results from equation (C.67): = ()×(( )× )( )×( ) (× ) ×() (C.70) ETSI ETSI TR 104 073 V2.2.1 (2026-02) 84 With the offset coefficients from equation (C.69) and equation (C.70) the time tn corresponding to a certain time step number n can be calculated for B<1 according to equation (C.71): = × × × × + − × ×( ) (C.71) For B<1 equation (C.54) and equation (C.71) can be used to calculate a continuous time solution of the channel load Cla(t) that is to be expected when the network nodes are not synchronized in time. Figure C.18 shows this continuous time solution for the same × values and the same DCC configuration that were used for the results shown as dashed lines in Figure C.16 for the parameter values from Table 3 of ETSI TS 102 687 [i.14]. Figure C.19 shows for the parameter values given in Table 3 of ETSI TS 102 687 [i.14] the convergence time calculated with equation (C.71) from the number of time steps obtained in clause C.2.2.3 (see Figure C.15) to reach 95 % of the channel load equilibrium for an initial channel load of Cla = 0 and an initial packet rate of ( = 0) = 0. Figure C.18: Channel load as function of time for a basic adaptive rate controller with an initial packet rate of (! = ) = 0.8 0.6 0.4 0.2 0.0 Channel load CLa 14x10 3 12 10 8 6 4 2 0 time / ms No filtering and gain saturation N x Ton = 500 ms N x Ton = 100 ms N x Ton = 10 ms ETSI ETSI TR 104 073 V2.2.1 (2026-02) 85 Figure C.19: Time tc to reach 95 % of the equilibrium channel load when starting from an empty channel for the adaptive congestion control C.3 Power control Adapting the output power level to control the channel congestion is used in centrally managed radio networks since the central controller has the knowledge of the power settings of transmitters that are out of radio range of each other. Because a power control has no direct impact on the local channel load, the local channel load cannot be used without modifications as feedback in a decentralized power control loop. There are two possibilities to use local data as feedback for such a control loop. Either the channel load measurement threshold is modified according to the transmit power level as described in [i.36], or since this is not possible with most of the chipsets, the number of DSRC stations within a transmit power level dependent radius is counted and used together with the channel occupancy time of these transmitters as feedback metric. Even with a modified local feedback value, the power control cannot control the local channel load independently from the vehicle density to an optimum value, since it estimates the remote vehicle density based on the local one. Hence, the most reasonable use of a power control is to avoid channel overload at distant ITS stations. This implies that the channel load measured by these remote transmitters should be known, since a "power control" based on local channel load information lacks negative feedback necessary to "control" the power level. This is because when reducing the transmit power level, a transmitter gets invisible to other transmitters far away. In turn these transmitters will increase their transmit power level since the channel load at this position decreases. This increase in transmit power level has no impact on the channel load in the vicinity of the transmitting station, but it causes even more channel load at positions far away. A decentralized power control therefore needs a management channel that has a longer range than the radio channel used for message dissemination. This could be done by multi hop dissemination of the power settings with a robust MCS. A power control has further difficulties. The vehicle density can vary by more than factor 10, implying that a power range of more than 20 dB is necessary to cope with these variations by keeping always the same number of vehicles within range. Different transmit power levels result in different ranges causing a lot of stations to be hidden, and the CSMA/CA algorithm to fail. Therefore, the packet collision rate statistics with power control will get worse and exhibit a more ALOHA like behaviour. Therefore, a power control should only be used to support a rate control. 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 1 10 100 1000 tc / ms N x Ton / ms initial channel load = 0 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 86 Adapting the output power level only based on the local channel load can lead to spatial oscillations. As an example Figure C.20 and Figure C.21 show the simulation result for a low traffic density scenario in two lanes with 35 m vehicle separation after 25 iterations (steady state) of the algorithm specified in SAE J2945/1 [i.39]. For this simulation a channel load measurement threshold of -85 dBm was chosen and a simple free space channel model was used. The packets of 0,8 ms duration were generated with a frequency of 10 Hz to show the behaviour of the power control algorithm only. The result for this scenario show in Figure C.20 and Figure C.21 oscillations of the transmit power level and the channel load as function of the vehicle position. This spatial oscillation is caused by the missing negative feedback when using the local channel load as feedback value in the control loop. Similar problems have been observed in [i.37] when simulating the ETSI joint power and rate congestion control specified in ETSI TS 102 687 [i.14]. Figure C.20: Power level oscillations in space for light street traffic Figure C.21: Channel load oscillations in space for light street traffic 20 18 16 14 12 10 Tx power level [dBm] -1500 -1000 -500 0 500 1000 1500 Position [m] 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Channel Load -1500 -1000 -500 0 500 1000 1500 Position [m] ETSI ETSI TR 104 073 V2.2.1 (2026-02) 87 Annex D: Channel load measurement D.1 Channel load measurement overview In Annex C it was assumed that the channel load can be measured in any point of time. In practice it is measured by determining the total channel active time " and the total channel idle time ## " over a certain measurement period and then by calculating the measured channel load $ according to equation (D.1). This measured channel load is often called channel busy ratio : $ = = (D.1) When the measurement period is shorter than the idle time ## between two consecutive messages transmitted by a certain node, the channel utilization of this node will be considered only in some of the channel load measurements. Hence, the measurement result will not be stable even when all nodes are not changing their channel utilization. Furthermore, such a sliding window approach to measure the channel load does exhibit a linear phase over frequency and has no flat frequency response. This has an influence on the control loop stability since some frequencies will be amplified in the control loop, leading to instabilities. Some authors [i.38] even proposed to inject noise into the measurement results to avoid such a feedback effect in the control loop. In addition, for an ITS-G5 access layer the measurement times are not synchronized what usually improves stability but complicates an analytic approach to determine stability. Even the unsynchronized measurement times will improve stability, they should not be taken for granted, since some implementation might synchronize them with the GPS time. Therefore, the stability determination should be done for the worst case of synchronized nodes to guaranty stability for any random configuration. Such an investigation has been done in Annex C. In Annex D a simulation with 1 ms time slots and a random start configuration was used, where the channel was first empty and the initial ## value of each node has been determined by the first step of the adaptive algorithm plus a random value between -1 ms and +1 ms. The first transmission time of each node has been determined from the initial channel load resulting from the initial ## value by a random experiment. Starting with the first time slot and the first node a random number between 0 and 1 is drown, if this number is smaller than the initial channel load then the time slot is used for transmission of the actual node and the next node starts the same random experiment at the consecutive time slot, else the random experiment is repeated for the next time slot with the same node. This process runs until all nodes found a random transmission time slot. For the simulation an idealized MAC is used that avoids any packet collisions, hence when a time slot is already in use, the MAC looks for the next free slot and puts the transmission there. This means that each transmission increases the channel load. The amount of this increase is only given by the ## value. In contrast the ITS-G5 MAC allows for simultaneous transmissions resulting in a packet loss by self-interference and roughly no channel load increase by simultaneous transmissions. This causes a channel load saturation effect not investigated in Annex C, therefore the idealized MAC simplification was chosen to allow a comparison of the results from Annex C and Annex D. D.2 Influence of the channel load measurement on the reactive congestion control First simulation results of the table based reactive algorithm show chaotic behaviour, even under static traffic and radio channel conditions. Such a chaotic behaviour is typical for nonlinear differential equations as given by the step function resulting from the table entries. Figure D.1 shows such a behaviour for a reactive DCC according to Table A.1 in ETSI TS 102 687 [i.14] for 401 nodes, Ton = 1 ms, and 100 ms CBR measurement time. The blue bars at the bottom of Figure D.1 show the transmission events. Even the scale is such that not all channel idle periods can be seen in this figure, it is obvious that the white spaces in between the transmissions are not deterministic. The coloured line shows in blue an underutilization and in red an over utilization of the channel. In average it fluctuates chaotically around the expected 60 % channel load. The Average is calculated by a sliding window of 1 second length. The CBR result obtained with a 100 ms sliding window duration is shown in black, it jumps chaotically between different values in the range from 0 to 100 %. This fluctuating value is the input to the reactive DCC. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 88 To visualize these fluctuations for different number of nodes and thereby different ## values the shaded region in Figure D.2 shows the range of channel load values measured with a sliding window duration of 1 second that were calculated for a time span of 20 seconds after bounded stability was reached. The green line shows the average of these channel load values. When comparing this to Figure C.9 the theory only fits in the far-left part of Figure D.2. This is because the measurement time span of 100 ms is much shorter than the ## time of each node necessary to limit the channel load when there are more than 40 nodes in range. To reduce the chaotic behaviour of the measured channel load, the measurement period can be increase as specified in The C2C-CC Profile, where it was doubled to 200 ms. Figure D.3 shows the result for this doubled measurement period. As expected, the result is close to theory for around double the number of nodes compared to a measurement period of 100 ms. Still, for more than 80 nodes the behaviour gets again chaotic. Consequently, the CBR measurement duration should be long enough to account for the transmissions of all nodes to avoid erratic behaviour of the control loop. As outline in the beginning of the present clause and in clause C.2.2.4, the easiest and most effective way of doing this is to make the measurement duration equal to + ## . Since in practice all nodes use different values of and ## this will automatically desynchronize the measurement times, and even when all times are equal, each node will measure a (slightly) different CBR, resulting in (slightly) different ## time for each node. Figure D.4 shows that this leads for the reactive DCC for up to 300 nodes to theoretical result shown in Figure C.9. For more nodes the MAC of ITS-G5 starts to avoid simultaneous transmissions what makes the channel access less predictive and the channel load more erratic. Figure D.1: Time behaviour for a reactive DCC according to Table A.1 in ETSI TS 102 687 [i.14] for 401 nodes, Ton = 1 ms, and 100 ms CBR measurement time ETSI ETSI TR 104 073 V2.2.1 (2026-02) 89 Figure D.2: Bounded stability for a reactive DCC according to Table A.1 in ETSI TS 102 687 [i.14] for Ton = 1 ms and 100 ms CBR measurement time Figure D.3: Bounded stability for a reactive DCC according to C2C-CC [i.34] for Ton = 1 ms and 200 ms CBR measurement time 80 70 60 50 40 30 20 10 CBR / % 400 360 320 280 240 200 160 120 80 40 0 Number of Nodes n 80 70 60 50 40 30 20 10 CBR / % 400 360 320 280 240 200 160 120 80 40 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 90 Figure D.4: Bounded stability for a reactive DCC according to C2C-CC [i.34] for Ton = 1 ms and a CBR measurement time equal to Ton+Toff D.3 Influence of the channel load measurement on the adaptive congestion control D.3.1 CBR measurement with a fixed 100 ms measurement time interval When using a plain adaptive congestion control with a fixed CBR measurement time of 100 ms without gain saturation and without a Toff limitation the CBR averaged over 1 second stays after 30 seconds bounded within a 10 % margin for up to around 500 network nodes and a Ton of 1 ms as can be seen in Figure D.5. Figure D.5: Bounded stability for the adaptive DCC with α and β taken from Table 3 in ETSI TS 102 687 [i.14] for Ton = 1 ms and 100 ms CBR measurement time 80 70 60 50 40 30 20 10 CBR / % 400 360 320 280 240 200 160 120 80 40 0 Number of Nodes n 80 70 60 50 40 30 20 10 CBR / % 800 750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 91 Figure D.6 shows as an example the convergence of the CBR over time for 100 network nodes and a Ton of 1 ms. The black line shows the CBR measured by the network nodes and the coloured line shows the CBR average over 1 second. The coloured bars at the bottom show the transmission events of the network nodes. They are randomly distributed as expected. Figure D.6: CBR convergence for the adaptive DCC with α and β taken from Table 3 in ETSI TS 102 687 [i.14] for 100 nodes, Ton = 1 ms, and 100 ms CBR measurement time Figure D.7 shows the Toff variations between 100 different network nodes for Ton = 1 ms. The light grey shaded region marks the ± sigma range, the dark grey region shows the minimum to maximum range of Toff. For all 100 nodes Toff converges within 30 seconds to the same equilibrium. Figure D.7: Toff convergence for the adaptive DCC with α and β taken from Table 3 in ETSI TS 102 687 [i.14] for 100 nodes, Ton = 1 ms and 100 ms CBR measurement time Figure D.8 shows what happens for a fixed 100 ms CBR measurement time when × ## is increased to 600 ms. The measured CBR shown as black line is oscillating. Also, the 1 second average shows a strong stable oscillation with CBR values between 45 % and 80 % after 30 seconds as also shown in Figure D.5. The coloured bars at the bottom of Figure D.8 show the transmission events of the network nodes. They are randomly distributed, what means that the oscillations are generated by the whole assemble of the network nodes. 60 40 20 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms N = 100 1600 1400 1200 1000 800 600 400 200 0 Toff variations / ms 30x10 3 25 20 15 10 5 Time / ms N = 100 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 92 Figure D.8: CBR convergence for the adaptive DCC with α and β taken from Table 3 in ETSI TS 102 687 [i.14] for 600 nodes, Ton = 1 ms, and 100 ms CBR measurement time Figure D.9 shows the Toff variations between the network nodes for node numbers up to 800 and Ton = 1 ms evaluated between 20 seconds and 30 seconds after the CBR step function. The light grey shaded region marks the ± sigma range, the dark grey region shows the minimum to maximum range of Toff. For more than 100 nodes the Toff values of the individual nodes start to diverge, even the average CBR shown in Figure D.5 converges. This is because the time constant for the Toff convergence is longer than the 30 second observation time. For around more than 200 nodes a Toff runoff of some individual nodes can be observed. A few nodes were increasing the Toff value to large numbers, so that they were far off the 100 ms measurement time interval and could not be recognized by the other network nodes. This effect gets even worse when the average CBR starts to oscillate. Figure D.9: Toff variations for the adaptive DCC with α and β taken from Table 3 in ETSI TS 102 687 [i.14] for Ton = 1 ms and 100 ms CBR measurement time 100 80 60 40 20 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms N = 600 1600 1400 1200 1000 800 600 400 200 0 Toff variations / ms 800 700 600 500 400 300 200 100 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 93 D.3.2 CBR measurement with a fixed 100 ms measurement time interval and a limited Toff range Figure D.9 shows for the adaptive DCC with α and β taken from Table 3 in ETSI TS 102 687 [i.14] that the Toff values of some individual network nodes can get out of control. This can be avoided when pinning Toff to a maximum value when the adaptive algorithm tries to exceed a certain Toff range. A reasonable value for this upper Toff limit Toff max is 1 second. Frequency regulation assumes that the duty cycle of ITS transmissions does not exceed 1 %, this requirement gives a lower limit for Toff. For Ton = 1 ms this limit Toff min results to 99 ms. In Figure D.10 can be seen for the same α and β that this Toff limitation avoids excessive oscillations of the 1 second average of the CBR as was shown in Figure D.5. Nevertheless, when looking at the CBR convergence for 400 nodes in Figure D.11 it can be seen that the measured CBR over 100 ms shown as black line exhibits a strong stable oscillation up to 100 % CBR. Figure D.12 shows the Toff variations between 400 different network nodes for Ton = 1 ms. The light green shaded region marks the ± sigma range, the dark green region shows the minimum to maximum range of Toff. It is obvious, that the Toff values do not converge at all. This can also be seen in Figure D.13 where for high node numbers the Toff values are even within the full range between Toff min and Toff max. Figure D.13 shows the Toff variations between the network nodes for node numbers up to 800 and Ton = 1 ms evaluated between 20 seconds and 30 seconds after the CBR step function. The light green shaded region marks the ± sigma range, the dark green region shows the minimum to maximum range of Toff. For more than 100 nodes the Toff values of the individual nodes start to diverge. This is because the time constant for the Toff convergence is longer than the 30 second observation time. For around more than 200 nodes a Toff runoff of some individual nodes can be observed. A few nodes were increasing the Toff value to the upper limit Toff max, so that they were far off the 100 ms measurement time interval and could not be recognized by the other network nodes. This effect gets even worse when the average CBR starts to oscillate. Figure D.10: Bounded stability for the adaptive DCC for a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms and 100 ms CBR measurement time 80 70 60 50 40 30 20 10 CBR / % 800 750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 94 Figure D.11: CBR convergence for the adaptive DCC for 400 nodes, a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, and 100 ms CBR measurement time Figure D.12: Toff convergence for the adaptive DCC for 400 nodes, a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms and 100 ms CBR measurement time 100 80 60 40 20 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms N = 400 1600 1400 1200 1000 800 600 400 200 0 Toff variations / ms 30x10 3 25 20 15 10 5 Time / ms N = 400 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 95 Figure D.13: Toff variations for the adaptive DCC for a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms and 100 ms CBR measurement time D.3.3 CBR measurement and adaptive algorithm as specified in ETSI TS 102 687 ETSI TS 102 687 [i.14] specifies the CBRITS-S to be taken as input for the adaptive algorithm according to equation (D.2): %& &( ) = × + × −100 + ( −200 ) (D.2) Where t is the current time and is the CBR measured over the time from t - 100 ms to t. ETSI TS 102 687 [i.14] specifies in Table 3 the DCC algorithm parameters α, β, δmax, δmin, $' , $' , and in clause 5.4 the filter function used in this clause to evaluate the simulation results. In addition, the Toff limits Toff max = 1 s and Toff min = 99 ms were used in the simulation with Ton = 1 ms. Figure D.14 shows no significant improvement compared to Figure D.10. The improvement by averaging 3 consecutive CBR measurements can be seen when comparing Figure D.15 with Figure D.13. 1000 800 600 400 200 0 Toff variations / ms 800 700 600 500 400 300 200 100 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 96 Figure D.14: Bounded stability for the adaptive DCC with a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, and 100 ms CBR measurement time Figure D.15 shows the Toff variations between the network nodes for node numbers up to 800 and Ton = 1 ms evaluated between 20 seconds and 30 seconds after the CBR step function. The light blue shaded region marks the ± sigma range, the dark blue region shows the minimum to maximum range of Toff. For more than 200 nodes the Toff values of the individual nodes start to diverge. This is because the time constant for the Toff convergence is longer than the 30 second observation time. For around more than 300 nodes a Toff runoff of some individual nodes can be observed. A few nodes were increasing the Toff value to the upper limit Toff max, so that they were off the measurement time interval of 3 times 100 ms and could not be recognized by the other network nodes. This effect gets even worse when the average CBR starts to oscillate for configurations with more than around 400 network nodes. Figure D.15: Toff variations for the adaptive DCC with a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, and 100 ms CBR measurement time Figure D.16 shows the Toff variations between 300 different network nodes for Ton = 1 ms. The light blue shaded region marks the ± sigma range, the dark blue region shows the minimum to maximum range of Toff. The Toff value slowly converge to a common equilibrium while for 400 network nodes as shown in Figure D.17 no such convergence is observable within 30 seconds. 80 70 60 50 40 30 20 10 CBR / % 800 750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 0 Number of Nodes n 1000 800 600 400 200 0 Toff variations / ms 800 700 600 500 400 300 200 100 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 97 Figure D.16: Toff convergence for the adaptive DCC for 300 nodes, a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, and 100 ms CBR measurement time Figure D.17: Toff convergence for the adaptive DCC for 400 nodes, a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, and 100 ms CBR measurement time D.3.4 Dynamic CBR measurement time and adaptive algorithm as specified in ETSI TS 102 687 As outlined in clause D.1 and D.2 the CBR measurement time should be + ## . The results presented in clause D.3.4 were obtained by averaging the CBR over + ## instead of the 100 ms measurement interval with the filtering given in ETSI TS 102 687 [i.14]. The parameters α, β, δmax and δmin were taken from table 3 in ETSI TS 102 687 [i.14], $' , and $' were not used. The Toff limits were Toff max = 1 s and Toff min = 99 ms, and Ton was set to 1 ms. Figure D.18 shows only small CBR variations in the time window of 10 seconds length starting 20 seconds after the channel load step for all node numbers between 0 and 800. Figure D.19 confirms that ## for all nodes is bounded in a defined range given by the convergence time. 1600 1400 1200 1000 800 600 400 200 0 Toff variations / ms 30x10 3 25 20 15 10 5 Time / ms N = 300 1600 1400 1200 1000 800 600 400 200 0 Toff variations / ms 30x10 3 25 20 15 10 5 Time / ms N = 400 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 98 Figure D.18: Bounded stability for the adaptive DCC with a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, and a CBR measurement time of Ton + Toff Figure D.19: Toff variations for the adaptive DCC with a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, and a CBR measurement time of Ton + Toff Figure D.20 shows as an example that for 550 network nodes the CBR is converging to an equilibrium value. Figure D.21 shows that the convergence time of the Toff values for 550 network nodes exceeds 30 seconds. 80 70 60 50 40 30 20 10 CBR / % 800 750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 0 Number of Nodes n 1000 800 600 400 200 0 Toff variations / ms 800 700 600 500 400 300 200 100 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 99 Figure D.20: CBR convergence for the adaptive DCC for 550 nodes, a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, and a CBR measurement time of Ton + Toff Figure D.21: Toff convergence for the adaptive DCC for 550 nodes, a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, and a CBR measurement time of Ton + Toff D.3.5 Dynamic CBR measurement with enhanced filtering and adaptive algorithm as specified in ETSI TS 102 687 The results can be further improved when using a weighted average to determine the CBRITS-S over a time span of Ton + Toff. Equation (D.3) shows such a weighted average, where # is a function that is 1 when the channel is active and 0 when it is free. %& &( ) = 0,95 × + 0,1 × × (D.3) When using the weighted average filter for the same configuration as given in clause D.3.4 the results can be slightly improved compared to the ones given in clause D.3.4 as shown in Figure D.22, Figure D.23, Figure D.24, and Figure D.25. 80 60 40 20 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms N = 550 1600 1400 1200 1000 800 600 400 200 0 Toff variations / ms 30x10 3 25 20 15 10 5 Time / ms N = 550 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 100 Figure D.22: Bounded stability for the adaptive DCC for a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, a CBR measurement time of Ton + Toff, and enhanced filtering Figure D.23: Toff variations for the adaptive DCC with a 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, a CBR measurement time of Ton + Toff, and enhanced filtering 80 70 60 50 40 30 20 10 CBR / % 800 750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 0 Number of Nodes n 1000 800 600 400 200 0 Toff variations / ms 800 700 600 500 400 300 200 100 0 Number of Nodes n ETSI ETSI TR 104 073 V2.2.1 (2026-02) 101 Figure D.24: CBR convergence for the adaptive DCC for 550 nodes, 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, a CBR measurement time of Ton + Toff, and enhanced filtering Figure D.25: Toff convergence for the adaptive DCC for 550 nodes, 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms, no gain saturation, a CBR measurement time of Ton + Toff, and enhanced filtering D.4 Influence of the MAC on the adaptive congestion control In the previous clauses an ideal MAC was assumed that avoids packet collisions by always looking for an empty time slot to transmit. The MAC of ITS-G5 choses for each transmission a random number between 0 and 15 and counts it down as long as a time slot (contention window) is occupied on the radio channel. When this counter reaches zero or the time slot is empty, the packet is transmitted. This behaviour can lead to packet collisions, when two or more ITS-S start to transmit at the same time. These transmissions interfere with each other, and only stations close to the transmitter can receive them correctly. For random uncorrelated transmissions and low channel load the ITS-G5 MAC can efficiently avoid such collisions. Therefore, a working DCC limits the channel load and keeps the transmissions random. In the previous clauses it was shown that the transmissions can synchronize with each other, leading to an oscillating channel load. Hence, even the average of the channel load is low, short term channel congestion can happen. 80 60 40 20 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms N = 550 1600 1400 1200 1000 800 600 400 200 0 Toff variations / ms 30x10 3 25 20 15 10 5 Time / ms N = 550 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 102 Figure D.26 shows as black line the short term CBR taken as DCC input for the DCC parameters specified in ETSI TS 102 687 [i.14] for 600 ITS-S. While the short term CBR is oscillating between 45 % and 90 %, the average channel load over one second shown as coloured line is quite stable between 62 % and 69 % after a steady state is reached. Still a lot of packet collisions occur, shown as coloured bars in Figure D.26. The hight of the bars represent the number of simultaneous transmissions. In contrast, Figure D.27 shows less packet collisions for the same DCC parameters but a dynamic CBR measurement with enhanced filtering as described in clause D.3.5 with an ITS-G5 MAC. To reduce the packet collision rate further, either the equilibrium channel load could be lowered or the contention window counter increased. Figure D.28 shows the result for the same DCC parameters that were used to generate Figure D.27 for a MAC that is using a contention window counter range of 0 to 127 compared to 0 to 15 used for Figure D.27. This increase of the number of contention windows has no significant impact on the convergence, but it reduces the number of packet collisions drastically as shown in Figure D.29. There the difference between the packet collision ratio for an adaptive DCC with dynamic CBR measurement and enhanced filtering as described in clause D.3.5 and the DCC as specified in ETSI TS 102 687 [i.14] is shown. While the ETSI adaptive DCC shows a strong increase of the collision ratio to more than 50 %, the results for the adaptive DCC with dynamic CBR measurement stay around a collision ratio of 10 % for a contention window counter value from 0 to 15. When allowing a maximum number of 128 contention window retries, the packet collision ratio stays below 1,3 % for up to 700 ITS-S. The drawback of increasing the contention window counter is that this increases the average transmission delay. Further investigations are necessary to find the right balance between transmission delay and packet collision ratio. Figure D.26: CBR convergence for the adaptive DCC for 600 nodes, 1 % duty cycle limit, Toff max = 1 s, Ton = 1 ms and the number of simultaneous transmissions for each time step for an ITS-G5 MAC 100 90 80 70 60 50 40 30 20 10 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms 20 18 16 14 12 10 8 6 4 2 0 Simultaneous transmissions N = 600 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 103 Figure D.27: CBR and number of simultaneous transmissions for the same setup as in Figure D.26 with additional gain saturation, enhanced filtering, and a CBR measurement time of Ton + Toff, Figure D.28: CBR and number of simultaneous transmissions for the same setup as in Figure D.27 for a modified MAC using a CW counter from 0 to 127 100 90 80 70 60 50 40 30 20 10 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms 20 18 16 14 12 10 8 6 4 2 0 Simultaneous transmissions N = 600 100 90 80 70 60 50 40 30 20 10 0 CBR / % 30x10 3 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Time / ms 20 18 16 14 12 10 8 6 4 2 0 Simultaneous transmissions N = 600 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 104 Figure D.29: Packet collision ratio for different DCC and MAC configurations 60 50 40 30 20 10 0 800 700 600 500 400 300 200 100 0 Number of Nodes n 60 50 40 30 20 10 0 Packet collision ratio / % Adaptive DCC with parameters as given in TS 102 687 Adaptive DCC with dynamic CBR measurement and enhanced filtering: CW counter = 0 ... 15 CW counter = 0 ... 127 ETSI ETSI TR 104 073 V2.2.1 (2026-02) 105 Annex E: A real-time evaluation of basic resource management solution E.1 Evaluation set-up The basic solution described in clause 7.5 is evaluated. The evaluation was performed using a virtualized testing environment to mimic real-world C-ITS operations. The RM module was integrated into the Vanetza C-ITS protocol stack at the FL with only minimal modifications - leveraging an extended version of the Socktapp tool - to support the additional congestion control functionalities at the FL. To enable concurrent testing, the entire stack was encapsulated in a Docker container image, allowing multiple instances to run on a single hardware platform. Each container represented a distinct C-ITS station with five active Message services, connected via a Docker virtual network. A scaling factor was applied to emulate low, medium, and high channel load scenarios by adjusting the effective channel load, evaluated through the Channel Busy Ratio (CBR). This approach also took into account realistic processing delays and hardware limitations, ensuring that the evaluation reflected operational conditions. E.2 Results The experimental results confirmed that the basic solution achieves convergence and effective congestion control under varying channel loads. Under low channel load conditions, the system converged to a resource usage value (δ) of approximately 2,74 % (Figure E.1(a)). In contrast, for medium and high channel load scenarios, δ stabilized at around 0,66 % and 0,45 %, respectively. The evaluation showed that the implemented RM module could reduce the CBR from a potential 94 % - if all Message services transmitted at maximum rate - to about 62 % under high load (Figure E.1(b)). Additionally, while there were slight deviations between the configured and actual transmission intervals (attributable to real-time processing delays, see Figure E.2, these discrepancies did not impact the overall stability or robustness of the congestion control mechanism. The results validate that the integration of the RM within Vanetza and its execution in a Docker-based virtualized environment effectively bridge simulation-based studies and real-world testing. (a) Available resources (delta) (b) Channel load or CBR (Channel Busy Ratio) Figure E.1: Stability and convergence of the implementation 0 50 100 150 200 250 300 Time [s] 0 0.5 1 1.5 2 2.5 3 Low channel load Medium channel load High channel load CBR ETSI ETSI TR 104 073 V2.2.1 (2026-02) 106 (a) Low channel load (b) Medium channel load Figure E.2: Time evolution of transmission interval of different Message services 50 100 150 200 250 300 Time [ms] 90 100 110 120 130 140 150 V2X service 1 V2X service 2 V2X service 3 V2X service 4 V2X service 5 All V2X services ETSI ETSI TR 104 073 V2.2.1 (2026-02) 107 Annex F: Maximum size of Facilities layer message size For transmission of FL messages over ITS-G5, Release 1 considers the maximum size of these messages. Two main reasons for message size limitation exist: • Existing specifications from lower layer standards, specifically PHY, MAC, LLC. • Radio performance considerations. From an ITS-G5 PHY perspective, the size of a PHY frame is limited by the 12-bit PHY header "Length" field, which results in a maximum PHY frame size 4 095 Bytes. From an ITS-G5 MAC perspective, the MTU corresponds to the maximum MSDU as defined in IEEE 802.11 [i.9], i.e. the payload that higher layers deliver to the IEEE 802.11 MAC layer, and equals 2 304 Bytes. This MTU is further reduced by LLC with the SNAP option, GeoNetworking and BTP headers, yielding a theoretical maximum FL message size of 2 252 Bytes (see Table 3, column non-NGV ITS-G5), which still need to be reduced by the variable size of signature and certificate (see also clause 5.5.5). In comparison to the non-NGV ITS-G5, the NGV ITS-G5 allows a larger MTU size, which results in a maximum FL-SDU size of 7 883 Bytes, minus size of signature and certificate (see column NGV ITS-G5 in Table 3). Table 3: Maximum FL message size for ITS-G5 without consideration of size for signature and certificate Ethertype-compatible ITS-G5 non-NGV ITS-G5 NGV ITS-G5 MTU (max MSDU) 1 500 B 2 304 B 7 935 B LLC (802.2 LLC + SNAP) - 8 B - 8 B - 8 B GeoNetworking Header - 88 B (maximum) - 40 B (SHB) - 40 B (SHB) BTP-Header - 4 B - 4 B - 4 B Maximum FL-SDU 1 400 B 2 252 B 7 883 B Another consideration in the message size is the convention that in IEEE 802.11 networks, IP typically sets the MTU size to 1 500 Bytes (see IETF RFC 1042 [i.48]). It is important to emphasize that the 1 500 Bytes limitation does not originate from the IEEE 802.11 standard but has historical reasons: When Wi-Fi® emerged as a wireless LAN technology, it was designed to interoperate transparently with wired Ethernet networks - using the same IP subnet, Layer-2 bridging, and protocol stack without additional configuration. To maintain this compatibility, implementors adopted Ethernet's maximum payload size of 1 500 Bytes, even though the IEEE 802.11 MAC technically allows a MSDU of up to 2 304 Bytes. This line of argumentation was used in Release 1 to effectively limit the maximum FL message in Release 1 to 1 400 Bytes, minus size of signature and certificate (see Table 3, column Ethertype-compatible ITS-G5). For Release 2, stakeholders in TC ITS should be consulted whether the argumentation for the 1 500 Bytes still applies for Release 2 implementations or whether a larger size can be permitted. For LTE-V2X and 5G NR-V2X, an equivalent analysis of the maximum FL message size is not feasible because the PC5 interface does not define a fixed MTU at layer 2. Instead, data is segmented into MAC SDUs, which are then mapped into so-called Transport Blocks (TBs). The size of a TB is determined by the selected MCS, the available bandwidth, and the allocated radio resources, rather than by any fixed limit. In defining the maximum FL message size, radio-related aspects should be considered. In practice, the transmission of large frames may suffer from synchronization and equalization issues, particularly at high vehicle speeds. Since different access-layer technologies exhibit different physical-layer capabilities, the maximum FL message size may also differ across these technologies. For example, a larger FL message size may be feasible for NGV ITS-G5 than for 11p-mode, taking into account the improved radio performance of NGV ITS-G5 - such as wider 20 MHz channels (instead of 10 MHz), the use of mid-ambles (instead of preambles only), and improved channel coding. Consequently, a maximum FL message size could be set that is lower than the theoretical values in Table 3, representing a compromise between maximizing application payload and minimizing adverse radio effects. ETSI ETSI TR 104 073 V2.2.1 (2026-02) 108 History Version Date Status V2.1.1 July 2025 Publication V2.2.1 February 2026 Publication |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 1 Scope | The present document provides a mapping between the Consumer Mobile Device Protection Profile (CMDPP) in the ETSI TS 103 732 series ([i.2], [i.3], [i.4], [i.5], [i.6] and [i.7]) security requirements and the essential cybersecurity requirements from the Annexes of the Cyber Resilience Act (CRA) [i.1]. The present document will also analyse the gaps between the CMDPP ([i.2], [i.3], [i.4], [i.5], [i.6] and [i.7]) (if any) and the CRA [i.1], considering how to address them where necessary. |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 2 References | |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 2.1 Normative references | Normative references are not applicable in the present document. |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 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] Regulation (EU) 2024/2847 of the European Parliament and of the Council of 23 October 2024 on horizontal cybersecurity requirements for products with digital elements and amending Regulations (EU) No 168/2013 and (EU) No 2019/1020 and Directive (EU) 2020/1828 (Cyber Resilience Act). [i.2] ETSI TS 103 732-1 (V2.1.2) (11-2023): "CYBER; Consumer Mobile Device; Part 1: Base Protection Profile". [i.3] ETSI TS 103 732-2 (V1.1.2) (11-2023): "CYBER; Consumer Mobile Device; Part 2: Biometric Authentication Protection Profile Module". [i.4] ETSI TS 103 932-1 (V1.1.2) (11-2023): "CYBER; Consumer Mobile Devices Base PP- Configuration; Part 1: CMD and Biometric Verification". [i.5] ETSI TS 103 732-3 (V1.1.1) (10-2023): "CYBER; Consumer Mobile Device; Part 3: Multi-user Protection Profile Module". [i.6] ETSI TS 103 732-4 (V1.1.1) (06-2024): "CYBER; Consumer Mobile Device; Part 4: Preloaded Applications Protection Profile Module". [i.7] ETSI TS 103 732-5 (V1.1.1) (07-2024): "Cyber Security (CYBER); Consumer Mobile Device; Part 5: Bootloader & Root of Trust Protection Profile Module". [i.8] Regulation (EU) 2019/881 of the European Parliament and of the Council of 17 April 2019 on ENISA (the European Union Agency for Cybersecurity) and on information and communications technology cybersecurity certification and repealing Regulation (EU) No 526/2013 (Cybersecurity Act). [i.9] Commission Implementing Regulation (EU) 2024/482 of 31 January 2024 laying down rules for the application of Regulation (EU) 2019/881 of the European Parliament and of the Council as regards the adoption of the European Common Criteria-based cybersecurity certification scheme (EUCC). ETSI ETSI TR 104 071 V1.1.1 (2025-07) 6 [i.10] ISO/IEC 15408-1:2022: "Information security, cybersecurity and privacy protection -Evaluation criteria for IT security — Part 1: Introduction and general model". [i.11] ISO/IEC 15408-2:2022: "Information security, cybersecurity and privacy protection -Evaluation criteria for IT security — Part 2: Security functional components". [i.12] ISO/IEC 15408-3:2022: "Information security, cybersecurity and privacy protection -Evaluation criteria for IT security — Part 3: Security assurance components". [i.13] ISO/IEC 15408-4:2022: "Information security, cybersecurity and privacy protection -Evaluation criteria for IT security — Part 4: Framework for the specification of evaluation methods and activities". [i.14] ISO/IEC 15408-5:2022: "Information security, cybersecurity and privacy protection -Evaluation criteria for IT security — Part 5: Pre-defined packages of security requirements". [i.15] ISO/IEC 18045:2022: "Information security, cybersecurity and privacy protection - Evaluation criteria for IT security — Methodology for IT security evaluation". [i.16] ENISA Cyber Resilience Act implementation via EUCC and its applicable technical elements (Final version: 27/01/2025). [i.17] GSMA™: "SGP.06 GSMA eUICC Security Assurance Principles v2.2". [i.18] GSMA™: "SGP.07 GSMA eUICC Security Assurance Methodology v2.2". [i.19] GSMA™: "SGP.25 eUICC for Consumer and IoT Device Protection Profile v2.1". |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 3 Definition of terms, symbols and abbreviations | |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 3.1 Terms | For the purposes of the present document, the following terms apply: embedded UICC: UICC which is not easily accessible or replaceable, is not intended to be removed or replaced in the terminal, and enables the secure changing of subscriptions preloaded application: application provided by the TOE manufacturer as part of the system software that cannot be uninstalled by the user UICC: smart card that conforms to the specifications written and maintained by the ETSI Secure Element Technologies Technical Body NOTE: UICC is neither an abbreviation nor an acronym. |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 3.2 Symbols | Void. |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 3.3 Abbreviations | For the purposes of the present document, the following abbreviations apply: ADP Application Distribution Platform CC Common Criteria CEM Common Evaluation Methodology CMD Consumer Mobile Device CMDPP Consumer Mobile Device Protection Profile CRA Cyber Resilience Act DoS Denial of Service ETSI ETSI TR 104 071 V1.1.1 (2025-07) 7 ECR Essential Cybersecurity Requirement eSA eUICC Security Assurance EUCC European Union Cybersecurity Certification GCF Global Certification Forum GDPR General Data Protection Regulation MNO Mobile Network Operator OS Operating System PP Protection Profile PTCRB PCS Type Certification Review Board RDP Remote Data Processing SAR Security Assurance Requirement SBOM Software Bill Of Material SFR Security Functional Requirement SIM Subscriber Identity Module ST Security Target TLS Transport Layer Security TOE Target Of Evaluation TSF TOE Security Function |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 4 Methodology | The present document compares each CRA essential cybersecurity requirements with the SARs and SFRs of the CMDPP documents ([i.2], [i.3], [i.4], [i.5], [i.6] and [i.7]). Two other dimensions are considered in the comparison: the SARs and SFRs defined in the Common Criteria version 2022 ([i.10], [i.11], [i.12], [i.13], [i.14] and [i.15]) than may be used instead of those in the CMDPP due to the fact that CMDPP is based on a previous CC version (i.e. Common Criteria version 3.1R5) and the content of the ENISA document Cyber Resilience Act implementation via EUCC [i.16]. ETSI TS 103 732-1 [i.2], in clause 4.5 claims conformance to CC v3.1 Release 5 and the CC and CEM addenda and it is conformed to the package EAL2 augmented with ALC_DVS_EXT.1 & ALC_FLR.3. As per EUCC [i.9] the CMDPP ([i.2], [i.3], [i.4], [i.5], [i.6] and [i.7]), which implement AVA_VAN.2 vulnerability assessment, is considered at assurance level 'substantial'. |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 5 Scope of the assessment | The CMDPP TOE is a subset of the Consumer Mobile Device (CMD) seen as product with digital elements in the context of the CRA [i.1]. Although the CMDPP TOE includes hardware, the Operating System and the preloaded apps (see clause 4.1 of ETSI TS 103 732-1 [i.2]), the radio interface of the CMD including its security functionality (UICC/SIM) related to the cellular mobile communication are not included in the TOE. The cellular mobile communication functions are out of the scope of the CMDPP. However, as suggested in [i.16], the PP owner can justify the difference between the CMDPP TOE and the CMD on the basis of the risk analysis linked to the CMDPP Security Problem Definition. If a gap still remain it has to be covered with an extension of the CMDPP or other means. The security of the mobile communication credentials is delegated to the secure element which stores them. Depending on the secure element form factor there are two scenarios: a) The secure element is a UICC. In this case the UICC is not provided by the CMD manufacturer but from the MNO chosen by the user purchasing the CMD. The UICC is therefore not part of the CMD but it is itself a different product with digital elements. The UICC manufacturer is therefore responsible of the CRA conformity of the UICC. b) The secure element is an embedded UICC (eUICC) or an integrated eUICC. In this case the secure element is a component included by the CMD manufacturer in the CMD. The eUICC and the integrated eUICC have their own Protection Profile and moreover they are products with digital elements that need to be supplied to the CMD manufacturer with proof of CRA conformity (i.e. CE mark). It is a duty of the CMD manufacturer to verify that the eUICC or the integrated eUICC is conform to CRA. ETSI ETSI TR 104 071 V1.1.1 (2025-07) 8 In both cases the secure element which stores the mobile communication credentials may be certified independently by dedicated security assessment schemes and is not introducing a gap. The other component involved in the CMD cellular mobile communication functions is the cellular modem. The modem interacts with the (e)UICC and provides the cellular mobile connectivity with the mobile network implementing the 3GPP standards. The implementation of such standard is guaranteed by the certification provided by Global Certification Forum (GCF) and PCS Type Certification Review Board (PTCRB). The functional compliance to the 3GPP standards is therefore granted. The cellular modem is however part of the CMDPP TOE because the later includes all the CMD hardware; it is therefore subject to the vulnerability management of the CMDPP TOE. Based on the above consideration it appears that the gaps in the scope of the CMDPP TOE are covered by the (e)UICC certification and the modem functional compliance provided by the GCF or PTCRB. 6 CRA Annex I Essential Cybersecurity Requirements Part I comparison This clause compares the Essential Cybersecurity Requirements set out in CRA Annex I Part I "Cybersecurity requirements relating to the properties of products with digital elements" with the CMDPP SFR/SAR. The CRA ECR in Table 1 below are provisions defined in the Cyber Resilience Act [i.1]. Table 1: Mapping of CMDPP SFRs and SARs versus CRA ECR Annex I Part I CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (1) Products with digital elements shall be designed, developed and produced in such a way that they ensure an appropriate level of cybersecurity based on the risks; - ASE_SPD.1 Security problem definition; ASE_OBJ.2 Security objectives; ASE_REQ.2 Derived security requirements. ENISA request: ASE_SPD.1 Security problem definition; ASE_OBJ.1 Security objectives; ASE_REQ.1 Direct rationale security requirements. Covered (2) On the basis of the cybersecurity risk assessment referred to in Article 13(2) and where applicable, products with digital elements shall: - - - - (a) be made available on the market without known exploitable vulnerabilities; - AVA_VAN.2 Vulnerability analysis ENISA request minimum: AVA_VAN.1 Vulnerability survey Covered ETSI ETSI TR 104 071 V1.1.1 (2025-07) 9 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (b) be made available on the market with a secure by default configuration, unless otherwise agreed between manufacturer and business user in relation to a tailor-made product with digital elements, including the possibility to reset the product to its original state; FPT_RCV.2 Automated recovery ADV_ARC.1 Security architecture description The CMD is provided with the possibility to reset it to the factory configuration. The factory configuration implements the security architecture described in ADV_ARC.1. This is not the case where the CMD is deliberately provided in a non-secure configuration. FPT_RCV.2 mandate the TOE to return to a secure state using automated procedures. Covered I ensure that vulnerabilities can be addressed through security updates, including, where applicable, through automatic security updates that are installed within an appropriate timeframe enabled as a default setting, with a clear and easy-to-use opt-out mechanism, through the notification of available updates to users, and the option to temporarily postpone them; FDP_UPF_EXT.1: Update check frequency; FMT_SMF.1/APP_ Update Specification of Management Functions FMT_SMF.1/SSW_ Update Specification of Management Functions - The FDP_UPF_EXT.1 defines requirements for the frequency the TOE checks for updates of apps, system software and actions if an update is available. This ensures that security updates are installed within an appropriate timeframe (e.g. "an interval of no more than 1 month"). FMT_SMF.1/APP_Upda te and FMT_SMF.1/SSW_Upd ate defines requirements for the TSF to specify if the software update (application or system software) is automatically installed, or the user is notified after the download, or the user is notified about the software update without the download. Several further actions are specified after the user selection. Covered (d) ensure protection from unauthorised access by appropriate control mechanisms, including but not limited to authentication, identity or access management systems, and report on possible unauthorised access; FIA_UAU.1 Timing of authentication; FIA_UAU.1 Timing of authentication; FDP_ACC.1/UserD ataAsset Subset access control; FDP_ACF.1/UserD ataAsset Security attribute based access control; FIA_AFL.1 Authentication failure handling - Authentication and access control requirements are fulfilled by the CMDPP. FIA_AFL.1 the allowed threshold of authentication attempts and the actions when the final attempt fails. Covered ETSI ETSI TR 104 071 V1.1.1 (2025-07) 10 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion I protect the confidentiality of stored, transmitted or otherwise processed data, personal or other, such as by encrypting relevant data at rest or in transit by state of the art mechanisms, and by using other technical means; Stored data confidentiality: FDP_ACF.1/UserD ataAsset Security attribute based access control; Confidentiality of communication: FTP_ITC_EXT.1/B T Inter-TSF trusted channel; FTP_ITC_EXT.1/H TTPS Inter-TSF trusted channel; FTP_ITC_EXT.1/T LS Inter-TSF trusted channel; FTP_ITC_EXT.1/W LAN Inter-TSF trusted channel; Cryptographic mechanisms: FCS_RNG_EXT.1 Random number generation; FCS_CKM.1/Asym metric Cryptographic key generation; FCS_CKM.1/Sym metric Cryptographic key generation; FCS_COP.1/SigGe n Cryptographic operation; FCS_COP.1/KeyE st Cryptographic operation; FCS_COP.1/Symm etric Cryptographic operation; FCS_COP.1/Deriv ation Cryptographic operation; FCS_COP.1/Hash Cryptographic operation; FCS_COP.1/Keyed Hash Cryptographic operation - FDP_ACF.1 defines the conditions under which the CMD decrypt the User Data Assets based on their classification; this assumes that the User Data Assets are encrypted on the CMD. The confidentiality of the communication is defined within the TOE scope for Bluetooth, HTTPS, TLS and WLAN. Other supported TSF trusted channel might be added by the manufacturer in its ST, but this is out of the scope of the present TOE. Cryptographic mechanisms are covering the state of the art. Covered ETSI ETSI TR 104 071 V1.1.1 (2025-07) 11 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (f) protect the integrity of stored, transmitted or otherwise processed data, personal or other, commands, programs and configuration against any manipulation or modification not authorised by the user, and report on corruptions; Integrity of communication: FTP_ITC_EXT.1/B T Inter-TSF trusted channel; FTP_ITC_EXT.1/H TTPS Inter-TSF trusted channel; FTP_ITC_EXT.1/T LS Inter-TSF trusted channel; FTP_ITC_EXT.1/W LAN Inter-TSF trusted channel; Cryptographic mechanisms: FCS_RNG_EXT.1 Random number generation; FCS_CKM.1/Asym metric Cryptographic key generation; FCS_CKM.1/Sym metric Cryptographic key generation; FCS_COP.1/SigGe n Cryptographic operation; FCS_COP.1/KeyE st Cryptographic operation; FCS_COP.1/Symm etric Cryptographic operation; FCS_COP.1/Deriv ation Cryptographic operation; FCS_COP.1/Hash Cryptographic operation; FCS_COP.1/Keyed Hash Cryptographic operation The Integrity of the communication is defined within the TOE scope for Bluetooth, HTTPS, TLS and WLAN. Other supported TSF trusted channel might be added by the manufacturer in its ST, but this is out of the scope of the present TOE. Cryptographic mechanisms are covering the state of the art. The integrity of the stored data is not clearly mentioned in ETSI TS 103 732-1 [i.2]. A possible way forward is to require in FCS_CKH_EXT.1 an explanation about the algorithm used to grant the data integrity. Alternatively, to use a new SFR from CC2022 like FDP_SDI. The report of the data corruption is not applicable to Consumer Mobile Device scenario. The user will not be able to use the data if they are corrupted and most likely the application of service affected by the problem will not work. Possible gap ETSI ETSI TR 104 071 V1.1.1 (2025-07) 12 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (g) process only data, personal or other, that are adequate, relevant and limited to what is necessary in relation to the intended purpose of the product with digital elements (data minimisation); ENISA defined an extended SAR for this purpose: ADV_PDM.1: Processed Data Minimisation (Extended). However, as described in section 8 for RDP, the CMD is expecting that the preloaded application will demonstrate their conformity to CRA separately from the CMD. This means that the data minimisation concept applies only to the CMD main OS and system services (e.g. ADP). Possible gap (h) protect the availability of essential and basic functions, also after an incident, including through resilience and mitigation measures against denial-of- service attacks; FPT_FLS.1 Failure with preservation of secure state FPT_RCV.2 Automated recovery - The TSF preserves a secure state in case of failures due to software update. The secure state can be the state before the update is executed or a state for recovery as defined in FPT_RCV.2 Automated recovery. FPT_RCV.2 mandate the TOE to return to a secure state using automated procedures. The TSF checks its integrity running a suite of self-tests at the initial start-up to demonstrate its correct operation. Incidents at that stage are countered with FPT_RCV.2. However, the DoS attacks are considered to be network attacks (or network-based, anyway), and these two SFRs are not applicable to that scenario. They are considered only to handle the case where there is a failure in the system software. The scenario where the CMD is used to perform DdoS attack against a network is protected by the authentication and authorization SFRs. Covered or NA ETSI ETSI TR 104 071 V1.1.1 (2025-07) 13 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (i) minimise the negative impact by the products themselves or connected devices on the availability of services provided by other devices or networks; The TSF protects itself trough: FPT_PHP.3 Resistance to physical attack FPT_TST.1 TSF testing The attacks scenarios are local, physical attacks, not relevant to network services. This is not really relevant for the mobile device as it is not providing services to other devices. NA (j) be designed, developed and produced to limit attack surfaces, including external interfaces; - AVA_VAN.2 Vulnerability analysis; ADV_TDS.1 Basic design; ADV_FSP.2 Security- enforcing functional specification; ADV_ARC.1 Security architecture description. ENISA request: AVA_VAN.1 Vulnerability survey; ADV_TDS.1 Basic design; ADV_FSP.1 Basic functional specification; ADV_ARC.1 Security architecture description. Covered (k) be designed, developed and produced to reduce the impact of an incident using appropriate exploitation mitigation mechanisms and techniques; FPT_FLS.1 Failure with preservation of secure state FPT_RCV.2 Automated recovery ADV_ARC.1 Security architecture description; ADV_TDS.1 Basic design; ADV_FSP.2 Security- enforcing functional specification. ENISA request: FPT_FLS.1 Failure with preservation of secure state; FPT_RCV.1 Manual recovery; ADV_ARC.1 Security architecture description; ADV_TDS.1 Basic Design; ADV_FSP.1 Basic functional specification. Covered (l) provide security related information by recording and monitoring relevant internal activity, including the access to or modification of data, services or functions, with an opt-out mechanism for the user; ENISA request: FMT_SMR.1 Security roles. In FAU_GEN.1 Audit data generation it is necessary to indicate which events are logged. In FMT_SMF.1, the opt-out mechanism (enable/disable audit function) needs to be included. Possible Gap ETSI ETSI TR 104 071 V1.1.1 (2025-07) 14 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (m) provide the possibility for users to securely and easily remove on a permanent basis all data and settings and, where such data can be transferred to other products or systems, ensure that this is done in a secure manner. FCS_CKM.4 Cryptographic key destruction - The permanent removal of the user data is one of the CMDPP Security Objectives: O.SECURE_WIPE The TOE is able to make user data assets permanently unreadable. This objective is achieved by FCS_CKM.4 specifying that keys from the key hierarchy for each class of data can be deleted on request of the user, making the data unreadable. The user data included in the User Data Assets are transferred in a secure manner using the mechanisms listed for ESR 2I. The ENISA requested SFRs are not applicable as user data are neither exported nor transmitted. The SFR FDP_RIP.1 subset residual information protection is replaced with the alternative FCS_CKM.4 achieving the equal result. Covered 7 CRA Annex I Essential Cybersecurity Requirements Part II comparison This clause compares the Essential Cybersecurity Requirements set out in CRA Annex I Part II "Vulnerability handling requirements" with the CMDPP SFR/SAR. The CRA ECR in Table 2 below are provisions defined in the Cyber Resilience Act [i.1]. ETSI ETSI TR 104 071 V1.1.1 (2025-07) 15 Table 2: Mapping of CMDPP SFRs and SARs versus CRA ECR Annex I Part II CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion Manufacturers of products with digital elements shall: - - - - (1) identify and document vulnerabilities and components contained in products with digital elements, including by drawing up a software bill of materials in a commonly used and machine- readable format covering at the very least the top-level dependencies of the products; - ALC_FLR.3 Systematic flaw remediation ALC_FLR.3 covers the first part of the ESR. The second part related to a software bill of materials is not currently covered. ENISA is defining an extended SAR: ALC_SBM.1: Software bill of materials (Extended), but it would be useful to evaluate if this can be achieved with alternative SARs defined in CC2022. Possible Gap (2) in relation to the risks posed to products with digital elements, address and remediate vulnerabilities without delay, including by providing security updates; where technically feasible, new security updates shall be provided separately from functionality updates; - ALC_FLR.3 Systematic flaw remediation ENISA request at minimum ALC_FLR.1 Basic flaw remediation. The extended component defined by ENISA about the distinction between security and functional updates is not applicable for the CMD scenario where in several cases the way to remediate a vulnerability is a mix between a security and a functional update. Covered (3) apply effective and regular tests and reviews of the security of the product with digital elements; - - ENISA is defining an extended SAR: ALC_PSR.1 Periodic security review and testing. However, the EUCC surveillance mechanism implicitly fulfil the requirement. The EUCC scheme itself is covering the requirement. Covered ETSI ETSI TR 104 071 V1.1.1 (2025-07) 16 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (4) once a security update has been made available, share and publicly disclose information about fixed vulnerabilities, including a description of the vulnerabilities, information allowing users to identify the product with digital elements affected, the impacts of the vulnerabilities, their severity and clear and accessible information helping users to remediate the vulnerabilities; in duly justified cases, where manufacturers consider the security risks of publication to outweigh the security benefits, they may delay making public information regarding a fixed vulnerability until after users have been given the possibility to apply the relevant patch; - ALC_FLR.3 Systematic flaw remediation ENISA request at minimum ALC_FLR.1 Basic flaw remediation. Covered (5) put in place and enforce a policy on coordinated vulnerability disclosure; - - This is not applicable to the product certification. NA (6) take measures to facilitate the sharing of information about potential vulnerabilities in their product with digital elements as well as in third party components contained in that product, including by providing a contact address for the reporting of the vulnerabilities discovered in the product with digital elements; - ALC_FLR.3 Systematic flaw remediation ENISA request at minimum ALC_FLR.2 Flaw reporting procedures. Covered (7) provide for mechanisms to securely distribute updates for products with digital elements to ensure that vulnerabilities are fixed or mitigated in a timely manner and, where applicable for security updates, in an automatic manner; - ALC_FLR.3 Systematic flaw remediation ENISA request at minimum ALC_FLR.3 Systematic flaw remediation. The extended component defined by ENISA about the distinction between security and functional updates is not applicable for the CMD scenario where in several cases the way to remediate a vulnerability is a mix between a security and a functional update. Covered ETSI ETSI TR 104 071 V1.1.1 (2025-07) 17 CRA ECR CMDPP SFRs CMDPP SARs Rationale Conclusion (8) ensure that, where security updates are available to address identified security issues, they are disseminated without delay and, unless otherwise agreed between a manufacturer and a business user in relation to a tailor-made product with digital elements, free of charge, accompanied by advisory messages providing users with the relevant information, including on potential action to be taken. - ALC_FLR.3 Systematic flaw remediation ENISA request at minimum ALC_FLR.3 Systematic flaw remediation. The extended component defined by ENISA about the distinction between security and functional updates is not applicable for the CMD scenario where in several cases the way to remediate a vulnerability is a mix between a security and a functional update. The commercial aspects of the ESR is not applicable to the CMDPP. Covered |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 8 Remote data processing | The remote data processing service is defined in the Cyber Resilience Act [i.1] as a data processing at a distance for which the software is designed and developed by the manufacturer, or under the responsibility of the manufacturer, and the absence of which would prevent the product with digital elements from performing one of its functions. In the context of the CMDPP it is necessary to make some assumptions on the TOE defined in ETSI TS 103 732-1 [i.2] concerning the CMD functionalities that involve a remote data processing: • The functionalities related to cellular mobile communication (that are actually out of the scope of CMDPP) have their counterpart in the cellular mobile network that performs a remote data processing to allow the cellular mobile communication service. However, the cellular mobile network is not designed and developed by the CMD manufacturer and it is under the responsibility of the Mobile Network Operators (MNOs), therefore this remote data processing cannot be considered as part of the CMD intended as product with digital element. • The preloaded application are part of the TOE and their SFR/SAR are described in the base PP ETSI TS 103 732-1 [i.2] and the PP Module ETSI TS 103 732-4 [i.6]. The scope of the preloaded applications SFR/SAR is to grant they are not negatively affecting the main OS and the CMD user data. The functionalities of each preloaded application are not considered as part of the TOE and it is expected they are separately tested or certified (whatever will be the scheme used). This means that in the context of the CRA [i.1] the preloaded application remote data processing, if any, is to be handled within the preloaded application CRA conformity and not in the CMD CRA conformity. Moreover, some preloaded applications, even if they are included in the CMD by the CMD manufacturer, are designed and developed by a third party that design, develop and operate the remote data processing; therefore, in these cases, this remote data processing cannot be considered as part of the CMD intended as product with digital element. • The remote services provided to the CMD in order to perform its functionalities are described in clause 4 of ETSI TS 103 732-1 [i.2]. The CMD uses and Application Distribution Platform (ADP) to allow the user to download the mobile device applications. Such ADP is usually provided by the CMD manufacturer and provides also the functionality to install the mobile application on the device. When the ADP is designed, developed and operated by the CMD manufacturer the related remote data processing is considered part of the CMD intended as product with digital element. The actual version of the CMDPP is not covering this part therefore there is a gap to cover for this functionality. The case where the user decides to use a third party ADPs is out of the scope of the CMDPP and it is also not in the scope of the CRA [i.1] due to the fact these ADPs are not under the control of the manufacturer. ETSI ETSI TR 104 071 V1.1.1 (2025-07) 18 • Other remote servers may be present in the CMD ecosystem; one on them is the trustworthy update server which provides secure update to the CMD system software. This server is fully under the control of the CMD manufacturer therefore the remote data processing is considered part of the CMD intended as product with digital element. The actual version of the CMDPP is not covering this part therefore there is a gap to cover for this functionality. Other servers, if any, has to be handled case by case. The concern with the Remote data processing being included within the direct boundary of the CMDPP is that it is difficult to properly scope the requirements between both the client and remote server into a single set of requirements. This will have to be discussed as to the best way to cover these requirements. |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 9 Gap analysis | |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 9.1 Product scope and TOE | The parts that are today out of the scope of the CMDPP have to be handled to cover the entire Consumer Mobile Device products in light of the CRA conformity. In particular the two aspects to cover are: • eUICC: the eUICC is part of the CMD and the CMD manufacturer has to grant its conformity. This can be achieved reusing the eUICC Common Criteria certification based on the related protection profile [i.19] or considering the GSMA eSA eUICC certification scheme [i.17] and [i.18]. • Cellular modem: the 3GPP functionalities of the cellular modem are out of the scope of the CMDPP. This has to be handled even due to the fact that the modem is an important product in the CRA context. There are some potential solutions that can be considered to fill this gap in the scope: - Re-use GCF and/or PTCRB certification. - Consider the CRA compliancy of the modem (e.g. using the vertical harmonised standard for the routers, modems intended for the connection to the internet, and switches [i.1]). - A combination of both. These ways forward could be part of the CMDPP Assumption section. 9.2 Security Functional Requirements There are some gaps to be filled in the next version of the CMDPP: 1) CRA ECR Annex I Part I article 2(f) related to the integrity of the stored data; this is not explicitly covered by the CMDPP. This could be solved with the CC2022 FDP_SDI or with an application note of the existing FCS_CKH_EXT.1. 2) CRA ECR Annex I Part I article 2(g) related to data minimization. It is important to clarify that the CMDPP data minimization is not covering the mobile application behaviours This means that the data minimization concept applies only to the CMD main OS and system services (e.g. ADP). 3) CRA ECR Annex I Part I article 2(l) related to recording and monitoring relevant internal activity; there could be a gap because today the CMDPP does not provide any SFR/SAR on this topic. However further investigation is needed to understand which level of monitoring is needed also considering the monitoring done by online services linked to the user accounts. The opt-out mechanism for the user is not relevant in the Consumer Mobile Device case because it assumes a specific knowledge of the logged information. Lastly GDPR implication needs to be considered. |
0d62cae81ffc0606506b6bb7d993a1b1 | 104 071 | 9.3 Security Assurance Requirements | The main gap identifies in the SAR section is related to the CRA ECR Annex I Part II article 1 related to the SBOM requirement. This gap can be filled with an extended SAR or with a combination of existing SAR in CC2022. ETSI ETSI TR 104 071 V1.1.1 (2025-07) 19 9.4 Remote Data Processing Accordingly, with the analysis did in clause 8 of the present document, the remote data processing solution to be considered a part of the CMDPP are the ADP provided by the CMD manufacturer and the remote server used to provide secure update of the system software when this involve the processing of the user data. ETSI ETSI TR 104 071 V1.1.1 (2025-07) 20 Annex A: Change history Date Version Information about changes September 2024 V0.0.1 Introduction, Scope, References and skeleton of the present document April 2025 V0.0.7 Stable draft April 2025 V0.0.9 Stable draft after the rapporteur call May 2025 V0.0.11 Final draft for approval May 2025 V0.0.12 Answer to TO comments May 2025 V0.0.13 Solved the last editorial comments ETSI ETSI TR 104 071 V1.1.1 (2025-07) 21 History Document history V1.1.1 July 2025 Publication |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 1 Scope | The purpose of the present document is to identify technical terms used within Rail Telecommunications (RT) Technical Specifications for the purpose of: - Ensuring that editors use terminology that is consistent across specifications. - Providing a reader with convenient reference for technical terms that are used across multiple documents. - Preventing inconsistent use of terminology across documents. The present document is a collection of terms and definitions and provides a tool for further work on FRMCS technical documentation and facilitates their understanding. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 2 References | |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 2.1 Normative references | Normative references are not applicable in the present document. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 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. In the case of a reference to a TC RT document, a non-specific reference implicitly refers to the latest version of that document. 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] UIC FRMCS SRS: "Future Railway Mobile Communication System; System Requirements Specification; AT-7800". [i.2] ETSI TS 123 280: "LTE; Common functional architecture to support mission critical services; Stage 2 (3GPP TS 23.280)". [i.3] UIC FRMCS FFFIS: "FRMCS FFFIS; Form Fit Functional Interface Specification; FFFIS-7950". [i.5] Directive 2012/34/EU of the European Parliament and of the Council of 21 November 2012 establishing a single European railway area. [i.6] Recommendation ITU-T I.112: "Integrated services digital network (ISDN); General structure; Vocabulary for ISDNs". [i.7] UIC FRMCS TOBA FRS: "On-Board FRMCS; Functional Requirements Specification; TOBA- 7510". [i.8] UIC FRMCS FIS: "Future Railway Mobile Communication System; Functional Interface Specification; FIS-7970". ETSI ETSI TR 103 791 V1.1.1 (2025-06) 6 |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 3 Definition of terms, symbols and abbreviations | |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 3.1 Terms | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 3.2 Symbols | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 3.3 Abbreviations | 3GPP 3rd Generation Partnership Project EU European Union FFFIS Form Fit Functional Interface Specification FIS Functional Interface Specification FRMCS Future Railway Mobile Communication System FSCP FRMCS Service Control Plane FSUP FRMCS Service User Plane H2N Host-to-Network RMR Railway Mobile Radio UIC Union Internationale de Chemin de fer SRS System Requirements Specification TOBA Telecom On-Board Architecture TR Technical Report TS Technical Specification |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4 Basic FRMCS definitions | |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.1 0-9 | 5G Core: a mandatory component of the FRMCS domain in the Transport Stratum. 5G Core system architecture and functions are defined in 3GPP, specifying how mobile core network should evolve to support the needs of 5G New Radio (NR A component in the Transport Stratum. 5G Core system architecture and functions are defined by 3GPP to support the needs and use cases of e.g. 5G New Radio (NR). 5GS: a 5th Generation System that represents a network architecture encompassing 5G Core, Next-Generation Radio Access Network (NG-RAN) and User Equipment (UE). |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.2 A | domain of applicability: FRMCS domain(s) from which a "non-interoperable application" (i.e. of application type II or IV) is able to operate. NOTE: In other FRMCS domains which are not part of the application domain of applicability, service attempts by the application are expected to fail. application function: the FRMCS service server within a service domain which performs signalling for applications. application type: category of application identifying whether an application is interoperable and whether an application is an IM application or a RU application. NOTE: Term derived from UIC FRMCS SRS [i.1]. application plane: interaction plane providing the data exchange between endpoint applications. ETSI ETSI TR 103 791 V1.1.1 (2025-06) 7 |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.3 B | base station: equipment responsible for radio transmission and reception. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.4 C | communication services: services enabling two-way communication between two or more authorised service users (i.e. applications) from applications towards other applications/entities reachable through various networks. NOTE: Term derived from UIC FRMCS SRS [i.1]. complementary services: services providing support to communication services and the railway application stratum (such as providing and/or utilizing the location of the service user, etc.). NOTE: Term derived from UIC FRMCS SRS [i.1]. coupling mode: operating mode, either loose-coupled mode or tight-coupled mode, used by an application to interact with the FRMCS system. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.5 D | domain: the highest-level group of functional entities. Reference points are defined between domains. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.6 E | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.7 F | foreign FRMCS domain: FRMCS domain which is not the Home FRMCS domain for a given train. FRMCS domain: administrative domain which comprises a service domain and a transport domain under the control of an FRMCS operator. NOTE: Term derived from UIC FRMCS SRS [i.1]. FRMCS operator: railway infrastructure manager, or an operator delegated by a railway infrastructure manager, who manages the FRMCS transport domain and/or FRMCS service domain for which FRMCS policies and FRMCS user subscriptions are applicable. NOTE: Term derived from UIC FRMCS SRS [i.1]. FRMCS service client: client that enables the use of the communication services and/or complementary services for the railway applications. NOTE: MC service client is one example of an FRMCS service client. FRMCS Service Control Plane (FSCP): interaction plane providing the signalling for session establishment and teardown between FRMCS service client and FRMCS service server. FRMCS service server: server application functions acting as counterparts to FRMCS service clients. FRMCS Service User Plane (FSUP): - interaction plane for data exchange between endpoint applications providing loose-coupled applications with tunnelling through FRMCS service clients; - interaction plane equivalent to the application user plane for tight-coupled applications. FRMCS system: telecommunication system conforming to FRMCS specifications, consisting of transport stratum and service stratum. ETSI ETSI TR 103 791 V1.1.1 (2025-06) 8 FRMCS user: human or machine making use of communication services and/or complementary services. FRMCS on-board application profile: set of parameters associated to an application within the On-Board FRMCS which allows or forbids specific operations over OBAPP or characterize the application for specific operations (e.g. Startup Application or not). FRMCS railway on-board profile: set of parameters associated to an application within the On-Board FRMCS which enables the communication services or prescribe specific behaviours in domain transitions. FRMCS railway trackside profile: set of parameters associated to an application within the FRMCS trackside Gateway which enables the communication services. FRMCS trackside application profile: set of parameters associated to an application within the FRMCS trackside gateway which allows or forbids specific operations over TSAPP. 4.8 G Void. 4.9 H home FRMCS domain: FRMCS domain which is considered by default as the home of the train. NOTE 1: Term derived from UIC FRMCS SRS [i.1]. NOTE 2: To the notion of Home are associated elements of profile at the transport stratum level (such as SIM profile) and at the Service Stratum level (such as primary MC system). H2N network endpoint: FRMCS service client hosted by a FRMCS Trackside Gateway that enables connectivity towards a specific IP network without requiring individual hosts of the said network to be individually locally-bound to the FRMCS trackside gateway. 4.10 I IM application: an application that is either interoperable or non-interoperable and associated to an Infrastructure Manager (IM). NOTE: Infrastructure manager is defined in point 2 of Article 3 in Directive 2012/34/EU [i.5]. 4.11 J Void. 4.12 K Void. 4.13 L local binding: procedure between an application and the On-Board FRMCS (respectively the FRMCS trackside gateway) enabling the establishment of a secure mutually-authenticated link as a pre-requisite to subsequent OBAPP (respectively TSAPP) control plane information exchanges. NOTE: Term derived from UIC FRMCS FFFIS-7950 [i.3]. loose-coupled application: application which interacts with the FRMCS System through the On-Board FRMCS via OBAPP or through the FRMCS trackside gateway via TSAPP after a successful Local Binding and calls API features of OBAPP / TSAPP to use FRMCS communication services. ETSI ETSI TR 103 791 V1.1.1 (2025-06) 9 loose-coupled mode: operating mode used by a loose-coupled application to interact with the FRMCS system. 4.14 M MC service client: a generic name for the client application function of a specific MC service. MC service client could be replaced by MCPTT client, or MCVideo client, or MCData client depending on the context [i.2]. MC service user: an authorized user, who can use an MC service UE to participate in one or more MC services. NOTE: Term defined in ETSI TS 123 280 [i.2]. MC system: the collection of applications, services, and enabling capabilities required to provide a single mission critical service or multiple mission critical services to one or more mission critical organizations [i.2]. NOTE: In an FRMCS System, an MC System is one example of a service domain for the specific support of mission critical communications using MCPTT and MCData communication services. MC user: a user identified by an MC ID, who after authorization obtains mission critical service(s). NOTE: Term derived from UIC FRMCS FIS [i.8]. 4.15 N Void. 4.16 O on-board FRMCS: system enabling FRMCS communication services for on-board applications [i.7]. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.17 P | proxy: person or entity that is acting or being used in the place of someone or something else. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.18 Q | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.19 R | railway application stratum: railway-specific functionalities using services offered by the service stratum reference point: conceptual point applicable for interaction between functional services that enables authorized functions, e.g. in the network, to access their services. reference point: a conceptual point at the conjunction of two non-overlapping functional groups. NOTE 1: A reference point only becomes a physical interface when the functional entities on either side of it are contained in different physical equipment units. NOTE 2: Term defined in Recommendation ITU-T I.112 [i.6]. RU application: an application that is either Interoperable or Non-interoperable and associated to a Railway Undertaking (RU). NOTE: Railway undertaking is defined in point 1 of Article 3 of [i.5]. ETSI ETSI TR 103 791 V1.1.1 (2025-06) 10 |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.20 S | service domain: implementation of the Service Stratum belonging to a unique organization and is operated by a unique organization. service stratum: set of functions to enable communication services, complementary services and operations and maintenance services for the FRMCS system. startup application: application for which the On-board FRMCS takes specific measures as part of the FRMCS Start of Operation procedure. NOTE 1: Term derived from UIC FRMCS SRS [i.1]. NOTE 2: Startup Applications are identified as such as part of the FRMCS onboard application profile. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.21 T | tight-coupled application: application which interacts with the FRMCS System through the On-Board FRMCS via OBAPP or through the FRMCS Trackside Gateway via TSAPP after a successful local binding and directly uses standard reference points of the service domain. tight-coupled mode: operating mode used by a tight-coupled application to interact with the FRMCS system. Transport domain: implementation of the transport stratum belonging to a unique organization and is operated by a unique organization. Transport Stratum: set of access and corresponding core functions applicable for the FRMCS system. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.22 U | user equipment: communication unit providing the physical and logical functions necessary to access the FRMCS network. It includes the radio interface, processing elements, and interfaces to train systems, enabling railway-specific communication services. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.23 V | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.24 W | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.25 X | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.26 Y | Void. |
2feca01c1cd3ce8a61f96d1e5a9f34d9 | 103 791 | 4.27 Z | Void. ETSI ETSI TR 103 791 V1.1.1 (2025-06) 11 History Document history V1.1.1 June 2025 Publication |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 1 Scope | The present document describes applications for SRDs in the 76 - 77 GHz which may require a change in the present regulatory framework for the proposed band. It includes in particular: • Market information; • Technical information regarding equipment type and typical installation; • Regulatory issues. For the applications described, the intended and unwanted emissions are within the current harmonized regulations for SRDs. The regulatory changes that would be required for their realization are relaxations on usage restrictions. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 2 References | |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 2.1 Normative references | Normative references are not applicable in the present document. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 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 TR 103 148 (V1.1.1): "Electromagnetic compatibility and Radio spectrum Matters (ERM); System Reference document (SRdoc); Technical characteristics of Radio equipment to be used in the 76 GHz to 77 GHz band; Short-Range Radar to be fitted on fixed transport infrastructure". [i.2] ECC Report 262 (2017): "Studies related to surveillance radar equipment operating in the 76 to 77 GHz range for fixed transport infrastructure". [i.3] ERC/REC 70-03 (7 June 2024): "ERC Recommendation of 1997 relating to the use of Short Range Devices (SRD)". [i.4] Commission Implementing Decision (EU) 2022/180 of 8 February 2022 amending Decision 2006/771/EC updating harmonised technical conditions in the area of radio spectrum use for short- range devices. [i.5] ETSI EN 301 091-1 (V2.1.1): "Short Range Devices; Transport and Traffic Telematics (TTT); Radar equipment operating in the 76 GHz to 77 GHz range; Harmonised Standard covering the essential requirements of article 3.2 of Directive 2014/53/EU; Part 1: Ground based vehicular radar". [i.6] ETSI EN 301 091-2 (V2.1.1): "Short Range Devices; Transport and Traffic Telematics (TTT); Radar equipment operating in the 76 GHz to 77 GHz range; Harmonised Standard covering the essential requirements of article 3.2 of Directive 2014/53/EU; Part 2: Fixed infrastructure radar equipment". ETSI ETSI TR 104 052 V1.1.1 (2025-06) 8 [i.7] ETSI EN 301 091-3 (V1.1.1): "Short Range Devices; Transport and Traffic Telematics (TTT); Radar equipment operating in the 76 GHz to 77 GHz range; Harmonised Standard covering the essential requirements of article 3.2 of Directive 2014/53/EU; Part 3: Railway/Road Crossings obstacle detection system applications". [i.8] ETSI EN 303 360 (V1.1.1): "Short Range Devices; Transport and Traffic Telematics (TTT); Radar equipment operating in the 76 GHz to 77 GHz range; Harmonised Standard covering the essential requirements of article 3.2 of Directive 2014/53/EU; Obstacle Detection Radars for Use on Manned Rotorcraft". [i.9] ETSI EN 303 661 (V1.1.1): "Short Range Devices (SRD); Ground Based Synthetic Aperture Radar (GBSAR) in the frequency range 17,1 GHz to 17,3 GHz and High Definition Ground Based Synthetic Aperture Radar (HD-GBSAR) in the frequency range 76 GHz to 77 GHz; Harmonised Standard for access to radio spectrum". [i.10] ETSI TR 103 664 (V1.1.1): "System reference document (SRdoc); Security Scanners (SSc) within the frequency range from 60 GHz to 90 GHz". [i.11] ECC Report 344 (2022-10): "Sharing and compatibility studies of Security Scanners (SScs) within frequency range 60-82 GHz". [i.12] ECC Report 222 (2014-09): "The impact of Surveillance Radar equipment operating in the 76 to 79 GHz range for helicopter application on radio systems". [i.13] ECC Report 315 (2020-05): "Feasibility of spectrum sharing between High-Definition Ground Based Synthetic Aperture Radar (HD-GBSAR) application using 1 GHz bandwidth within 74-81 GHz and existing services and applications". [i.14] ECC/DEC/(16)01 (4 March 2016): "The harmonised frequency band 76-77 GHz, technical characteristics, exemption from individual licensing and free carriage and use of obstacle detection radars for rotorcraft use". [i.15] ERC/REC 74-01 (May 2022): "Unwanted emissions in the spurious domain". [i.16] ETSI TR 104 078: "System Reference document (SRdoc); Short Range Devices; Radar equipment operating in 57 GHz to 64 GHz and 76 GHz to 77 GHz for applications on drones". |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 3 Definition of terms, symbols and abbreviations | |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 3.1 Terms | Void. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 3.2 Symbols | Void. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 3.3 Abbreviations | For the purposes of the present document, the following abbreviations apply: CP Critical Part e.i.r.p. equivalent isotropic radiated power FFT Fast Fourier Transform FIR Fixed Infrastructure Radar FMCW Frequency Modulated Carrier Wave FOD Foreign Object Debris FoF Friend or Foe FoV Field of View ETSI ETSI TR 104 052 V1.1.1 (2025-06) 9 FSSA Fixed Security and Safety Applications GSV Ground Support Vehicles HD-GBSAR High-Definition Ground Based Synthetic Aperture Radar LDC Low Duty Cycle LPR/TLPR (Tank) Level Probing Radar PIDS Perimeter Intrusion Detection System RSM Runway Surface Movement SSc Security Scanner TTT Transport and Traffic Telematics WAM Wide Area Monitoring |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 4 Comments on the System Reference Document | No ETSI member raised any comments. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5 Presentation of the system or technology | |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.1 General overview | The band 76 - 77 GHz is already used by many applications including ground-based vehicle and TTT infrastructure systems (ERC/REC 70-03 [i.3] Annex 5), obstruction/vehicle detection via radar sensor at railway level crossings (ERC/REC 70-03 [i.3] Annex 4), obstacle detection radars for rotorcraft use (ERC/REC 70-03 [i.3] Annex 5), HD-GBSAR (ERC/REC 70-03 [i.3] Annex 6) and LPR/TLPR (ERC/REC 70-03 [i.3] Annex 6). The technology that is discussed in the present document is that which is already in use in the 76 - 77 GHz band but limited to certain applications. There therefore exists a large body of experience in manufacturing and use of radars in this band. Dedicated semiconductor devices are available from several manufacturers. Design and manufacture of antenna systems has been perfected. Deployed equipment typically falls into two types: high value fixed installations that are professionally installed and operated and mass market, price sensitive devices. In terms of technology, both types are typically FMCW radars with a total RF power of the order of 10 mW. They all rely on digital processing, such as FFTs, to extract target information from the reflected radar signals and for post processing of this information. Many fixed installations have a single antenna that forms a very narrow beam, and the antenna is scanned mechanically to cover the Field of View (FoV). Fixed antennas with single or multiple wider beams are also found. Mass market systems are highly integrated. They may have multiple small antennas combined with the RF circuitry into a small module. In terms of systems, the main applications for fixed installations are TTT infrastructure (ERC/REC 70-03 [i.3] Annex 5), HD-GBSAR (ERC/REC 70-03 [i.3] Annex 6) and radar sensors at railway level crossings (ERC/REC 70-03 [i.3] Annex 4). The main application for mass market equipment is on ground-based vehicles (ERC/REC 70-03 [i.3] Annex 5). With the band listed in multiple Annexes in ERC/REC 70-03 [i.3], there is potential for uncertainty in exactly what applications and uses are permitted. Is, for instance, monitoring an airport runway for debris a TTT function or a safety function? In the list of applications that follow, there will be some that are permitted by many administrations. Others may be edge cases or lie in a grey area. One purpose of the present document is to seek to remove such uncertainties and to move to a harmonised position among CEPT member states. ETSI ETSI TR 104 052 V1.1.1 (2025-06) 10 |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2 Fixed Security and Safety Applications (FSSA) | 5.2.1 System description The scanning radar systems provide an Automatic Incident Detection solution for a range of safety/security applications for strategic or sensitive sites. These can include but are not limited to airfields, power stations, data centres, mines, and other critical national infrastructure. By continually measuring and tracking objects through a wide FoV using high frequency radar the system can generate incident alerts, whilst maintaining extremely low nuisance alarm rates. Systems of this nature would be similar to existing fixed infrastructure radars for TTT applications as in Annex 5 of ERC/REC 70-03 [i.3]. They would meet the technical requirements in ETSI EN 301 091-2 [i.6] for fixed infrastructure radars. Given the applications and typical deployment locations described below, fewer than 10 systems per site would be expected. The expected number of sites is around 3 per major city. These assumptions are discussed further in clause 6. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.2 Site and perimeter protection | |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.2.1 Radar for site security and safety applications | Fixed infrastructure radar can be used to detect and track vehicles or people in and around critical national infrastructure and other important sites. This would include but not be limited to airports, power stations, refineries, ports/harbours or data centres. The threat is not limited to possible terrorist activity. For example, in UK airports, environmental protestors have breached the perimeter and caused major delay and disruption. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.2.2 Perimeter Intrusion Detection System (PIDS) | The specific application is to detect breaches of the site perimeter. This could be deployed at airports, critical National Infrastructure sites as well as some civilian installations such as car storage facilities, data centres and private energy sites where a threat would typically originate from outside of the site perimeter. A key advantage of a radar for this application is the ability to detect and track in all weather conditions with continual monitoring of multiple objects in real time within a site. Traditional PIDS solutions offer an alert to breach but a radar solution can offer live and historic tracking of multiple targets of interest to enable a fast response for interception as well as historic forensic analysis of breach locations. As has been identified in numerous airfield breaches, knowing that an intruder(s) has entered the site is one thing but the ability to monitor live locations of multiple targets once inside can prove vital to maintaining integrity of vulnerable areas of the site. In cases such as this a full situational awareness is vital in leading an effective response. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.2.3 Wide Area Monitoring including Wildlife Detection | In the general security installations, the system objective is for wide area monitoring for detection and alerting to unusual movement of vehicles, people and wildlife within a site. A typical installation in an airport may see one or more scanning radars installed such that the radar FoV would cover the open area between the perimeter fence to the runway or terminal/maintenance buildings. In this case the system would be looking to detect pedestrians in restricted zones or larger wildlife that could pose a danger to aircraft or ground support equipment. Typical systems may be longer range > 1 km line of sight. This results in fewer physical systems required to achieve desired coverage as well as minimizing costs associated with ground works required for data and power to each system. Typical installation mounting height of 4 - 8 m with a 360° FoV. Systems such as WAM and PIDS may be combined with FoF solutions in order to reduce nuisance alarms. ETSI ETSI TR 104 052 V1.1.1 (2025-06) 11 |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.2.4 Border Protection | Monitoring of border areas over long distances is becoming particularly important with increasing migration levels. Early detection and monitoring positions of objects post-breach can enable the relevant response personnel to respond in the most appropriate and timely manner. 5.2.2.5 Port and Harbour Protection Perimeter protection is difficult for areas crossing or bordering water. Manned patrols on water or shoreline cannot cover the entire surface in time. Constant monitoring of all activities, including the water surface, can be necessary for effective security. A typical installation may include harbours shared with both military and civilian watercraft. Monitoring surface movement in shared environments is a difficult task where public access is permitted in close proximity to restricted zones around military assets. Short range fast detection of object entering or approaching restricted zones in all weather and sea conditions can be vital in ensuring integrity of those assets. Port and harbour protect may also consist of overland monitoring or a combination of both to protect and monitor access from both. In this example the equipment used to monitor the site may be the same as that used to monitor traffic from the roadside. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.3 Airport - airside protection | |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.3.1 Runway Surface Movement | RSM monitoring is primarily a safety application used to aid control tower monitoring of aircraft and GSV in and around the runway and taxiways at major airports where line of sight can be restricted by poor whether such as fog and rain. This is a safety application aimed at reducing the chance of collisions. Typical installations would be 50 - 100 m from the runway at 2 - 3 m installation height. Such a system would almost certainly be seen as TTT as it is in a transport environment and monitoring traffic. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.3.2 Critical Part Line Monitoring | Monitoring the CP line in airports is a complex task due to the constant movement and strict security requirements. Managing the operational flow and preventing potential security threats poses significant challenges for airport authorities. Airports create a virtual line where physical barriers are not possible and are monitored to alert operators of unauthorized access. Accurately detecting and tracking objects continuously, even in adverse weather and varying light conditions is paramount importance. Typical installations would require a relatively short range of operation with a relatively high update rate. Radar mounting height would be expected at 2 - 3 m mounting height. In this example the equipment used to monitor ground traffic may be the same as that used to monitor traffic from the roadside. Such a system would probably be seen as TTT as it is in a transport environment and monitoring traffic, although there is a major safety aspect. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.3.3 FOD Detection | In this scenario a scanning radar would be used to automatically identify unwanted objects (FOD) on runways. At most airports this is a manual task completed by human observers inspecting the runway at set intervals. An automatic FOD detection system reduces the requirement for manual inspections and reduces the management overhead of ensuring safety of the inspection teams operating on active runways. In this application the installation would be expected to single digit numbers per runway with an installation height of less than 2 m and within 100 m of the runway. Here it is less certain that all administrations would see this as TTT. It is in a transport environment, but the application is safety rather than traffic management. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.4 Ports & Maritime | 5.2.4.1 Situational awareness Perimeter and site security is discussed in clause 5.2.2.5 above. Monitoring of movement within a port or harbour is also vital. ETSI ETSI TR 104 052 V1.1.1 (2025-06) 12 Because it is difficult to place infrastructure in the water and because of the need to operate in all weather and sea conditions, short range radar is the best, and possibly the only feasible, solution. This application is in a transport environment and the situational awareness function would almost certainly be seen as TTT. But in practice the security function is integrated into the same equipment and system, so the installation as a whole may be seen as a grey area. 5.2.4.2 Quayside collision prevention Vessels manoeuvring too fast and/or on the wrong heading repeatedly cause damage to port infrastructure, piers, sluices, floodgates, fairway limitation/buoy and other ships. To prevent this, a radar on the infrastructure side (quay, jetty or pontoon) could be foreseen as part of an assistance / protection system. Such system integrates highly robust maritime radar sensors that are installed at various points on the maritime infrastructure and is in direct contact with the crew on the ships via various end devices (signal lamps, displays, etc.). The information on local conditions helps the crew to navigate the ship safely even during adverse weather conditions affecting the ship captain's sight (fog, heavy rain, snow, nighttime etc) or affecting the ship's manoeuvrability (heavy winds / gusts). Such a system could be seen as part of TTT infrastructure, although some might argue the application is safety. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.2.5 Scanning antennas | With a very narrow beam antenna, a long-standing method of monitoring a wide FoV is to mechanically rotate (scan) the antenna assembly in azimuth. This has the effect, for a receiver in the FoV, of simulating a very low duty cycle with a repetition frequency dependent upon the scan rate. A narrow beam can also be scanned electronically, though usually only over a limited angle. Other electronic techniques include processing multiple fixed beams. Antennas of a scanning nature have been mandated for fixed infrastructure radars. The reason is that the effect in the time domain of a mechanically scanned narrow beam provides mitigation to vehicular radars. One proposal of the present document is that this requirement is applied only to roadside installations. Away from vehicular radar there would be no benefit to this restriction. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.3 Uncrewed aircraft systems | There is a precedent for Obstacle Detection Radar in the 76 - 77 GHz band, for use upon manned rotorcraft, following ERC/REC 70-03 [i.3] and ECC/DEC/(16)01 [i.14]. Extending such usage to uncrewed aircraft, e.g. drones, could be seen as a subject for the present document. ETSI members, however, have agreed to produce a separate SRdoc on airborne radar systems that considers a wider frequency range. Information on the use of 76 - 77 GHz for airborne radar is included in this separate ETSI SRdoc ETSI TR 104 078 [i.16]. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 5.4 Millimetre Wave Security Scanners | The proponents of the present document are aware of previous and current work in CEPT and ETSI on Security Scanners (SScs), including: • ETSI TR 103 664 [i.10]. • ECC Report 344 [i.11]. • ETSI is also developing a draft harmonised standard for security screening applications. This clause explains the differences between SScs and the systems described here. ETSI ETSI TR 104 052 V1.1.1 (2025-06) 13 Use Case The SSc referred to above are intended for detection and examination of small objects (e.g. concealed weapons) at close range. The systems described here are intended for detection of larger objects (e.g. vehicles, people) at larger ranges. Technology SSc are wider bandwidth and lower peak e.i.r.p. than the systems described here. In addition, the signal format and antenna type are usually different. The proponents of the present document therefore conclude that the requirements and applications described here are not able to be met within the parameters assigned to mmW SSc. The systems are in different categories, in terms both of use case and the technology. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 6 Market information in the EU | 6.1 FSSA Market Size and Value 6.1.1 General The European security radar market is expected to reach USD 2,4 billion by 2027, growing at a CAGR of 7,8 % during the forecast period (2023-2027). Government security spending and increasing concerns about border security, critical infrastructure protection, perimeter intrusion detection, and aviation safety are driving the market growth. Key market segments include: • Critical infrastructure protection (33 %): Airports, power stations, refineries, data centres, etc. • Border security (27 %): Land and maritime borders, including surveillance and intrusion detection. • Airport security (18 %): Perimeter intrusion detection, runway surface movement monitoring, foreign object debris (FOD) detection, etc. • Perimeter Intrusion Detection (17 %): Sensitive facilities, high-security zones (excluding airports). • Others (5 %): Military applications, industrial security, and other critical infrastructure sites. Focusing on Airport Security: • The EU airport security radar market, specifically for runway FOD detection, is estimated to reach USD 0,2 billion by 2027. This estimation considers an average of 15 radar sensors per runway for full FOD detection coverage and the total number of runways in the EU. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 6.1.2 Traffic and Equipment Density Forecasts | The deployment of security radar systems in the EU is expected to increase significantly in the coming years. EU initiatives like the Smart Borders Package and the European Investment Plan are driving investments in security technologies. The number of critical infrastructure sites requiring advanced security solutions is also growing. Radar adoption is expected to be higher in countries with extensive land and maritime borders, critical infrastructure networks, and busy airports. It is difficult to estimate the exact equipment density per site as it depends on factors like size, security requirements, and budget. However, reports suggest an average of: • 2 - 5 radar systems per critical infrastructure site (excluding airports) • 5 - 10 systems per large border area ETSI ETSI TR 104 052 V1.1.1 (2025-06) 14 • 3 radars per single-runway airport for full site coverage • 15 radar sensors per single runway for complete FOD detection These installations would be high value, low volume fixed installations that are professionally installed and operated. 6.1.3 Specific Application Data 6.1.3.1 Airport Security The European airport security market presents a significant opportunity for advanced radar technologies, driven by the need to: • Strengthen perimeter security: Airports face diverse threats, including unauthorized personnel intrusions, drone incursions, and ground vehicle breaches. Our radar technology can offer a powerful layer of defence by providing: - Early detection and tracking: Detect and track suspicious activity and objects approaching the perimeter in real-time, regardless of weather conditions. - Improved situational awareness: Gain a comprehensive view of the entire perimeter, enabling faster response to potential threats. - Data fusion: Integrate seamlessly with existing security systems like cameras and access control for a unified view. • Elevate wide area monitoring: Effectively monitoring large areas within the airport can be challenging. Radar technology can address this by: - Monitoring vast spaces: Cover expansive areas like runways, taxiways, and cargo zones continuously and reliably. - Identifying suspicious activity: Detect unusual movements, loitering individuals, or potential hazards like wildlife entering restricted areas. - Supporting resource allocation: Optimize security personnel deployment based on real-time insights from radar data. • Optimize access control: Secure sensitive areas within the airport with enhanced efficiency and effectiveness. Radar technology can: - Monitor restricted zones: Provide continuous surveillance of access points and identify unauthorized attempts to enter. - Enhance checkpoint screening: Integrate with existing screening systems to improve detection accuracy and efficiency. - Reduce manual oversight: Automate routine monitoring tasks, freeing up security personnel for critical interventions. Beyond these core applications, radar technology may also have the potential to contribute to areas like: - Vehicle and personnel screening: Enhance existing screening procedures with more advanced detection capabilities. - Foreign Object Debris (FOD) detection: While primarily an airport safety concern, radar can offer a proactive approach to FOD detection, potentially contributing to improved safety outcomes. - Runway incursion monitoring: Mitigate the risk of runway incursions by unauthorized vehicles or aircraft through real-time detection and tracking on and around runways. This can significantly enhance runway safety and prevent accidents. ETSI ETSI TR 104 052 V1.1.1 (2025-06) 15 6.1.3.2 Other Applications The adoption of radar for border protection, critical infrastructure protection (excluding airports), and perimeter intrusion detection (excluding airports) is also expected to grow in the coming years. Each application has its own specific market dynamics and challenges. 6.1.4 Conclusion The European security radar market offers significant growth potential, particularly for applications that go beyond traffic telematics. Radar technology has a unique ability to address specific safety and security challenges in airports, ports and high value installations. The technology's cost-effectiveness, together with automation and all-weather capabilities can create a compelling advantage in many markets. These installations would be high value, low volume fixed installations that are professionally installed and operated. |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 7 Technical information | |
c61edf069c94e8cd635d6dbdeb17b902 | 104 052 | 7.1 FSSA technical description |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.