hash
stringlengths
32
32
doc_id
stringlengths
5
12
section
stringlengths
5
1.47k
content
stringlengths
0
6.67M
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.22.3 Impacts on services, entities and interfaces
MTLF: - Act as server or client for VFL model training. - Register VFL related capabilities at NRF. - As VFL model training server, select features and related VFL model training clients, and request clients to train VFL model for a feature. - As VFL model training server, for UEs as sample, determine reachable UEs to ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23 Solution #23: Cross domain VFL involving NWDAF and AF
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.1 Description
Editor's note: This clause will describe the solution principles and architecture assumptions for corresponding key issue(s). Sub-clause(s) may be added to capture details. This solution describes how to enable ML Model training and inference using VFL with the following characteristics: - ML Model training is distribu...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.1.1 Terminology
The terminology described in clause 3.1 applies. VFL Active participant: The same as in clause 3.1, in addition the VFL Active Participant is also involved in other VFL tasks such as inference. VFL Passive participant: The same as in clause 3.1, in addition the VFL Active Participant is also involved in other VFL tasks...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.1 NWDAF acts as FL Server with VFL capabilities
Figure 6.23.2.1-1: VFL training and inference when NWDAF acts as a FL Server with VFL capabilities 1. AFs and NWDAFs with VFL capabilities register to NRF its NF profile that includes whether it can act as a FL server or client with VFL capabilities, and per supported Analytics ID, the type of intermediate results to a...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.2 AF acts as FL Server with VFL capabilities
Figure 6.23.2.2-1: VFL training and inference when the AF acts as a FL Server with VFL capabilities 1. Same as step 1 in figure 6.23.2-1, the AF is the NF that registers to NRF possibly via NEF. The NEF translates the identification of the samples such as UE identifier or External Group Id into the internal identificat...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.3 VFL capabilities within NF profile
The passive participants such as the AF or the NEF (on behalf of the AF) or the NWDAF registers its VFL capabilities into NRF: - FL capability information (extending existing one), whether the AF or NWDAF can act as a FL Server with VFL Capabilities or FL Client with VFL Capabilities or both. (MANDATORY). - If FL capab...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.4 Training procedure
The training process in VFL including agreement between the FL server and the FL clients on the training method is illustrated below. The training process is repeated until the server decides to terminate it based on the local constraints that are set by the server and distributed to the clients. When the ML Model is t...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.4.1 NWDAF as FL Server with VFL capabilities
In this procedure NWDAF is the FL Server/active participant that has the label, i.e. that is determined by NWDAF based on existing OAM provided KPIs related to QoE such packet loss or packet delay as some more listed in clause 6.4 of TS 23.288 [5] or is provided by the FL Client/passive participant i.e. AFs or NWDAFs t...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.4.2 AF as FL Server with VFL capabilities
In this procedure the AF is the FL Server with VFL capabilities/active participant that has the label, i.e. the AF collects QoE metrics obtained at the application layer, e.g. user opinion scores per video session per given time interval, the AF determines the perceived QoE that is used as label. The NWDAF is a passive...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.5 Inference procedure
The inference process occurs once the training process has been done, the server knows that there is a ML Model for an Analytics ID trained and determines which ones from the participants in the training, that shall participate in the inference. The differences between the VFL training and inference are that both a) th...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.5.1 NWDAF(AnLF) triggers the inference procedure
The NWDAF can also initiate the inference procedures for a ML Model that is trained with active and passive NWDAFs, but it is not shown in the figure below. NOTE 1: This solution considers that the NWDAF that trains the model is the same NWDAF as performs inference, as such it is considered that the trained ML Model is...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.2.5.2 AF triggers the inference procedure
Figure 6.23.2.5.2-1: The AF triggers the inference procedure 1. The AF performs analytics, then knows that data is needed from NWDAF, but instead a trained ML Model for these analytics using VFL with NWDAF is available for the same samples as the analytics request. The AF decides to ask NWDAF to perform inference of an...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.23.3 Impacts on services, entities and interfaces
AF: - Support for new Naf service operations for training. NWDAF with MTLF: - Updates Nnwdaf_MLModelTraining service to e.g. provide intermediate results. - Updates Nnwdaf_MLModelTraining to include a new service operation to notify the participants in the VFL training that a trained ML Model is available. - New Nnwdaf...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.24 Solution #24: How to support Vertical Federated Learning between NWDAF and AF
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.24.1 Terminology
VFL server: - owns labels for a VFL task and coordinate the training and inference procedures. - distributes initial ML models to each VFL client. - Performs sample and feature alignment with VFL client. - discovers and selects VFL clients to participant in an VFL procedure. - requests VFL clients to do local model tra...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.24.1 Description
This solution addresses Key Issue #2: 5GC Support for Vertical Federated Learning. This solution focuses on the scenario where the AF initiates the VFL training process. In this scenario, the AF acts as the VFL server, considered the VFL active participant, owning the label information. The NWDAF(s) act as the VFL clie...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.24.2 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.24.2.1 Procedure of NF discovery and selection
Figure 6.24.2-1 shows the procedure of NF discovery and selection. Figure 6.24.2-1: Procedure for of NF discovery and selection 1. NWDAF with the capability to participate as VFL clients, registers its NF profile with the NRF. This profile may include VFL capability information indicating whether it can support trainin...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.24.3 Impacts on services, entities and interfaces
NWDAF: - Supports sample and/or feature alignment. - Retrieves initial ML models for training from the AF. - Performs local ML model training with the initial ML models provisioned from the AF. - Performs ML model inference collaboration with other NWDAF(s) and/or the AF. NEF: - Receives VFL client NF filtering criteri...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.25 Solution #25: NF Registration and Discovery Enhancement for VFL
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.25.1 Description
This solution is proposed for KI#2 to support NF Registration and Discovery Enhancement for VFL. Considering the distributed inference nature, the inference procedure involves multiple NF instances. Therefore, the inference for VFL can be categorized into two cases, in terms of the difference between the NF instances t...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.25.2 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.25.2.1 Registration and discovery of NWDAF and AF for VFL
The VFL model training operation consists of NF discovery and VFL preparation. In the NF Discovery, the "VFL Active Participant" discovers the candidates of "VFL Passive Participant" NFs that will participate in the VFL training or inference procedure. In the VFL preparation, the "VFL Active Participant" determines the...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.25.2.1.1 NWDAF-initiated registration and discovery of AF for VFL in Scenario 1
Figure 6.25.2.1.1-1: VFL passive participant discovery procedure in Scenario 1 Steps 1 and 2 are the "VFL Passive Participant" NF(s) (i.e. AF) Discovery procedure in Scenario 1. 1-2. "VFL Active Participant" NWDAF determines to use VFL model training or inference based on operator policy, Analytic ID, Service Area/DNAI...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.25.2.1.2 AF-initiated registration and discovery of NWDAF for VFL in Scenario 2
Figure 6.25.2.1.2-1: VFL passive participant discovery procedure in Scenario 1 Steps 1 and 2 are the "VFL Passive Participant" NF(s) (i.e. AF and NWDAF) Discovery procedure in Scenario 2. 1-2. "VFL Active Participant" AF determines to use VFL model training or inference if data that cannot be obtained/exposed directly ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.25.3 Impacts on Existing Nodes and Functionality
For NWDAF and AF: - Register the "VFL capability" in NF profile to NRF, (in case of untrusted AF, register the "VFL capability" in NF profile to NRF via NEF profile). For NWDAF and AF with "VFL active participant" capability: - Support "VFL passive participant" NF (i.e. NWDAF and AF) discovery procedure.
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.26 Solution #26: NWDAF-assisted policy control with Recommendation logical function
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.26.1 Description
This solution focuses on Use Case #1: NWDAF-assisted QoS enhancement and addresses Key Issue #3: NWDAF-assisted policy control and QoS enhancement. As documented in Use Case #1, NWDAF can assist the PCF in determining QoS parameters that can achieve the expected service experience requirements. This solution introduces...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.26.1.1 Recommendation logical function
Recommendation Logic Function (ReLF) is new NWDAF functionality which may request a plurality of analytics (i.e. multiple analytic ID) from one or more NWDAFs or collect data directly from one or more NF(s) and derives recommendation information (e.g. recommended QoS parameters) that can fulfil the optimization goals (...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.26.1.2 Policy control enhancement with the assistance of NWDAF (ReLF)
According to TS 23.501 [2], TS 23.502 [3] and TS 23.503 [4], there are some cases related to NWDAF-assisted policy control: - The PCF, based on "Service Experience" analytics per UP path, determines for each DNAI, a traffic steering policy ID. SMF may use "Service Experience" analytics per UP path to select UPF. - The ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.26.2 Procedures
Figure 6.26.2-1: Procedure for NWDAF providing QoS recommendation 1. Consumer NF (e.g. PCF) sends a Recommendation request/subscribe (Recommendation ID = QoS recommendation, Recommendation Filter Information, Target of Recommendation Reporting) to NWDAF (ReLF) by invoking a Nnwdaf_RecommendationInfo_Request or a Nnwdaf...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.26.3 Impacts to Services, Entities and Interfaces
This solution introduces the Recommendation logical function (ReLF), which exposes the following services: - Nnwdaf_RecommendationSubscription service. - Nnwdaf_RecommendationInfo service. Consumer NF: - Supports to consume recommended QoS parameters from NWDAF.
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.27 Solution #27: NWDAF assisted QoS policy generation
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.27.1 Description
This solution addresses Key Issue #3 "NWDAF-assisted policy control and QoS enhancement". As described in clause 5.1.1 "Use Case #1: NWDAF-assisted QoS recommendation", the PCF may determine whether the current QoS can fully satisfy the service requirements and update the QoS parameters based on the Service Experience ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.27.2 Procedures
Figure 6.27.2-1: Procedure for QPGF assisted QoS policy generation 1. The PCF requests or subscribes to assisted QoS policy generation from the NWDAF containing QPGF, providing the service requirements (received from the AF), optionally QoS policy and targeted service experience. 2. The NWDAF containing QPGF requests o...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.27.3 Impacts on services, entities and interfaces
NWDAF: - Contains a new logical function QPGF which generates optimized QoS policy using the input data such as service requirements and optionally QoS policy and targeted service experience from PCF, and analytics information of service experience and network performance. PCF: - Requests or subscribes to assisted QoS ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.28 Solution #28: QoS Flow Analytics
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.28.1 Description
In 5GC, QoS Flows are managed by SMF. Based on the requests from UE or AF and also notification from RAN, SMF in collaboration with PCF, based on PCC rules, determines the appropriate QoS characteristics which are then sent by SMF to the RAN, UE and also configured in UPF(s) to establish the QoS Flow. Therefore, SMF ha...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.28.2 Procedures
The consumer of the proposed "QoS Flow analytics" may indicate in the request: - Analytics ID = "QoS Flow". - Target of Analytics Reporting: a single UE (SUPI) or a group of UEs (an Internal Group ID). - Analytics Filter Information optionally including: - S-NSSAI; - DNN; - Application ID; - Area of Interest; - an opti...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.28.3 Impacts on existing services, entities and interfaces
The solution has the following impacts: NWDAF: - Extending functionalities to produce QoS Flow analytics. SMF: - Extending functionalities and service to report QoS Flow events and corresponding information. PCF: - Extending functionalities to use QoS Flow analytics for QoS and policy control. - Extending functionaliti...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.29 Solution #29: How to evaluate NWDAF-assisted policy control and QoS enhancement
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.29.1 Description
Currently, the quality of NWDAF analytics is monitored by the analytics accuracy. This accuracy monitoring assumes that higher analytics accuracy leads to better performance when analytics consumers take action(s) based on the provided analytics. However, this assumption may not hold true, especially when the target pe...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.29.2 Procedures
Figure 6.29.2-1 shows the procedure to support the use case where NWDAF service consumer (e.g. PCF) subscribes performance feedback from NWDAF to evaluate the quality of QoS recommendation. Other use cases could be also supported by this procedure. Editor's note: Whether to introduce a new logical function in the NWDAF...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.29.3 Impacts on services, entities and interfaces
NWDAF: - Performance evaluation capability to compute performance evaluation metrics based on the performance feedback request information. PCF: - Provides performance feedback request information to subscribe performance feedback from NWDAF and determines QoS policy based on the performance feedback provided by the NW...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.30 Solution #30: NWDAF-assisted PDU Set assistance information
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.30.1 Description
As part of existing XRM procedures, PDU Set based handling is expected. A PDU Set is comprised of one or more PDUs carrying an application layer payload such as a video frame or video slice. The PDU Set based QoS Handling can be applied for GBR and non-GBR QoS Flows. The AF should provide PDU Set related assistance inf...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.30.2 Procedures
Existing procedure as captured in clause 4.15.6.2 of TS 23.502 [3] with addition of - DN Performance (see clause 6.14 of TS 23.288 [5]) in step 0 as a possible option for AF subscription. - NF (e.g. PCF or SMF) to request XRM Application-Specific Expected UE Behaviour parameters in step 0. - New set of XRM Application-...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.30.3 Impacts on services, entities and interfaces
AF: - Use NWDAF Analytics IDs (e.g. Network Performance or DN performance) and provide XRM Application-Specific Expected UE Behaviour parameters to 5GC (see clause 6.30.1). UDR/ UDM: - Maintain new set of XRM Application-Specific Expected UE Behaviour parameters. PCF /SMF: - Use NWDAF Analytics IDs (e.g. Network Perfor...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.31 Solution #31: PCF as RL Agent and NWDAF as Interpreter
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.31.1 Key Issue mapping
This solution addresses Key Issue #3: "NWDAF-assisted policy control and QoS enhancement".
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.31.2 Description
This solution (high level view depicted in Figure 6.31.2-1) proposes a 5GC enhancement to support the mechanism of reinforcement learning by: - Enhancing PCF as RL Agent. - Enhancing NWDAF as RL Interpreter. Figure 6.31.2-1: Procedure for PCF acting as RL Agent In this proposal the RL technique is used to learn the mos...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.31.3 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.31.3.1 PCF as RL Agent
Figure 6.31.3-1 depicts the procedure for the proposed solution. Figure 6.31.3.1-1: Procedure for PCF as RL Agent 1. PCF starts the process by invoking the new Nnwdaf_RLInterpreter service providing the definition of state(NW state) and Reward (based on QoE), targetReward (including the appId/service or the list of the...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.31.4 Impacts on services, entities and interfaces
Editor's note: This clause captures impacts on existing 3GPP nodes and functional elements. PCF: - Acting as RL Agent. NWDAF: - Acting as RL Interpreter, implements a new service to provide State and Reward to assist in RL procedures. - NWDAF supports the calculation of state=NW state. - NWDAF supports the calculation ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.32 Solution #32: NWDAF-assisted optimized QoS policies determination
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.32.1 Description
This solution addresses aspects of key issue #3 on NWDAF-assisted policy control and QoS enhancement. This solution proposes a new network analytic that combines multiple analytics produced within a set time window, such as Observed service experience and QoS sustainability. These new analytic may be called "Joint PCC ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.32.2 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.32.2.1 Procedure for optimized QoS determination based on combined Analytics
The following shows an example procedure where the PCF is triggered to produce optimized QoS policies, and requests multiple analytics, e.g. Observed Service Experience and QoS Sustainability analytics, and it uses these analytic results to determine optimized QoS parameters. Figure 6.32.2.1-1: Procedure for PCF to be ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.32.3 Impacts on services, entities, and interfaces
Editor's note: This clause captures impacts on existing services, entities, and interfaces. NWDAF: - Support for new Joint PCC Determination analytics. PCF: - Configured to be triggered to determine to optimize QoS policies.
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.33 Solution #33: QoS/policy enhancements assisted by NWDAF
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.33.1 Description
Editor's note: The terminology of this solution needs to align with TS 23.503 [4]. This solution aims to address the issues in KI#3: NWDAF-assisted policy control and QoS enhancement. As described in Use Case #2, each PDU session is associated with a default QoS rule which is normally sufficient for basic browsing or i...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.33.2 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.33.2.1 Procedures for NWDAF assisted QoS and policy determination
Figure 6.33.2.1-1: Procedures of NWDAF-assisted policy control and QoS enhancement 1. To collect the relevant information to assist with QoS and policy control decision, PCF subscribes to or send request to different data sources, e.g. NWDAF, AMF, SMF, AF, etc. This procedure might be triggered by different reasons, e....
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.33.2.2 Procedures to deploy NWDAF assisted QoS and policy determination during PDU session establishment
Figure 6.33.2.1-2: Procedures of NWDAF-assisted policy control and QoS enhancement during PDU session establishment In this clause, the PDU session establishment procedures are used as an example to illustrate the procedures of the how the 5GC will deploy the NWDAF assistance with QoS and policy determination. 1a - 1b....
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.33.3 Impacts on services, entities and interfaces
PCF: - Consider a combination of NWDAF analytics, including new or enhanced analytics from NWDAF to generate PCC Rules. - Store and update the determined QoS in a PCC Rule. NWDAF: - Collect new inputs to generate assistance information of QoS and policy control. - Generate new outputs to assist with PCF for QoS and pol...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.34 Solution #34: BDT Policy Recommendations
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.34.1 Description
The proposed solution focuses on an analytics service related to BDT policy, which can be used for: (i) policy selection assisting the PCF by providing recommendations related to which BDT polices are fit and optimal to choose from and (ii) policy negotiation by assisting the PCF to identify when a BDT policy is affect...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.34.2 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.34.2.1 General Procedure
The procedures for deriving BDT service recommendations are shown in Figure 6.34.2.1-1. Figure 6.34.2.1-1: Procedure for BDT service recommendation analytics 1. The analytics consumer is either pre-configured with the appropriate ReLF (Analytics ID = BDT policy) or it discovers it through NRF. Then it issues a subscrip...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.34.2.2 Input Data
The ReLF supporting recommendations on BDT policy shall be able to collect UE location, mobility, and policy information from the 5GC and AFs, as well as energy saving information from the OAM. Table 6.34.2.2-1: UE and Network information collected from 5GC, related AFs and OAM Information Source Description Data Volum...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.34.2.3 Output Data
The ReLF supporting data analytics on BDT policy shall be able to provide recommendations to the PCF. Table 6.34.2.3-1: BDT Policy Recommendations and Warning notifications Information Description ASP or AF ID Identify of the ASP or AF per BDT policy. > start time Indicates the recommended start time for applying a BDT...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.34.3 Impacts on services, entities and interfaces
Editor's note: This clause captures impacts on existing services, entities and interfaces. NWDAF: - Supporting prescriptive analytics engine.
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35 Solution #35: NWDAF-assisted Network Abnormal Behaviour Mitigation and Prevention
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35.1 Description
This solution resolves KI#4 for NWDAF enhancements to support network abnormal behaviours (i.e. signalling storm) mitigation and prevention. In terms of scenarios, there are cases where it is difficult to troubleshoot and or understand whether the NFs themselves are behaving correctly or not. Often, they may not know t...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35.2 Procedures
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35.2.1 General
The NWDAF can provide information on network abnormal behaviours (i.e. signalling storm) as follows. Figure 6.35.2.1-1: Procedure for NWDAF-assisted Network Abnormal Behaviour Mitigation and Prevention 1. The consumer NF or e.g. MDAF/MDAS subscribes to or sends a request to NWDAF assistance information for the Abnormal...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35.2.2 Input Data
NFs can extract various information when acting as a consumer or producer. It is assumed to consider Input Data from a Producer NF perspective. The NWDAF collects signalling storm information as listed in Table 6.35.2.2-1 to 6.35.2.2-3. Table 6.35.2.2-1: Input data collected by NWDAF for AMF Information Source Descript...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35.2.3 Output Analytics
The output analytics of signalling storm statistics of NWDAF is defined in Table 6.35.2.3-1. Table 6.35.2.3-1: Signalling storm statistics Information Description Report (1..max) List of observed signalling storm. > Target NF ID The target of signalling storm detected by NWDAF. > Abnormality ID The potential cause of N...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35.2.4 Mitigation or Prevention
The following examples of existing mechanism can be used to mitigate and prevent the signaling storm. Table 6.35.2. 4-1: Example mechanisms to mitigate and prevent the signalling storm Abnormality ID Actions of NFs Massive UE access AMF sets MM NAS related timer (e.g. back-off, T3512) with suggested time range for a se...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.35.3 Impacts on services, entities and interfaces
NWDAF: - Supports providing subscriptions and/or requests for Abnormal NF signalling storm Analytics. - Supports deriving statistics and/or predictions of based on Analytics consumer subscriptions and/or requests. Consumer NF: - Supports subscribing or requesting Abnormal NF signalling storm Analytics from NWDAF using ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.36 Solution #36: Registration Signalling Analytics to support detection and prevention of signalling storm
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.36.1 Description
This solution is proposed to address Key Issue#4: NWDAF enhancements to support network abnormal behaviours (i.e. Signalling storm) detection and prevention. As pointed in use case #3, in some scenarios, UEs are possible to send small data over NAS signalling at same time leading to potential signalling storm. For exam...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.36.2 Procedures
Figure 6.36.2-1 illustrates the procedure for registration signalling analytics provided by NWDAF. Figure 6.36.2-1: Registration signalling analytics provided by NWDAF 1. The consumer NF subscribes to registration signalling analytics by invoking Nnwdaf_AnalyticsInfo or Nnwdaf_MLModelProvision_Subscribe service (Analyt...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.36.3 Impacts on services, entities and interfaces
NWDAF: - Supports of providing new analytics ID "Registration signalling". - Collects registration signalling information and timer information from AMF. - Collects activation time information from AF. AMF: - Provides registration signalling information and timer information. AF: - Provides activation time information.
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.37 Solution #37: Signalling storm detection and mitigation based on O&M data
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.37.1 Description
The proposed solution leverages O&M information about signalling storm (or, in 3GPP SA5 terminology "control plane congestion", "signalling congestion"), and based on that information, enables signalling storm prediction, detection, prevention and mitigation by taking into account the control plane congestion analysis ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.37.2 Procedures
Editor's note: This clause describes high-level procedures and information flows for the solution. The procedure of how NWDAF subscribes for the analytics information available at MDAF/MDAS defined by clause 6.2.14 of TS 23.288 [5], will be fully leveraged by this solution for the newly introduced analytics type of sig...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.37.3 Impacts on services, entities and interfaces
Editor's note: This clause captures impacts on existing services, entities and interfaces. NWDAF is enhanced to request Control Plane Congestion Analysis data/signalling storm analytics from MDAF/MDAS. AMF/SMF/PCF and the corresponding AMF-NWDAF, PCF-NWDAF and SMF-NWDAF interfaces are enhanced to subscribe/receive sign...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38 Solution #38: NWDAF-assisted signalling storm analytics, predictions, prevention and mitigation
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1 General description
This is a solution to Key Issue#4: NWDAF enhancements to support network abnormal behaviours (i.e. Signalling storm) mitigation and prevention. This solution includes two aspects: - NWDAF-assisted signalling storm detection: - Introduce a new Analytics ID, i.e. Analytics ID = Signalling Storm Detection, for the NWDAF t...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.1 NWDAF-assisted signalling storm detection
From the development process of the signalling storm in the above scenario, it can be seen that signalling storms may be caused by repeated requests from UEs. Therefore, whether a signalling storm occurs in the network can be predicted by analysing the statistical characteristics of the received signalling request and ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.1.1 Input Data
The total signalling received by the NFs and the request type should be collected to determine the statistical characteristic of the NF. And the finer granularity information such as signalling from UE should also be collected to determine whether the UE is abnormal. Table 6.38.1.1.1-1: Input data from 5GC for signalli...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.1.2 Output Data
The NWDAF provides analytics and/or predictions for signalling storm detection to the consumer, as defined in Table 6.38.3-1. Table 6.38.1.1.2-1: Signalling Storm Detection (Statistics/Predictions) Information Description Analytics Area List of TA(s) or Cell ID(s) within the requested area of interest. Time period list...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.1.3 NWDAF-assisted signalling storm analytics Procedure
Figure 6.38.1.1.3-1 depicts a procedure for signalling storm analytics/prediction service provided by NWDAF. Figure 6.38.1.1.3-1: Procedure for NWDAF providing signalling storm analytics and predictions 1. The NF Consumer and the SCP sends an Analytics request (Analytics ID = Signalling Storm Detection, Analytics Filte...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.2 NWDAF-assisted signalling storm prevention and mitigation
Based on the signalling storm analytics and predictions provided by the NWDAF containing AnLF and the request from the consumer, the ReLF generates the corresponding recommendation parameters or polices to assist the NF consumer to prevent or mitigate the signalling storm. The service consumer of signalling storm preve...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.2.1 Input Data
The total signalling received by the NFs and the request type should be collected to determine the statistical characteristic of the NF. And the finer granularity information such as signalling from UE should also be collected to determine whether the UE is abnormal. NF instance IDs of DNN/S-NSSAI should be collected b...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.2.2 Output Data
The NWDAF provides analytics and/or predictions for signalling storm detection to the consumer, as defined in Table 6.38.3-1. Table 6.38.1.2.2-1: Signalling Storm Recommendation Information Information Description NF instance ID Identification of the NF instance. Validity Period The validity period for the recommended ...
c8f50ee24a14933c33290076ea3625ad
23.700-84
6.38.1.2.3 NWDAF-assisted signalling storm prevention and mitigation procedure
Figure 6.38.1.2.3-1 depicts a procedure for signalling storm recommendation provided by NWDAF containing ReLF. Figure 6.38.1.2.3-1: Procedure for NWDAF-assisted signalling storm prevention and mitigation 1. The NF consumer and the SCP sends a Recommendation request/subscribe (Recommendation ID = Signalling Storm Recomm...