[{"name":"R5-260435","title":"Discussion on Enabling AI ML UAI Based Signalling Tests","source":"Qualcomm Incorporated","contact":"Yogesh Tugnawat","contact-id":63855,"tdoctype":"discussion","for":"Agreement","abstract":"The Rel 19 Work Item on Artificial Intelligence \/ Machine Learning (AI\/ML) for the NR air interface introduced several enhancements to the UE Assistance Information (UAI) signaling framework to support AI\/ML lifecycle management (LCM) based on UE initiated triggers. These enhancements enable the UE to indicate applicability status, data collection preferences, and resource related constraints to the network through UAI signaling.\nTable 1.1 summarizes the key functional areas from the scoped Rel 19 AI\/ML RAN5 signaling work plan [1] that rely on UE triggered UAI signaling to exercise and validate the specified procedures, including verification of the UE capability to correctly construct and transmit UAI messages.\nA substantial portion of the RAN5 signaling work plan [1] for Rel 19 depends on UE initiated triggering of UAI to validate call flows, signaling behavior, and UE conformance to the specified procedures. These mechanisms are expected to play a critical role in practical AI\/ML deployment scenarios. However, due to the UE autonomous nature of the triggers, there is currently no deterministic or reliable method within RAN5 testing scope to explicitly invoke these UAI procedures. This creates challenges for defining robust and repeatable test cases.\nTherefore, there is a need to discuss mechanisms that enable deterministic triggering of UE initiated UAI signaling for conformance testing purposes within RAN5 scope.\n2.2?Solution \u2013 Test Command\nTo address the challenges identified in Section 2.1, this discussion paper proposes the introduction of a single test command capable of triggering the full set of UAI procedures associated with Rel 19 AI\/ML functionality.\nBased on a set of logical indicators and associated parameters included in the SET AIML UAI REQUEST message, the test system can instruct the UE to trigger a specific UAI procedure, including any associated preferences, configurations, or reporting behavior. This approach allows multiple UAI scenarios to be exercised using a common test mechanism.\nObservations and Proposals\t\t\nObservation #1\nThere is currently no standardized mechanism within RAN5 scope to deterministically trigger UE initiated UAI signaling procedures for the scenarios outlined in Section 2.1.\nProposal #1\nBased on Observation #1, introduce a new test command in TS 38.509 [2] capable of triggering all relevant AI\/ML related UAI procedures required to support the Rel 19 RAN5 signal level test cases.\nObservation #2\nUE behavior upon receiving such a test command is not currently specified.\nProposal #2\nDefine UE requirements in TS 38.509 [2] for behavior upon receipt of the proposed test command, covering all applicable AI\/ML UAI trigger scenarios.\nIt is proposed, in principle, to introduce a new test command in TS 38.509 [2] to support deterministic triggering of AI\/ML related UAI procedures thus enabling Work Plan progress with unified approach of test case design. \nCorresponding CRs addressing Proposals #1 and #2 will be introduced and evaluated on their individual technical merits","secretary_remarks":"","agenda_item_sort_order":1088,"ainumber":"6.3.28.11","ainame":"Discussion Papers, Work Plan, TC lists","tdoc_agenda_sort_order":4350,"status":"noted","reservation_date":"2026-01-29 01:11:47","uploaded":"2026-02-13 10:41:08","revisionof":"","revisedto":"","release":"Rel-19","crspec":"","crspecversion":"","workitem":[{"winame":"NR_AIML_air-UEConTest"}],"crnumber":"","crrevision":"","crcategory":"","tsg_crp":"","lsreplyto":"","lsto":"","Cc":"","lsoriginalls":"","lsreply":"","link":"https:\/\/www.3gpp.org\/ftp\/tsg_ran\/WG5_Test_ex-T1\/TSGR5__110_Gothenburg\/Docs\/R5-260435.zip","group":"R5","meeting":"R5-110","year":2026,"uicc_affected":"","me_affected":"","ran_affected":"","cn_affected":"","clauses_affected":"","crsinpack":null,"crsinpacknumber":0},
{"name":"R5-260794","title":"Testability Aspects for AI\/ML Based Beam Management UE sided Models","source":"Nokia Corporation","contact":"Rupali Gupta","contact-id":108430,"tdoctype":"discussion","for":"Endorsement","abstract":"->4.5.4\nIn RAN#107, the revised SID of the Study on Artificial Intelligence (AI)\/Machine Learning (ML) for NR air interface Phase 2 was agreed [2]. In this revised SID, the following scope was agreed for RAN4 and RAN5. Hence, the WID [3] was revised accordingly.\nRecently in RAN5#109, the new AIML WI [8] has opened for discussing on UE conformance test aspects related to the Artificial Intelligence (AI)\/Machine Learning (ML) for NR Air Interface. The following are the objectives of this WI.\n...\nFollowing this in our contribution, we share our views on the AIML conformance test framework on the following topics:\n\u2022\tNew UE Special Conformance Test function for AIML use cases\nIn this contribution, we share our views on UE conformance testing framework on below topics:\n-\tNew UE Conformance Test Function for AI\/ML use cases\nThe following observations and proposals were made:\nObservation 1: The existing UE test functions are not sufficient to handle the Rel-19 AI\/ML use cases and modes.\nObservation 2: RAN1 has completed Rel-19 work and has consolidated the list of parameters [6] and there is an available Summary of UE features for AI\/ML for NR Air Interface [7].\nObservation 3: Offline training of UE models may introduce behaviours that are not directly controlled by the network operator. Therefore, appropriate conformance testing is required to ensure NW compliant UE behaviour.\nObservation 4: In case of handover, when model sharing is not possible, the network shall verify, through inference based conformance testing, that the UE\u2019s active model continues to operate correctly after handover.\nProposal 1: RAN5 to endorse a UE special conformance test function to cater AIML use cases from Rel-19 onwards.\nProposal 2: RAN5 to endorse a new UE special conformance test function to cater to BM UE sided model for LCM AIML-enabled functionalities.\nProposal 3: RAN5 to endorse a new AIML BM inference related test function.","secretary_remarks":"","agenda_item_sort_order":1088,"ainumber":"6.3.28.11","ainame":"Discussion Papers, Work Plan, TC lists","tdoc_agenda_sort_order":7940,"status":"noted","reservation_date":"2026-01-30 06:47:00","uploaded":"2026-01-30 06:56:26","revisionof":"","revisedto":"","release":"Rel-19","crspec":"","crspecversion":"","workitem":[{"winame":"NR_AIML_air-UEConTest"}],"crnumber":"","crrevision":"","crcategory":"","tsg_crp":"","lsreplyto":"","lsto":"","Cc":"","lsoriginalls":"","lsreply":"","link":"https:\/\/www.3gpp.org\/ftp\/tsg_ran\/WG5_Test_ex-T1\/TSGR5__110_Gothenburg\/Docs\/R5-260794.zip","group":"R5","meeting":"R5-110","year":2026,"uicc_affected":"","me_affected":"","ran_affected":"","cn_affected":"","clauses_affected":"","crsinpack":null,"crsinpacknumber":0}]