Content for TR 23.700-80 Word version: 18.0.0
5 Key Issues p. 13
5.1 Key Issue #1: Monitoring of network resource utilization for support of Application AI/ML operations p. 13
5.2 Key Issue #2: 5GC information exposure to UE p. 13
5.3 Key Issue #3: 5GC Information Exposure to authorized 3rd party for Application Layer AI / ML Operation p. 14
5.4 Key Issue #4: Enhancing External Parameter Provisioning p. 14
5.5 Key Issue #5: 5GC Enhancements to enable Application AI/ML Traffic Transport p. 14
5.6 Key Issue #6: QoS and Policy enhancements p. 15
5.7 Key Issue #7: 5GS Assistance to Federated Learning Operation p. 15
6 Solutions p. 17
6.0 Mapping Solutions to Key Issues p. 17
6.1 Solution #1: 5GS Monitoring Capabilities for AI/ML-based Services p. 19
6.1.1 Description p. 19
6.1.2 Procedures p. 20
6.1.2.1 Procedure for QoS monitoring p. 20
6.1.2.2 Procedure for monitoring of session inactivity time and traffic volume events p. 22
6.1.3 Impacts on services, entities and interfaces p. 23
6.2 Solution #2: User Plane solution for 5GC information exposure to UE p. 23
6.2.1 Description p. 23
6.2.2 Procedures p. 24
6.2.2.1 General p. 24
6.2.2.2 UE ID retrieval p. 25
6.2.2.3 Network consent and Multiple IEAF p. 30
6.2.2.4 Using DCCF to deliver analytics for UEs p. 32
6.2.3 Impacts on services, entities and interfaces p. 33
6.3 Solution #3: 5GC information exposure to UE by SMF p. 34
6.3.1 Description p. 34
6.3.2 Procedures p. 36
6.3.2.1 Procedure for enabling analytics information exposure to UE p. 36
6.3.2.2 Procedure for disabling analytics information exposure to UE p. 39
6.3.3 Impacts on services, entities and interfaces p. 40
6.4 Solution #4: 5GC Analytics exposure to UE p. 40
6.4.1 Description p. 40
6.4.2 Procedures p. 41
6.4.3 Impacts on services, entities and interfaces p. 41
6.5 Solution #5: 5GC information exposure to UE p. 42
6.5.1 Description p. 42
6.5.2 Procedures p. 44
6.5.3 Impacts on services, entities and interfaces p. 45
6.6 Solution #6: Federated learning analytics as assistance to AI/ML application server p. 46
6.6.1 Description p. 46
6.6.2 Procedures p. 47
6.6.3 Impacts on services, entities and interfaces p. 48
6.7 Solution #7: 5GS information exposure for AI/ML operation p. 49
6.7.1 Description p. 49
6.7.2 Procedures p. 49
6.7.2.1 Data rate monitoring scheme based on AF session with required QoS p. 49
6.7.2.2 Data rate monitoring scheme based on service information provision p. 51
6.7.3 Impacts on services, entities and interfaces p. 51
6.8 Solution #8: 5GC information/analytics notification to the AF and the UE p. 51
6.8.1 Introduction p. 51
6.8.2 Description p. 52
6.8.3 Procedures p. 53
6.8.3.1 AF Subscription for 5GC information or analytics p. 53
6.8.3.2 5GC information or analytics notification to the AF and the UE p. 55
6.8.4 Impacts on services, entities and interfaces p. 56
6.9 Solution #9: AF influence on AI/ML operations related policies p. 57
6.9.1 Description p. 57
6.9.2 Procedures p. 57
6.9.3 Impacts on services, entities and interfaces p. 60
6.10 Solution #10: Application Data Transfer to support planned and event driven AI/ML traffic transport p. 60
6.10.1 Description p. 60
6.10.2 Solutions p. 61
6.10.2.1 ADT Mechanism to support Application AI/ML data transfer p. 62
6.10.2.2 ADT Negotiation for future PDU session to support Application AI/ML data transfer p. 63
6.10.2.3 Activation or Updating/Removing ADT Transfer to a PDU Session p. 65
6.10.2.4 NWDAF Analytics support for ADT warning notification p. 66
6.10.3 Impacts on services, entities and interfaces p. 67
6.11 Solution #11: Traffic routing enhancements for Application AIML Traffic Transport p. 67
6.11.1 Description p. 67
6.11.2 Procedures p. 68
6.11.3 Impacts on services, entities and interfaces p. 69
6.12 Solution #12: Enhanced charging for Application AIML traffic transport p. 69
6.12.1 Description p. 69
6.12.2 Procedures p. 69
6.12.3 Impacts on services, entities and interfaces p. 71
6.13 Solution #13: Support Application AI/ML Traffic Transport p. 72
6.13.1 Description p. 72
6.13.2 Procedures p. 73
6.13.3 Impacts on services, entities and interfaces p. 74
6.14 Solution #14: Flock QoS based Federated Learning Operation p. 74
6.15 Solution #15: QoS monitoring for AIML application p. 75
6.15.1 Description p. 75
6.15.2 Procedures p. 76
6.15.3 Impacts on services, entities and interfaces p. 77
6.16 Solution #16: Solution for supporting the aggregated QoS parameters based on the Group-MBR p. 77
6.16.1 Description p. 77
6.16.2 Procedures p. 80
6.16.2.1 Defining Group-MBR Traffic Monitoring Parameter p. 80
6.16.2.2 Extensions to AF Influence Information Element in AF request to monitor and report aggregate bit rate among the specific group of PDU sessions p. 80
6.16.2.3 Extensions to AF Influence Network Function Service Procedure to enable the support for Group-MBR monitoring p. 81
6.16.2.4 Extensions to AF Influence on traffic routing procedure for Session not identified by an UE address to enable the support for Group-MBR monitoring p. 82
6.16.2.5 The procedure of Group-MBR monitoring p. 83
6.16.2.6 Extension to Nnef_EventExposure Services for Event Monitoring to support Group-MBR monitoring p. 84
6.16.3 Impacts on services, entities and interfaces p. 85
6.17 Solution #17: 5GS assistance for FL member selection using QoS monitoring p. 85
6.17.1 Description p. 85
6.17.2 Procedures p. 86
6.17.3 Impacts on services, entities and interfaces p. 87
6.18 Solution #18: 5GS assistance for FL member selection based on UE's visited AOI information p. 87
6.18.1 Description p. 88
6.18.2 Procedures p. 90
6.18.3 Impacts on services, entities and interfaces p. 91
6.19 Solution #19: 5G assisted FL member selection p. 91
6.19.1 Description p. 91
6.19.2 Procedures p. 92
6.19.3 Impacts on services, entities and interfaces p. 93
6.20 Solution #20: 5GS assistance to reduce the latency divergence p. 93
6.20.1 Description p. 93
6.20.2 Procedures p. 95
6.20.3 Impacts on services, entities and interfaces p. 96
6.21 Solution #21: Exposure of UE performance or a group of UEs performance p. 96
6.21.1 Description p. 96
6.21.2 Procedures p. 97
6.21.3 Impacts on services, entities and interfaces p. 98
6.22 Solution #22: 5GC-assisted Federated Learning over 5G VN group p. 98
6.22.1 Description p. 98
6.22.2 Procedures p. 98
6.22.3 Impacts on services, entities and interfaces p. 100
6.23 Solution #23: Federated Learning Assistance Function assisting on federated learning members selection p. 100
6.23.1 Description p. 100
6.23.2 Procedures p. 101
6.23.3 Impacts on services, entities and interfaces p. 103
6.24 Solution #24: 5GS Assistance to Federated Learning p. 104
6.24.1 Description p. 104
6.24.2 Procedures p. 104
6.24.3 Impacts on services, entities and interfaces p. 105
6.25 Solution #25: Providing 5GS Assistance Information to AF for Federated Learning Operation p. 105
6.25.1 Description p. 105
6.25.2 Procedures p. 106
6.25.3 Impacts on services, entities and interfaces p. 109
6.26 Solution #26: FL operation with 5GS coordination for UE selection p. 110
6.26.1 Description p. 110
6.26.2 Procedures p. 110
6.26.3 Impacts on Existing Nodes and Functionality p. 111
6.27 Solution #27: Assistance to selection of UEs for federated learning operation p. 111
6.27.1 Description p. 111
6.27.2 Procedures p. 112
6.27.3 Impacts on services, entities and interfaces p. 113
6.28 Solution #28: Application AI/ML Assistance Services p. 113
6.28.1 Description p. 113
6.28.2 AaaML Service Architecture p. 116
6.28.3 AaaML Services p. 117
6.28.4 Procedures for AaaML Services p. 119
6.28.5 Impacts on services, entities and interfaces p. 120
6.29 Solution #29: 5GC assistance to UE on performing AI/ML task p. 120
6.29.1 Description p. 120
6.29.2 Procedures p. 121
6.29.2.1 Procedure for AF provisioning information to 5GC p. 121
6.29.2.2 Procedure for 5GC exposing information to UE p. 121
6.29.2.3 Procedure for 5GC exposing information to UE via UE policy p. 122
6.29.3 Impacts on Existing Nodes and Functionality p. 124
6.30 Solution #30: AI/ML Translator (AI/ML-T) for 5GC Information Exposure to UE p. 125
6.30.1 Description p. 125
6.30.2 Procedures p. 126
6.30.3 Impacts on Existing Nodes and Functionality p. 127
6.31 Solution #31: Network information exposure to UE p. 127
6.31.1 Description p. 127
6.31.2 Procedures p. 128
6.31.3 Impacts on Existing Nodes and Functionality p. 129
6.32 Solution #32: 5GC Information Exposure to Authorized 3rd Party for Assisting AI/ML Operation Splitting between AI/ML Endpoints p. 129
6.32.1 Description p. 129
6.32.2 Procedures p. 131
6.32.3 Impacts on services, entities and interfaces p. 132
6.33 Solution #33: 5GC Exposure of Network Authorization for UE to participate in the Application AI/ML operation p. 133
6.33.1 Description p. 133
6.33.2 Procedure p. 134
6.33.2.1 Procedure for 5GC Exposure of Network Authorization Status of UE for Application AI/ML operation p. 134
6.33.2.2 New Subscription Data of Authorized Application AI/ML services p. 135
6.33.3 Impacts on existing services, entities and interfaces p. 136
6.34 Solution #34: Expose AI/ML traffic transmission status to the authorized 3rd party for Application Layer AI/ML Operation p. 137
6.34.1 Description p. 137
6.34.2 Procedures p. 137
6.34.3 Impacts on services, entities and interfaces p. 139
6.35 Solution #35: External parameter provisioning by AF for AI/ML data transport p. 139
6.35.1 Description p. 139
6.35.2 Procedures p. 140
6.35.3 Impacts on services, entities and interfaces p. 140
6.36 Solution #36: Enhancing External Parameter Provisioning p. 140
6.36.1 Description p. 140
6.36.2 Procedures p. 141
6.36.3 Impacts on services, entities and interfaces p. 143
6.37 Solution #37: Solution for 5GC Assistance to support Group-MBR Monitoring p. 143
6.37.1 Description p. 143
6.37.2 Procedures p. 145
6.37.2.1 Defining Group-MBR Traffic Monitoring Parameter p. 145
6.37.2.2 Extensions to AF Influence Information Element in AF request to monitor and report aggregate bit rate among the specific group of PDU sessions p. 146
6.37.2.3 Extensions to AF Influence Network Function Service Procedure to enable the support for Group-MBR monitoring p. 146
6.37.2.4 Extensions to UPF Services to support Group-MBR monitoring p. 147
6.37.2.5 Extensions to AF Influence on traffic routing negotiation procedure to enable the group service provisioning and the activation for Group-MBR monitoring p. 148
6.37.2.6 The procedure of Group-MBR monitoring p. 150
6.37.3 Impacts on services, entities and interfaces p. 151
6.38 Solution #38: Time dependent QoS for ML model distribution in federated learning p. 151
6.38.1 Description p. 151
6.38.2 Procedures p. 152
6.38.2.1 Procedure based on setting up an AF session for QoS p. 152
6.38.2.2 Procedure for UEs' staggering with notification to the AI/ML application server p. 154
6.38.3 Impacts on services, entities and interfaces p. 156
6.39 Solution #39: 5GC information assistance to FL members' selection p. 156
6.39.1 Description p. 156
6.39.2 Procedures p. 158
6.39.3 Impacts on services, entities and interfaces p. 161
6.40 Solution #40: Flexible Federated Learning Operation over 5GS p. 161
6.40.1 Description p. 161
6.40.1.1 Overview p. 161
6.40.1.2 Allocation of an AMSID to the UE p. 162
6.40.2 Procedures p. 163
6.40.2.1 Procedure for group QoS request for FL by AF p. 163
6.40.2.2 Procedure for enabling FL operation configuration recommendation by AFLSF p. 164
6.40.2.3 Procedure for UE-initiated request to join an AI/ML session for FL p. 166
6.40.2.3.1 Using a session management request p. 166
6.40.2.3.2 Using an application layer request p. 167
6.40.2.4 Procedure for multiple PCF discovery p. 168
6.40.3 Impacts on services, entities and interfaces p. 168
6.41 Solution #41: Solution for supporting aggregated UE performance monitoring and exposure for a group of UEs p. 169
6.41.1 Description p. 169
6.41.2 Procedures p. 169
6.41.3 Impacts on services, entities and interfaces p. 169
6.42 Solution #42: FL operation support by 5GS based on AF session with required QoS provided by Application Server p. 170
6.42.1 Description p. 170
6.42.2 Procedures p. 171
6.42.3 Impacts on services, entities and interfaces p. 173
6.43 Solution #43: FL member selection p. 173
6.43.1 Description p. 173
6.43.2 Procedures p. 173
6.43.3 Impacts on services, entities and interfaces p. 174
6.44 Solution #44: 5G assisted FL member selection for application request p. 174
6.44.1 Description p. 174
6.44.2 Procedures p. 175
6.44.3 Impacts on services, entities and interfaces p. 175
6.45 Solution #45: Improving FL performance by selecting central application server with 5GC's assistance p. 176
6.45.1 Description p. 176
6.45.2 Procedures p. 177
6.45.3 Impacts on services, entities and interfaces p. 178
6.46 Solution #46: 5GS assistance for FL member selection based on UE's location and direction p. 179
6.46.1 Description p. 179
6.46.2 Procedures p. 180
6.46.3 Impacts on services, entities and interfaces p. 181
6.47 Solution #47: Composite exposure service for 5GS assistance to FL member selection p. 182
6.47.1 Description p. 182
6.47.2 Procedures p. 182
6.47.3 Impacts on Existing Nodes and Functionality p. 183
7 Evaluation p. 183
7.1 Key Issue #1: Monitoring of network resource utilization for support of Application AI/ML operations p. 183
7.2 Key Issue #2: 5GC information exposure to UE p. 188
7.3 Key Issue #3: 5GC Information Exposure to authorized 3rd party for Application Layer AI/ML Operation p. 196
7.4 Key Issue #4: Enhancing External Parameter Provisioning p. 199
7.5 Key Issue #5: 5GC Enhancements to enable Application AI/ML Traffic Transport p. 200
7.6 Key Issue #6: QoS and Policy enhancements p. 202
7.6.1 QoS performance measurement assistance to Application AI/ML operation p. 202
7.6.2 Monitoring Group-MBR p. 204
7.6.3 Other topics p. 205
7.7 Key Issue #7: 5GS Assistance to Federated Learning Operation p. 205
7.8 Usage of analytics p. 209
8 Conclusion p. 214
8.0 General Architecture Principles p. 214
8.1 Key Issue #1: Monitoring of network resource utilization for support of Application AI/ML operations p. 215
8.2 Key Issue #2: 5GC information exposure to UE p. 215
8.3 Key Issue #3: 5GC Information Exposure to authorized 3rd party for Application Layer AI/ML Operation p. 216
8.4 Key Issue #4: Enhancing External Parameter Provisioning p. 216
8.5 Key Issue #5: 5GC Enhancements to enable Application AI/ML Traffic Transport p. 216
8.6 Key Issue #6: QoS and Policy enhancements p. 217
8.6.1 QoS performance measurement assistance to Application AI/ML operation p. 217
8.6.2 QoS request for a group of UEs p. 218
8.6.3 Performance KPIs mapping to 5GS QoS parameters p. 218
8.7 Key Issue #7: 5GS Assistance to Federated Learning Operation p. 218
$ Change history p. 220
