Method to handle inference configuration and information

Abstract

Techniques are provided for managing artificial intelligence and machine learning (AI / ML) configuration data and inference information in a wireless communication network during state transitions of a user equipment (UE). A UE performs a functionality configuration procedure with a source base station to establish one or more AI / ML functionalities, such as beam management or channel state information feedback. When the UE transitions to an inactive state or experiences a radio link failure, the AI / ML configurations—including inference information, inference configurations, and applicability reporting configurations—are handled through coordination between the UE, the source base station, and a target base station. During an RRC resume or RRC reestablishment procedure, the target base station retrieves the stored configurations from the source base station or requests them directly from the UE. The coordination ensures that AI / ML functionalities are accurately resumed, updated, or released based on the reply from the target base station.

Description

METHOD TO HANDLE INFERENCE CONFIGURATION AND INFORMATIONFIELD OF THE DISCLOSURE

[0001] This disclosure relates to wireless communications, and more particularly, to configuring artificial intelligence / machine learning (AI / ML) services.BACKGROUND

[0002] In order to carry out artificial intelligence / machine learning tasks (AI / ML) within wireless communication networks, it is proposed to coordinate communications relevant to such tasks between user equipments (UEs) and base stations of the network.

[0003] For example, during an applicable functionality reporting procedure, a UE may report applicable functionalities to AI / ML tasks the UE wishes to undertake. This may include transmitting to the network inference information. The network side may assess the appropriate resources and approach for achieving this functionality and may provide the UE with an inference configuration which allows the UE to configure itself for the AI / ML procedure.

[0004] However, at present there is not defined process to handle relevant information or configuration data when a UE enters an inactive state or a radio link failure occurs. For example, recovery from such states may follow an RRC Resume or RRC Reestablishment process and there is a need to define how the inference information and / or inference configuration may be retrieved / updated during such a process. For example, where a UE had been previously configured for AI / ML functionality in combination with a source base station, it may be possible to define subsequent interaction with a target base station.SUMMARY

[0005] According to a first aspect, there is provided a computer implemented method performed by a user equipment, UE, the method comprising: performing a functionality configuration procedure in communication with a source base station, the functionality configuration procedure resulting in at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with the source base station; receiving an RRCRelease message from thesource base station; entering an inactive state; transmitting an RRC Resume Request message to a target base station; receiving a reply to the RRCResumeRequest message from the target base station, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply.

[0006] Optionally, the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

[0007] Optionally, the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

[0008] Optionally, inference configuration comprises one or more configuration parameters to enable AI / ML such as channel state information, CSI, feedback, beam management, positioning, and mobility.

[0009] Optionally, inference applicability reporting configuration provides scheduling for the UE to report inference information.

[0010] Optionally, the method further comprise, after receiving the RRCRelease message, storing the at least one configuration UE.

[0011] Optionally, the reply to the RRCResumeRequest message indicates whether to configure AI / ML functionalities with the target base station using the at least one stored configuration.

[0012] Optionally, the reply to the RRCResumeRequest message indicates to use an updated configuration set to configure AI / ML funcitonalities with the target base station.

[0013] Optionally, the updated configuration comprises second inference configuration.

[0014] Optionally, the method further comprises: after receiving the reply to the RRCResumeRequest message from the target base station, transmitting to the target base station at least one of the at least one configuration.

[0015] Optionally, the method further comprises after receiving the reply to the RRCResumeRequest message from the target base station, transmitting to the target base station second inference information.

[0016] Optionally, a further message from the target base station indicates whether to configure AI / ML functionalities with the target base station using the at least one stored configuration.

[0017] Optionally, the further message is an RRCReconfiguration message.

[0018] Optionally, the further message indicates to use an updated configuration to configure AI / ML functionalities with the target base station

[0019] Optionally, the method further comprises, after receiving the RRCRelease message, releasing the at least one configuration at the UE.

[0020] Optionally, the method further comprises after receiving the reply to the RRCResumeRequest message from the target base station, performing a functionality configuration procedure in communication with the target base station, the functionality configuration procedure resulting in one or more further configurations for configuring AI / ML functionalities with the target base station.

[0021] Optionally, the reply to the RRCResumeRequest message from the target base station is an RRCResume message.

[0022] Optionally, the method further comprises, after receiving the RRCRelease message, storing the at least one configuration at the UE, and, after receiving the reply to the RRCResumeRequest message from the target base station, releasing the at least one configuration at the UE.

[0023] Optionally, the method further comprises, the reply to the RRCResumeRequest message from the target base station is an RRCSetup message.

[0024] Optionally, the method further comprises, after receiving the RRCRelease message, storing the at least one configuration at the UE, and after receiving the reply to the RRCResumeRequest message from the target base station, entering the inactive state.

[0025] Optionally, the reply to the RRCResumeRequest message from the target base station is an RRCReject message.

[0026] Optionally, the method further comprises, after receiving the RRCRelease message, storing the at least one configuration at the UE, and wherein the reply to the RRCResumeRequest message from the target base station is a RRCRelease message.

[0027] Optionally, the method further comprises, after receiving the RRCRelease message, releasing the at least one configuration at the UE.

[0028] According to a further aspect, there is provided a computer implemented method performed by a user equipment, UE, the method comprising: performing a functionality configuration procedure in communication with a source base station, the functionality configuration procedure resulting in at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with the source base station; transmitting an RRCReestablishmentRequest message to a target base station; receiving a reply to the RRCReestablishmentRequest message from the target base station, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply

[0029] Optionally, the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

[0030] Optionally, the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

[0031] Optionally, inference configuration comprises one or more configuration parameters to enable AI / ML for channel state information, CSI, feedback, beam management, positioning, and mobility.

[0032] Optionally, inference applicability reporting configuration provides scheduling for the UE to report inference information.

[0033] Optionally, the method further comprises, prior to transmitting the RRCReestablishmentRequest message, storing the at least one configuration at the UE.

[0034] Optionally, the method further comprises, receiving a message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the at least one configuration.

[0035] Optionally, the message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using at least one configuration indicates to use an updated configuration to configure AI / ML functionalities with the target base station.

[0036] Optionally, the updated configuration comprises second inference configuration.

[0037] Optionally, the message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the at least one configuration is an RRCReconfiguration message.

[0038] Optionally, the method further comprises, after receiving a reply to the RRCReestablishmentRequest message from the target base station, transmitting to the target base station at least one of the at least one configuration.

[0039] Optionally, the method further comprises, prior to the transmitting to the target base station at least one of the at least one configuration, receiving a message from the target base station requesting at least one of the at least one configuration.

[0040] Optionally, the message from the target base station requesting at least one of the at least one configuration is an RRCReconfiguration message.

[0041] Optionally, the method further comprises: after transmitting to the target base station at least one of the at least one configuration, receiving a message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the at least one stored configuration.

[0042] Optionally, the message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the one or more of the stored at least one configuration is an RRCReconfiguration message.

[0043] Optionally, the method further comprises, prior to transmitting the RRCReestablishmentRequest message, releasing the at least one configuration at the UE.

[0044] Optionally, the method further comprises, after receiving the reply to the RRCReestablishmentRequest message from the target base station, performing a functionality configuration procedure in communication with the target base station, the functionality configuration procedure resulting in at least one further configuration for configuring AI / ML functionalities with the target base station.

[0045] Optionally, the method further comprises, prior to transmitting the RRCReestablishmentRequest message, storing the at least one configuration at the UE, and after receiving a reply to the RRCReestablishmentRequest message from the target base station, releasing the at least one configuration at the UE.

[0046] Optionally, the reply to the RRCReestablishmentRequest message from the target base station is an RRCSetup message.

[0047] According to a further aspect, there is provided a computer implemented method performed by a target base station, T-BS, the method comprising: receiving an RRCResumeRequest message from a user equipment, UE, wherein UE being associated with at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with a source base station; transmitting a reply to the RRCResumeRequest message to the UE, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply.

[0048] Optionally, the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

[0049] Optionally, the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

[0050] Optionally, inference configuration comprises one or more configuration parameters to enable AI / ML such as channel state information, CSI, feedback, beam management, positioning, and mobility.

[0051] Optionally, inference applicability reporting configuration provides scheduling for the UE to report inference information.

[0052] Optionally, the method further comprises, responsive to receiving the RRCResumeRequest message, transmitting a RETRIEVE UE CONTEXT REQUEST to the source base station.

[0053] Optionally, the method further comprises, after transmitting the RETRIEVE UE CONTEXT REQUEST to the source base station, receiving a response from the source base station.

[0054] Optionally, the response to the RETRIEVE UE CONTEXT REQUEST is a RETRIEVE UE CONTEXT RESPONSE.

[0055] Optionally, the response to the RETRIEVE UE CONTEXT REQUEST comprises the at least one configuration, the at least one configuration being stored at the UE.

[0056] Optionally, the method further comprises, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / ML functionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

[0057] Optionally, the updated at least one configuration comprises a second inference configuration.

[0058] Optionally, after receiving the RRCResumeRequest message, the method further comprises transmitting a request for the at least one configuration to the UE.

[0059] Optionally, the method further comprises receiving the at least one configuration from the UE.

[0060] Optionally, the at least one configuration received from the UE comprises a first inference configuration.

[0061] Optionally, the method further comprises receiving first inference information and / or second inference information from the UE.

[0062] Optionally, the method further comprises, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / MLfunctionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

[0063] Optionally, the updated at least one configuration comprises a second inference configuration.

[0064] Optionally, the method further comprises, after transmitting the reply to the RRCResumeRequest message to the UE, performing a functionality configuration procedure in communication with the UE.

[0065] Optionally, the reply to the RRCResumeRequest message to the UE causes the UE to release the at least one configuration

[0066] Optionally, the reply to the RRCResumeRequest message is an RRCSetup message.

[0067] Optionally, the reply to the RRCResumeRequest message is an RRCReject message.

[0068] Optionally, the reply to the RRCResumeRequest message is an RRCRelease message.

[0069] According to a first aspect, there is provided a computer implemented method performed by a target base station, T-BS, the method comprising: receiving an RRCReestablishmentRequest message from a user equipment, UE, wherein UE being associated with at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with a source base station; transmitting a reply to the RRCReestablishmentRequest message to the UE, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply.

[0070] Optionally, the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

[0071] Optionally, the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

[0072] Optionally, inference configuration comprises one or more configuration parameters to enable AI / ML such as channel state information, CSI, feedback, beam management, positioning, and mobility.

[0073] Optionally, inference applicability reporting configuration provides scheduling for the UE to report inference information.

[0074] Optionally, the method further comprises, responsive to receiving the RRCReestablishmentRequest message, transmitting a RETRIEVE UE CONTEXT REQUEST to the source base station.

[0075] Optionally, the method further comprises, after transmitting the RETRIEVE UE CONTEXT REQUEST to the source base station, receiving a response from the source base station.

[0076] Optionally, the response to the RETRIEVE UE CONTEXT REQUEST is a RETRIEVE UE CONTEXT RESPONSE.

[0077] Optionally, the response to the RETRIEVE UE CONTEXT REQUEST comprises the at least one configuration, the at least one configuration being stored at the UE.

[0078] Optionally, the method further comprises, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / ML functionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

[0079] Optionally, the updated at least one configuration comprises a second inference configuration.

[0080] Optionally, after receiving the RRCReestablishmentRequest message, the method further comprises transmitting a request for the at least one configuration to the UE.

[0081] Optionally, the method further comprises receiving the at least one configuration from the UE.

[0082] Optionally, the at least one configuration received from the UE comprises a first inference configuration.

[0083] Optionally, the method further comprises receiving first inference information and / or second inference information from the UE.

[0084] Optionally, the method further comprises, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / ML functionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

[0085] Optionally, the updated at least one configuration comprises a second inference configuration.

[0086] Optionally, the method further comprises: after transmitting the reply to the RRCReestablishmentRequest message to the UE, performing a functionality configuration procedure in communication with the UE.

[0087] Optionally, the reply to the RRCReestablishmentRequest message to the UE causes the UE to release the at least one configuration

[0088] Optionally, the reply to the RRCReestablishmentRequest message is an RRCSetup message.

[0089] According a further aspect, there is provided a user equipment, UE, for performing the method of previous aspects.

[0090] According a further aspect, there is provided a base station for performing the method of previous aspects.BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Fig. 1 is a block diagram of an example system in which a base station and / or a user equipment (UE) can implement the techniques of this disclosure for managing Inference configuration and information in a UE;

[0092] Fig. 2 is a block diagram of an example protocol stack according to which the UE of Figs. 1 can communicate with base stations;

[0093] Fig. 3 illustrates an example scenario in which a BS configures an inference configuration while the UE is a connected mode, for a UE that may activate / deactivate AI / ML functionality, after receiving inference configuration from the BS.

[0094] Fig. 4A illustrates an example scenario in which a source BS configures an inference configuration while the LIE is a connected mode and a subsequent RRC resume procedure with a target BS.

[0095] Fig. 4B illustrates a scenario similar to that of Fig. 4A, but in which the base station releases stored inference configuration an inference information, after requesting the UE to enter an inactive mode;

[0096] Fig. 4C illustrates a scenario similar to that of Fig. 4A, but in which both UE and source BS release inference configuration and inference information;

[0097] Fig. 4D illustrates a scenario similar to that of Fig. 4B, but in which the target BS receives a RETRIEVE UE CONTECT FAILURE message from the source BS and subsequently transmits an RRCSetup message to the UE;

[0098] Fig. 4E illustrates a scenario similar to that of Fig. 4B, but in which the target BS transmits an RRCReject message to the UE;

[0099] Fig. 4F illustrates a scenario similar to that of Fig. 4B, but in which the target BS transmits an RRCRelease message to the UE;

[0100] Fig. 5A illustrates an example scenario in which a source BS configures an inference configuration while the UE is a connected mode and subsequently an RRC Reestablishment process is carried out with a target BS;

[0101] Fig. 5B illustrates a scenario similar to that of Fig. 5A, but in which the source base station that is unable to provide stored UE inference configuration and inference information to the target BS;

[0102] Fig. 5C illustrates a scenario similar to that of Fig. 5A, but in which UE releases inference configuration and inference information, after detecting a radio link failure;

[0103] Fig. 5D illustrates a scenario similar to that of Fig. 5A, but in which the target BS receives a RETRIEVE UE CONTECT FAILURE message from the source BS and subsequently transmits an RRCSetup message to the UE;

[0104] Fig. 6A is a flow diagram of an example method for UE receiving an inference configuration from a BS;

[0105] Fig. 6B is a flow diagram of an example method for BS transmitting an inference configuration to an UE;

[0106] Fig. 7A is a flow diagram of an example method for a UE performing RRC resume procedure and receives inference configuration from a target BS;

[0107] Fig. 7B is a flow diagram of an example method for a source BS receiving an RETRIEVE UE CONTEXT REQUEST from a target BS and transmitting an inference configuration and inference information to the target BS;

[0108] Fig. 7C is a flow diagram of an example method for a target BS receiving a RRC resume request from a UE and transmitting an inference configuration and inference information to the UE;

[0109] Fig. 7D is a flow diagram of an example method for a UE performing RRC resume procedure and transmitting stored inference configuration and inference information to a BS;

[0110] Fig. 7E is a flow diagram of an example method for a target BS performing a RRC resume procedure and receiving the stored inference configuration and inference information from the UE;

[0111] Fig. 7F is a flow diagram of an example method for a BS performing a RRC resume procedure and receives the stored inference configuration and inference information from a BS or an UE;

[0112] Fig. 7G is a flow diagram of an example method for a UE releasing the stored inference configuration and inference information, after entering an inactive mode;

[0113] Fig. 7H is a flow diagram of an example method for a UE receiving an RRCReject, RRCSetup or RRCRelease message after deciding to perform an RRC resume procedure.

[0114] Fig. 8A is a flow diagram of an example method for a UE performing a RRC re-establishment procedure and receiving an inference configuration from a target BS;

[0115] Fig. 8B is a flow diagram of an example method for a target BS receiving an RETRIEVE UE CONTEXT RESPONSE from a source BS including inference configuration and inference information;

[0116] Fig. 8C is a flow diagram of an example method for a UE performing a RRC re-establishment procedure and receiving an inference configuration / information request from a target BS;

[0117] Fig. 8D is a flow diagram of an example method for a target BS performing a RRC re-establishment procedure transmitting an inference configuration / information request to a UE;

[0118] Fig. 8E is a flow diagram of an example method for a target BS performing a RRC re-establishment procedure and receiving the stored inference configuration and inference information from either a source BS or an UE;

[0119] Fig. 8F is a flow diagram of an example method for a UE releasing the stored inference configuration and inference information, after detecting a radio link failure;

[0120] Fig. 8G is a flow diagram of an example method for a UE releasing or keeps the stored inference configuration and inference information, after a conditional LTM or CHO mobility;

[0121] Fig. 8H is a flow diagram of an example method for a UE performing a RRC re-establishment procedure and receiving an RRCSetup message from a target BS; and

[0122] Fig. 8I is a flow diagram of an example method for a BS identifying that a UE is out of sync and storing first inference information and / or first inference configuration.DETAILED DESCRIPTION

[0123] The following description is directed to certain implementations for the purpose of describing innovative aspects of this disclosure. However, a person having ordinary skill in the art will readily recognize that the teachings herein can be applied in a multitude of different ways. Some of the examples in this disclosure are based on wireless communication according to the 3rd Generation Partnership Project (3GPP) wireless standards, such as ambient internet-of- things (A-loT), the4th generation (4G) Long Term Evolution (LTE) and 5th generation (5G) New Radio (NR) standards. However, the described techniques can be implemented in any device, system, or network that is capable of transmitting and receiving radio frequency signals according to any of the wireless communication standards, including any of the Institute of Electrical and Electronics Engineers (IEEE) 802.11 or 802.16 wireless standards, or other known signals that are used to communicate within a wireless, cellular, or loT network, such as a system utilizing 4G, 5G, 6th generation (6G), ZigBee, Bluetooth, WiFi, or future radio technology.

[0124] When performing AI / ML tasks in a wireless communication context, a UE may need to coordinate its behaviour with the base station to which it is connected. In some examples, a UE engages in an applicable functionality reporting procedure to provide the base station with an indication of the applicable functionalities which must be supported. In some examples, a proactive reporting schedule may be provided (for example, the UE reports applicable functionalities based on an existing network configuration and when any change occurs the UE updates this) but in others a reactive reporting schedule is used (for example, the UE reports applicable functionalities as requested by the network).

[0125] It is desirable for the network to configure connectivity of the UE appropriately given the applicable functionalities that are needed. For example, the network should engage in beam management, scheduling and so forth. In some examples, in reporting applicable functionalities (or otherwise) the UE transmits inference information to the base station. In examples, the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML procedures, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

[0126] In general, the inference information provides sufficient information for the network to identify an appropriate inference configuration. The inference configuration comprises one or more configuration parameters to enable AI / ML such as channel state information, CSI, feedback, beam management, positioning, and mobility. The base station transmits the inference configuration to the UE so the UE may configure itself appropriately for a subsequent AI / ML procedure. Subsequently,the UE may provide further inference information, and the network / base station will determine whether to update the inference configuration in response.

[0127] In addition to the above, in examples the base station transmits (inference) applicable reporting configuration to the UE. The applicable reporting configuration provides scheduling for the UE to report (further) inference information.

[0128] There may be circumstances in which the UE enters an inactive state, perhaps on instruction from the base station. In such circumstances, it is important that the UE and base station can effectively coordinate operation in relation to the inference configuration (and associated information) when the UE resumes connectivity (through an RRC Resume process, for example). The following disclosure provides approaches to handle such circumstances. For example, a source base station may release the UE so that it enters and inactive state, and at a later time the UE may send a resume request to a target base station.

[0129] Similarly, there may be circumstances in which there is a radio link failure between the UE and a source base station. Should the UE subsequently look to reestablish connection with a target base station (for example, through an RRC Reestablishment procedure), it is again important to handle the inference configuration and associated information appropriately and in a coordinated manner and the present disclosure provides solutions to these challenges.

[0130] In the examples described below, inference information, inference configuration and applicable reporting configuration are examples of AI / ML configurations which may in general be used for the configuration of AI / ML functionalities and associated processes. While the techniques are described in the context of these examples, the skilled person will recognize that alternative AI / ML configurations may be handled in an equivalent manner.

[0131] FIG. 1 is a block diagram of an example wireless communication system 100 in which a network entity may perform RRC Resume and / or RRC Reestablishment procedures. The wireless communication system 100 includes a UE 102, a first network entity (e.g., base station (BS) 104), a second network entity (e.g., base station 106), and a core network (CN) 110. The UE 102 may initially connect to the base station 104. In some scenarios, the base station 104 may configure another base station (e.g., base station 106) as a secondary node (SN) and may configurethe UE 102 to operate in dual connectivity (DC) with the base station 104 and the base station 106. The base stations 104 and 106 may operate as a master node (MN) and an SN for the UE 102, respectively.

[0132] In addition to base stations 104 and 106, a network entity may be an Evolved Universal Terrestrial Radio Access Network Node B (E-UTRAN Node B), evolved Node B (eNodeB or eNB), Next Generation Node B (gNodeB or gNB), Next Generation E-UTRAN Node B (ng-eNB), access point, radio head, or the like. The network entity may be implemented in a macrocell, microcell, small cell, picocell, or the like, or any combination thereof. In some aspects, a base station such as base station 104 and / or base station 106 may be a monolithic base station in which the functionality of the base station is implemented as a single unit. In some other aspects, a base station may be a distributed base station in which the functionality of the base station may be distributed among two or more units such as a CU and one or more DUs.

[0133] In various configurations of the wireless communication system 100, the base station 104 may be implemented as a master eNB (MeNB) or a master gNB (MgNB), and the base station 106 may be implemented as a secondary gNB (SgNB). In some aspects, the UE 102 may communicate with the base station 104 and the base station 106 via the same radio access technology (RAT) such as evolved universal terrestrial radio access (EUTRA) or new radio (NR). In some other aspects, the UE 102 may communicate with the base station 104 and the base station 106 via different RATs. When the base station 104 is an MeNB and the base station 106 is a SgNB, the UE 102 may be in EUTRA-NR DC (EN-DC) with the MeNB and the SgNB.

[0134] In some aspects, an MeNB or an SeNB may be implemented as an ng-eNB. When the base station 104 is a master ng-eNB (Mng-eNB) and the base station 106 is a SgNB, the UE 102 may be in next generation (NG) EUTRA-NR DC (NGEN-DC) with the Mng-eNB and the SgNB. When the base station 104 is an MgNB and the base station 106 is an SgNB, the UE 102 may be in NR-NR DC (NR-DC) with the MgNB and the SgNB. When the base station 104 is an MgNB and the base station 106 is a Secondary ng-eNB (Sng-eNB), the UE 102 may be in NR-EUTRA DC (NE-DC) with the MgNB and the Sng-eNB.

[0135] In the scenarios where the UE 102 hands over from the base station 104 to the base station 106, the base stations 104 and 106 operate as a source base station (S-BS) and a target base station (T-BS), respectively. The UE 102 may operate in DC with the base station 104 and an additional base station (not shown in Fig. 1) for example prior to the handover. The UE 102 may continue to operate in DC with the base station 106 and the additional base station or operate in single connectivity (SC) with the base station 106 after completing the handover. The base stations 104 and 106 in this case operate as a source MN (S-MN) and a target MN (T-MN), respectively.

[0136] The CN 110 may be an evolved packet core (EPC) 111 or a fifth-generation core (5GC) 160, both of which are depicted in the example of Fig. 1. The base station 104 may be an eNB supporting an S1 interface for communicating with the EPC 111, an ng-eNB supporting an NG interface for communicating with the 5GC 160, or a gNB that supports an NR radio interface as well as an NG interface for communicating with the 5GC 160. To directly exchange messages with each other during the scenarios discussed below, the base stations 104 and 106 may support an X2 or Xn interface. Among other components, the EPC 111 may include a serving gateway (SGW) 112, a mobility management entity (MME) 114, and a packet data network gateway (PGW) 116. The SGW 112 is generally configured to transfer userplane packets related to audio calls, video calls, Internet traffic, and the like. The MME 114 is configured to manage authentication, registration, paging, and other related functions. The PGW 116 provides connectivity from the UE to one or more external packet data networks, e.g., an Internet network and / or an Internet Protocol (IP) multimedia subsystem (IMS) network. The 5GC 160 includes a user plane function (UPF) 162 and an access and mobility management function (AMF) 163, and / or session management function (SMF) 166. The UPF 162 is generally configured to transfer user-plane packets related to audio calls, video calls, Internet traffic, and the like. The AMF 163 is configured to manage authentication, registration, paging, and other related functions. The SMF 166 is configured to manage protocol data unit (PDU) sessions.

[0137] In the example shown in Fig. 1, the base station 104 supports cell 124A, and the base station 106 supports a cell 126A. The cells 124A and 126A can partially overlap, so that the UE 102 can communicate in DC with the base station 104 andthe base station 106, where one of the base stations 104 and 106 is an MN and the other is an SN. The base station 104 can support additional cell(s) such as cells 124B and 124C, and the base station 106 can support additional cell(s) (not shown in Fig. 1). The cells 124A, 124B, and 124C can partially overlap, so that the UE 102 can communicate in carrier aggregation (CA) with the base station 104. The base station 104 can operate the cells 124A, 124B, and 124C via one or more transmission and reception points (TRPs, not shown in Fig. 1). More particularly, when the UE 102 is in DC with the base station 104 and the base station 106, one of the base stations 104 and 106 operates as an MeNB, an Mng-eNB, or an MgNB, and the other operates as an SgNB or an Sng-eNB.

[0138] In general, the wireless communication system 100 may include any suitable number of base stations supporting NR cells and / or EUTRA cells. More particularly, the EPC 111 or the 5GC 160 may be connected to any suitable number of base stations supporting NR cells and / or EUTRA cells. Although the examples below refer specifically to specific CN types (EPC, 5GC) and RAT types (5G NR and EUTRA), in general the techniques of this disclosure also may apply to other suitable radio access and / or core network technologies such as sixth generation (6G) radio access and / or 6G core network or 5G NR-6G DC.

[0139] The base station 104 is equipped with processing hardware 130 that can include one or more general-purpose processors (e.g., CPUs) and a non-transitory computer-readable memory storing instructions that the one or more general-purpose processors execute. Additionally or alternatively, the processing hardware 130 can include special-purpose processing units. The processing hardware 130 can include a physical (PHY) controller 132 configured to transmit data and control signal on physical downlink (DL) channels and DL reference signals with one or more user devices (e.g. UE 102) via one or more cells (e.g., the cells 124A, 124B, and / or 124C) and / or one or more TRPs. The PHY controller 132 is also configured to receive data and control signal on physical uplink (UL) channels and / or UL reference signals with the one or more user devices via one or more cells (e.g., the cells 124A, 124B, and / or 124C) and / or one or more TRPs. The processing hardware 130, in an example implementation, includes a medium access control (MAC) controller 134 configured to perform MAC functions with one or more user devices. The MAC functions include a random access (RA) procedure, managing UL timing advance(TA) for the one or more user devices, and / or communicating UL / DL MAC PDUs with the one or more user devices. In some aspects, the MAC functions may include AI / ML related functions described herein. For example, the AI / ML related functions include one or more AI / ML functionalities for beam management and / or CSI feedback. The processing hardware 130 can further include a radio resource control (RRC) controller 136 to implement procedures and messaging at the RRC sublayer of the protocol communication stack. For example, the RRC controller 136 may be configured to support RRC messaging associated with AI / ML configuration, handover procedures, and / or to support the necessary operations when the base station 104 operates as an MN relative to an SN or as an SN relative to an MN. The base station 106 can include processing hardware 140 that is similar to processing hardware 130. In particular, components 142, 144, 146 of base station 106 may be similar to the components 132, 134, 136, and 137, respectively, of base station 104.

[0140] The UE 102 is equipped with processing hardware 150 that can include one or more general-purpose processors such as CPUs and non-transitory computer-readable memory storing machine-readable instructions executable on the one or more general-purpose processors and / or special-purpose processing units. The PHY controller 152 is configured to receive data and control signals on physical DL channels and / or DL reference signals with the base station 104 or 106 via one or more cells (e.g., the cells 124A, 124B, 124C, and / or 126A) and / or one or more TRPs. The PHY controller 152 is also configured to transmit data and control signals on physical UL channels and / or UL reference signals with the base station 104 or 106 via one or more cells (e.g., the cells 124A, 124B, 124C, and / or 126A) and / or one or more TRPs. The processing hardware 150 in an example implementation includes a MAC controller 154 configured to perform MAC functions with base station 104 or 106. For example, the MAC functions may include a random access procedure, managing UL timing advance, and communicating UL / DL MAC PDUs with the base station 104 or 106. In some aspects, the MAC functions may include AI / ML related functions. The processing hardware 150 may further include an RRC controller 156 to implement procedures and messaging at the RRC sublayer of the protocol communication stack. For example, the RRC controller 156 may be configured to support RRC messaging associated with AI / ML configuration, handover procedures, and / or to support the necessary operations when the UE 102 communicates in dualconnectivity with an MN and an SN. Although described as being included with processing hardware 150, some of the functionality described above may be implemented as firmware or in software modules (e.g., software modules stored in the computer-readable memory.

[0141] In operation, the UE 102 in DC can use a radio bearer to communicate with a base station such as base station 104 or base station 106. Generally speaking, the UE and a base station can use signaling radio bearers (SRBs) to exchange RRC messages as well as nonaccess stratum (NAS) messages. The UE 102 and the base station 104 or 106 can use data radio bearers (DRBs) to transport data on a user plane.

[0142] UEs may use several types of SRBs and DRBs. When operating in dual connectivity (DC), the cells associated with a base station operating as an MN define a master cell group (MCG), and the cells associated with a base station operating as a secondary node (SN) define a secondary cell group (SCG). SRB1 resources may carry RRC messages, which in some cases may include NAS messages exchanged (e.g., transmitted and / or received) over the dedicated control channel (DCCH). SRB2 resources support RRC messages that may include logged measurement information or NAS messages. SRB2 resources may also be exchanged over the DCCH but with lower priority than SRB1 resources. SRB1 and SRB2 resources allow the UE and the MN to exchange RRC messages related to the MN and embed RRC messages related to the SN. These SRB1 and SRB2 messages may be referred to as MCG SRBs. The UE and the SN may use SRB3 resources to exchange RRC messages related to the SN. These SRB3 resources may be referred to as SCG SRBs. SRBs used by the UE to exchange RRC messages directly with the MN via lower layer resources of both the MN and the SN may be referred to as split SRBs. Similarly, DRBs using the lower-layer resources of only the MN can be referred to as MCG DRBs. DRBs using the lower-layer resources of only the SN can be referred to as SCG DRBs. DRBs using the lower-layer resources of both the MCG and the SCG can be referred to as split DRBs.

[0143] In the example shown in Fig. 1 , the UE 102 operating in DC can use a radio bearer (e.g., a DRB or an SRB) that at different times terminates at the base station 104 operating as an MN or the base station 106 operating as an SN. The UE 102 canapply one or more security keys when communicating on the radio bearer, in the uplink (UL) direction (e.g., from the UE 102 to a base station) and / or downlink direction (e.g., from a base station to the UE 102).

[0144] Fig. 2 illustrates, in a simplified manner, an example protocol stack 200 according to which the UE 102 can communicate with an eNB / ng-eNB, a gNB or a 6G BS 232 (e.g., one or more of the base stations 104, 106).

[0145] In the example stack 200, an physical layer (PHY) 202 of provides transport channels to the MAC sublayer 204, which in turn provides logical channels to the NR RLC sublayer 206. The RLC sublayer 206 in turn provides RLC channels to an NR PDCP sublayer 210. The PDCP sublayer 210 in turn can provide data transfer services to Service Data Adaptation Protocol (SDAP) 212 or a radio resource control (RRC) sublayer (not shown in Fig. 2).

[0146] The PDCP sublayer 210 receives packets (e.g., from an Internet Protocol (IP) layer, layered directly or indirectly over the PDCP layer 210) that can be referred to as service data units (SDUs), and output packets (e.g., to the RLC layer 206) that can be referred to as protocol data units (PDUs). Except where the difference between SDUs and PDUs is relevant, this disclosure for simplicity refers to both SDUs and PDUs as “packets.”

[0147] On a control plane, the PDCP sublayer 210 can provide signaling radio bearers (SRBs) or RRC sublayer (not shown in Fig. 2) to exchange RRC messages or non-access-stratum (NAS) messages, for example. On a user plane, the PDCP sublayer 210 can provide Data Radio Bearers (DRBs) to support data exchange. Data exchanged on the PDCP sublayer 210 can be SDAP PDUs, Internet Protocol (IP) packets or Ethernet packets.

[0148] Corresponding reference numerals are used within the multiple figures referred to be a the same numeral (so that the same reference numeral signifies the same event in all of Figs. 4A to 4F, all of Figs. 5A to 5D, all of Figs. 7A to 7H and all of Figs. 8A to 8I). Moreover, generally speaking, equivalent events in Figs. 3-8ID are labeled with similar reference numbers (e.g., event 350 in Fig. 3 is similar to event 450 in Figs. 4A-4F and event 550 in Figs. 5A-5D; event 418A in Fig. 4A is similar to event 418B-418F in Figs. 4B-4F and event 518 in Figs. 5A-5D) with differences discussed below where appropriate. With the exception of the differences shown inthe figures and discussed below, any of the alternative implementations discussed with respect to a particular event (e.g., for messaging and processing) may apply to events labeled with similar reference numbers in other figures and also to both integrated and distributed base stations.

[0149] Now referring first to Fig 3, in a scenario 300, the UE initially operates 302 in a connected mode with a base station (BS) 104 of a RAN 105. While a UE is in connected mode, the UE 102 may receive 304 a UECapabilityEnquiry message from a BS 104. In response to the UECapabilityEnquiry message, the UE transmits 306 a UECapabilitylnformation message to the BS 104. In the UECapabilitylnformation message, the UE includes inference capability / capabilities indicating support one or more AI / ML functionalities. To simplify the following description, “capability” is used to represent “capability / capabilities”. Alternatively, the BS 104 receives 307 the inference capability from CN 110 or another BS instead of the UE 102.

[0150] The BS 104 then transmits 308 an RRCReconfiguration message to the UE 102. In the RRCReconfiguration message, the BS 104 includes at least one of a first inference configuration or an inference applicability reporting configuration (or called functionality applicability reporting configuration). In response, the UE 102 then transmits an RCReconfigurationComplete message to the BS 104. The first inference configuration configures and / or activates an AI / ML functionality. Such a functionality may be understood to include the applicable functionality reporting configuration. In some implementations, the Al functionality refers to a 3GPP feature or feature group for which Al (e.g., an Al model) is enabled. In some implementations, the 3GPP features or feature groups includes one or more communication functions such as CSI feedback, beam management, and / or positioning. In some implementations, the Al functionality is one of the following Al functionalities: Spatial-frequency domain CSI compression using Al; Time domain CSI prediction; Spatial-domain Downlink beam prediction for Set A of beams based on measurement results of Set B of beams; Temporal Downlink beam prediction for Set A of beams based on the historic measurement results of Set B of beams; Direct AI / ML positioning; and / or AI / ML assisted positioning.

[0151] When the RRCReconfiguration message includes the first inference configuration, the UE 102 may thus configure itself accordingly for the AI / MLfunctionality. However, it may be desirable to provide further information to confirm the appropriate configuration. As such, in some examples, the UE 102 then transmits 310 a functionality applicability report including first inference information to the BS 104 - this process is part of the AI / ML functionality as a whole. In some implementation, the first inference information indicates that the AI / ML functionality is applicable. In other implementations, the first inference information indicates that the AI / ML functionality is not applicable. In response to the functionality applicability report, the BS 104 may transmit 312 an RRCReconfiguration message which includes a second inference configuration to the UE 102 if necessary. The BS 104 may generate the second inference configuration based on the first inference information. If the first inference information indicates that the AI / ML functionality is applicable, the BS 104 may generate the second inference configuration configuring or activating the AI / ML functionality. In some implementations, the second inference configuration is same as the first inference configuration if the BS 104 determines so. In other implementations, the BS 104 omits to transmit further inference configuration (i.e. , event 314 is omitted) so that the UE 102 remains using the first inference configuration previously received. In some implementations, the BS 104 does not include the first inference configuration in the 308 RRCReconfiguration message. Instead, the BS 104 transmits 312 an inference configuration (e.g., the second inference configuration) to the UE 102 after (e.g., in response to) receiving 310 the functionality applicability report. In some implementations, the UE 102 activates 314 the AI / ML functionality in accordance with or based on the first or second inference configuration after (e.g., in response to) receiving the first or second inference configuration. In some other implementations, the BS 104 may transmit an activation command to the UE 102 in response to receiving the first inference information (e.g., event 312). The activation command may be a MAC control element or a DL control information (DCI). The UE 102 activates 314 the AI / ML functionality in accordance with or based on the first or second inference configuration after (e.g., in response to) receiving the activation command.

[0152] In general, an updated configuration may include updates to at least one of the inference configuration, inference information and / or applicability reporting information. Through processes described below, the target BS and UE arrive at astate where both understand whether an updated configuration will be used in place of the configuration initially created for the AI / ML functionality.

[0153] In some implementations, the UE 102 activates a first Al model forthe Al functionality in response to activating the Al functionality. In some implementations, after (e.g., in response to) activating the Al functionality or the first Al model, the UE 102 performs inference for the Al functionality using the first Al model. In some implementations, the Al functionality uses two-sided Al model. In such cases, the BS 104 activates a second Al model to communicate with the UE 102 in event 314 after transmitting the first or second inference configuration or the activation command. In some implementations, after (e.g., in response to) activating the Al functionality or the second Al model forthe UE 102, the BS 104 uses the second Al mode to perform inference for the Al functionality.

[0154] In some implementations, if the first inference information indicates that the AI / ML functionality is not applicable, the BS 104 may generate the second inference configuration releasing the first inference configuration or deactivating the AI / ML functionality. The UE 102 deactivates 314 the AI / ML functionality in response to receiving the second inference configuration. If the second inference configuration indicates releasing the first inference configuration, the UE 102 releases the first inference configuration. In other implementations, if the first inference information indicates that the AI / ML functionality is not applicable, the BS 104 may transmit a deactivation command to the UE 102, deactivating the AI / ML functionality. The UE 102 deactivates the AI / ML functionality in response to receiving the deactivation command. In some implementations, the deactivation command is a MAC CE ora DCI. In some implementations, the UE 102 deactivates the first Al model or stops using the first Al model in response to deactivating the AI / ML functionality. If the BS 104 activates the second Al model as described above, the BS 104 deactivates the second Al model or stops using the second Al model forthe UE 102 after (e.g., in response to) transmitting the deactivation command or the second inference configuration releasing the first inference configuration.

[0155] In some examples, the UE 102 may perform event 312 N times (N is equal or larger than 1). In response to each of the N times, the BS 104 may transmit an inference configuration (e.g., similar to the first inference configuration or the secondinference configuration), an activation command ora deactivation command to the UE 102 as described above.

[0156] The events 301 , 302, 304, 306, 307, 308, 310, 312 and 314 are collectively referred to in Fig. 3 as an AI / ML functionality configuration procedure 350.

[0157] Figs. 4A-4F are message sequences of example scenarios in which a UE performs RRC resume procedure after the AI / ML functionality configuration procedure.

[0158] Referring first to Fig 4A, in a scenario 400A, the UE 102 initially operates 450 an AI / ML functionality configuration procedure. The S-BS 104 then 416 decides to transition the UE to an inactive state. In response to this decision, the S-BS 104 stores 418A the first inference information, the first inference configuration, and / or the first inference applicability reporting configuration. In response to the decision 416, the S-BS 104 transmits 420 an RRCRelease message to the UE 102. After (e.g., in response to) receiving the RRCRelease, the UE 102 stores 422 the first inference information, the first inference configuration and / or the first inference applicability reporting configuration. In response to receiving the RRCRelease message, the UE 102 enters 424 an inactive state from the connected state.

[0159] Later in time, the UE 102 in the inactive state decides to 426 initiate an RRC resume procedure on a cell operated by the T-BS 106. In response to this decision, the UE 102 transmits 428 an RRCResumeRequest message to the T-BS 106. After (e.g., in response to) receiving the RRCResumeRequest message, the T-BS 106 transmits 430 a RETRIEVE UE CONTEXT REQUEST message to the BS 104 to retrieve a UE context of the UE 102. The T-BS 106 then receives 432A a RETRIEVE UE CONTEXT RESPONSE message from the S-BS 104 in response. The RETRIEVE UE CONTEXT RESPONSE includes the first inference information and / or the first inference configuration. In some implementations, the S-BS 104 includes additional configurations (e.g., non-AI configurations) of the UE 102 in the RETRIEVE UE CONTEXT RESPONSE. Before receiving the RRCRelease message, the UE 102 and S-BS 104 communicates with each other using the additional configurations. The UE 102 stores the additional configurations after (e.g., in response to) receiving the RRCRelease message or entering the inactive state. In response to receiving the first inference information and / or the first inference configuration, the T-BS 106 thenincludes the first inference configuration or a second inference configuration in an RRCResume message and transmits 434A an RRCResume message to the UE 102. In some implementations, the additional configurations includes physical layer configurations, MAC layer configurations, RLC layer configurations, PDCP layer configurations, SRB configuration(s), DRB configuration(s), and / or measurement configuration(s).

[0160] After receiving the RRCResume message, the UE 102 applies the first or the second inference configuration as indicated. In one implementation, the second inference configuration is provided as a delta configuration showing changes with respect to the first inference configuration. For example, in some implementations the second inference configuration augments the first inference configuration. The UE 102 then applies the second inference configuration to augment the stored first inference configuration. In another implementation, the second inference is not a delta configuration of the first inference configuration (i.e. , a full configuration may be provided). The UE 102 then applies the second inference configuration and releases the stored first inference configuration. In some implementations, the T-BS 106 includes an indication in the RRCResume message to indicate the second inference configuration is a delta configuration or not. In one implementation, the T-BS 106 does not include the second inference configuration in the RRCResume message and includes an indication in the RRC resume message to indicate the UE 102 to apply the stored first inference configuration. In another implementation, the UE 102 applies stored first inference configuration when the receiving RRCResume message does not include an inference configuration.

[0161] After receiving the 434A RRCResume message, the UE 102 transmits 436A an RRCResumeComplete message to the T-BS 106. The UE 102 and the T-BS 106 then are able to 414 activate / deactivate the AI / ML functionality with the first or second inference configuration as described for Fig. 3. In some implementations, the UE 102 and the T-BS 106 communicate with each other in accordance with at least a portion of the additional configurations. In the RRCResume message, the T-BS 106 include new configurations to update the remaining portion of the additional configurations. The UE 102 and the T-BS 106 communicate with each other in accordance with the new configurations. In some implementations, the newconfigurations include at least one of physical layer configurations and / or MAC layer configurations.

[0162] In one implementation, the RRCResumeComplete includes second inference information. In another implementation, the LIE 102 transmits 438 a functionality applicability report that includes a second inference information to the T-BS 106 after transmitting the RRCResumeComplete. The T-BS 106 then transmits 440 an RRCReconfiguration message that includes a third inference configuration to the UE 102. The UE 102 then applies the third inference configuration and transmits 442 an RRCReconfigurationComplete message to the T-BS 106. After this, the UE 102 and the T-BS 106 are able to 444 activate / deactivate the AI / ML functionality with the third inference. In one implementation, the UE 102 and the T-BS 106 performs event 414A first. After event 438 and 440, the UE 102 and the T-BS 106 perform event 444.

[0163] In one implementation, the T-BS 106 does not include the inference configuration (i.e. , the first or the second inference configuration) in the RRCResume message. Instead, the T-BS 106 includes the inference configuration in the 440 RRCReconfiguration message.

[0164] In some implementations, the UE 102 suspends or deactivates the AI / ML functionality after (i.e., in response to) entering the inactive state. The UE 102 resumes or activates the AI / ML functionality after receiving 434A the RRCResume message, transmitting 436A the RRCResumeComplete message, receiving 440 the RRCReconfiguration message, or after transmitting 442 the RRCReconfigurationComplete message. In another implementation, the T-BS 106 includes a field in RRCResume message or RRCReconfiguration message to indicate the UE 102 to resume or active the AI / ML functionality with the stored inference configuration. Suspending or deactivating the AI / ML functionality in this manner, together with the approaches to resuming such functionality applies equally to the examples of Figs. 4B to 4F described below.

[0165] Referring now to Fig. 4B, a scenario 400B is similar to 400A and similarly involves a S-BS 104, a T-BS 106, and UE 102 activating / deactivating AI / ML functionality and perform RRC resume procedure. However, unlike scenario 400A, the S-BS 104 may release the first inference information and / or first inferenceconfiguration after transmitting (e.g., in response to) an RRCRelease message to the UE 102. The differences between Fig. 4A and 4B are described below.

[0166] In response to the decision 416, the S-BS 104 releases 419 the first inference information, the first inference configuration, and / or the first inference applicability reporting configuration, unlike event 418A of Fig. 4A. .

[0167] In response to the RETRIEVE UE CONTEXT RESPONSE message, the S-BS 104 transmits 432B, to the T-BS 106, a RETRIEVE UE CONTEXT RESPONSE message that does not include inference information and inference configuration (since in the example of Fig. 4B this information has been released by the S-BS 104 prior to this step, unlike in the example of Fig. 4A). In response to this, the T-BS 106 transmits 434B an RRCResume message that includes a specific field (e.g., inference information and / or configuration (Infoconfig) indication (e.g., a request indicator)) to the UE 102 to request the stored inference information and / or inference configuration from the UE 102. In response to the inference Infoconfig request, the UE 102 transmits 436B an RRCResumeComplete message that includes the second inference information and / or the first inference configuration to the T-BS 106. The second inference information represents the currently applicable inference information for the UE 102; if there is no relevant change between the status of the UE 102 at this stage and during the functionality configuration procedure 450, the second inference information and the first inference information may be the same.

[0168] After receiving the RRCResumeComplete message, the T-BS 106 transmits 446B an RRCReconfiguration message that includes the first inference configuration or a second inference configuration to the UE 102. The second inference configuration is as described for Fig. 4A. After receiving the RRCReconfiguration message, the UE 102 applies the first or the second inference configuration. The UE 102 then transmits 448 an RRCReconfigurationComplete message to the T-BS 106. The UE 102 and the T-BS 106 then are able to 414 activate / deactivate the AI / ML functionality with the first or second inference configuration as described for Fig. 4A. In one implementation, the UE 102 applies the first inference configuration after / during the RRC resume procedure. In some implementations, the T-BS 106 does not transmit the first inference configuration to the UE 102 in the RRCReconfiguration message. Instead, the T-BS 106 may include a specific filed(e.g., an Usestoredinference field) in the RRCReconfiguration message or the RRCResume message to indicate the UE 102 to use the stored inference configuration. The UE 102 applies the first inference configuration in response to receiving the field.

[0169] Referring now to Fig. 4C, a scenario 400C is similar to 400B and similarly involves a S-BS 104, a target BS 106, and a UE 102 activating / deactivating AI / ML functionality and perform a RRC resume procedure. However, unlike scenario 400B, the UE 102 may release the first inference information and / or first inference configuration after receiving an RRCRelease message.

[0170] Prior to event 420 (the S-BS 104 transmitting the RRCRelease message to the UE 102) the method of Fig. 4C proceeds as in the method of Figure 4B. In response to the RRCRelease message 420, the UE 102 releases 423 the first inference information, the first inference configuration, and / or the first inference applicability reporting configuration.

[0171] The method then proceeds as Fig. 4B until T-BS 106 transmits 434C an RRCResume message to the UE 102. In response to receiving the RRCResume message, the UE 102 transmits 436C an RRCResumeComplete message to the UE 102. The T-BS 106 and the UE 102 then initiate an AI / ML functionality configuration procedure 451 , which proceeds in line with the procedure 450. The procedure 451 provides the UE 102 with an appropriate inference configuration and / or applicability reporting configuration which can be utilized in activating AI / ML functionality, similar to the procedure 350.

[0172] Referring now to Fig. 4D, processing up to and including the T-BS 106 transmitting the RETRIEVE UE CONTEXT REQUEST at step 430 proceed in line with the process of Fig. 4B, although it is noted that in this case step 419 may be considered optional. The S-BS 104 transmits 432D, to the T-BS 106, a RETRIEVE UE CONTEXT FAILURE message that does not include inference information or inference configuration.

[0173] After (e.g., in response to) receiving the RETRIEVE UE CONTEXT FAILURE message, the T-BS 106 transmits 435 an RRCSetup message to the UE 102. In response to receiving the RRCSetup message, the UE 102 releases 423 the first inference information, the first inference configuration and / or first inferenceapplicability reporting configuration. The UE 102 then transmits 437 RRCSetupComplete message to the T-BS 106. In one implementation, the UE 102 deactivates the AI / ML functionality after (e.g., in response to) receiving the RRCSetup message. In response to receiving the RRCSetup message, the UE 102 enters 401 the connected state. At a later stage, an AI / ML functionality configuration procedure may be carried out in line with the process 300 described in relation to Fig.3 in order to facilitate AI / ML functionality.

[0174] In some implementations, the RETRIEVE UE CONTEXT FAILURE message indicates no UE context found for the UE 102. In some other implementations, the S-BS 104 transmits, to the T-BS 106, a RETRIEVE UE CONTEXT RESPONSE message that does not include inference information or inference configuration, instead of the RETRIEVE UE CONTEXT FAILURE message. The RETRIEVE UE CONTEXT RESPONSE message replaces the RETRIEVE UE CONTEXT FAILURE message described above.

[0175] Referring now to Fig. 4E, processing up to and including the T-BS 106 transmitting the RETRIEVE UE CONTEXT REQUEST at step 430 proceed in line with the process of Fig. 4B, although it is noted that in this case step 419 may be considered optional. In response, the S-BS 104 transmits 432D, to the T-BS 106, the RETRIEVE UE CONTEXT F Al LI RE message.

[0176] After (e.g., in response to) receiving the RETRIEVE UE CONTEXT FAILURE message, the T-BS 106 then transmits 452 an RRCReject message to the UE 102. In response to receiving the RRCReject message, the UE 102 stays in the inactive state 424. In some implementations, the T-BS 106 and the S-BS 104 do not perform the events 430 and 432D. In such example, the T-BS 106 transmits the RRCReject message to the UE 102 directly after it receives the RRCResumeRequest message from the UE 102 (e.g., due to network congestion).

[0177] Referring now to Fig. 4F, processing up to and including the T-BS 106 transmitting the RETRIEVE UE CONTEXT REQUEST at step 430 proceed in line with the process of Fig. 4B, although it is noted that in this case step 419 may be considered optional. In response, the S-BS 104 transmits 432D, to the T-BS 106, the RETRIEVE UE CONTEXT FAILURE message.

[0178] After (e.g. in response to) receiving the RETRIEVE UE CONTEXT FAILURE message, the T-BS 106 then transmits 421 an RRCRelease message to the UE 102. After receiving the RRCRelease message, the UE 102 releases 423 the first inference information, the first inference configuration and / or the first inference applicability reporting configuration. In response to receiving the RRCRelease message, the UE 102425 enters an inactive state or an idle state. In one implementation, the UE 102 releases the first inference information, the first inference configuration, and / or the first inference applicability reporting configuration while or in response to entering idle state in 425. In another implementation, the UE 102 does not release the first inference information and the first inference configuration and the first inference applicability reporting configuration while or in response to entering inactive state in event 425.

[0179] Next. Figs. 5A-5D are message sequences of example scenarios in which a UE performs RRC connection re-establishment procedure after the AI / ML functionality configuration procedure. Generally speaking, equivalent events in Figs.5A-5D are labeled with similar reference numbers (e.g., event 532A in Fig. 5A is similar to event 532B, 532C and 532D in Figs. 5B, 5C and 5D) with differences discussed below where appropriate. With the exception of the differences shown in the figures and discussed below, any of the alternative implementations discussed with respect to a particular event (e.g., for messaging and processing) may apply to events labeled with similar reference numbers in other figures and also to both integrated and distributed base stations.

[0180] Referring first to Fig 5A, in a scenario 500A, the UE initially operates 550 an AI / ML functionality configuration procedure. The UE 102 then detects 516 a radio link failure and decides 518 to initiate a RRC connection re-establishment procedure. In response to the decision, the UE 102 stores 522 the first inference information, the first inference configuration, and / or the first inference applicability reporting configuration. In some implementations, the UE 102 also deactivates 552 the AI / ML functionality in response to the decision or the initiation of the RRC connection reestablishment procedure. In response to the detection 516 and / or the decision 518, the UE 102 transmits 528 an RRCReestablishmentRequest message to a T-BS 106. Note, the deactivation 552 can occur before or after event 528.

[0181] After receiving the RRCReestablishmentRequest, the T-BS 106 transmits 530 a RETRIEVE UE CONTEXT REQUEST message to the S-BS 104. The T-BS 106 then receives 532A an RETRIEVE UE CONTEXT RESPONSE message from the S-BS 104. The RETRIEVE UE CONTEXT RESPONSE includes at least a first inference information and / or a first inference configuration. After receiving the RETRIEVE UE CONTEXT RESPONSE, the T-BS 106 then transmits 534 an RRCReestablishment message to the UE 102 in response to the RRCReestablishmentRequest mesage. In response to receiving the RRCReestablishment message, the UE 102 transmits 536 an RRCReestablishmentComplete message to the T-BS 106.

[0182] After (e.g., in response to) receiving the RRCReestablishmentComplete message, receiving the RRCReestablishmentRequest message, or transmitting the RRCReestablishment message, the T-BS 106 then transmits 546A a RRCReconfiguration message that includes a first inference configuration or a second inference configuration to the UE 102. After receiving the RRCReconfiguration message, the UE 102 applies the inference configuration (the first or second inference configuration) and transmits 548A an RRCReconfigurationComplete message to the T-BS 106. The UE 102 and the T-BS 106 then are able to 514 activate / deactivate the AI / ML functionality with the first or the second inference configuration as appropriate. In one example, the UE 102 and / or the T-BS 106 activate the AI / ML functionality with the first inference configuration after successfully performing the RRC reestablishment procedure (i.e. , after successfully transmit / receive the RRCReestablishmentComplete message). The UE 102 includes a field in RRCReestablishmentComplete message to indicate the first inference configuration is active.

[0183] In one implementation, the UE 102 transmits 538 a functionality applicability report that includes second inference information to the T-BS 106 after it transmits the RRCReestablishmentComplete message or RRCReconfigurationComplete message or receives the RRCReconfiguration message. In some implementations, the UE 102 includes the functionality applicability report in the RRCReconfigurationComplete message. In other implementations, the UE 102 includes the functionality applicability report or second inference information in a UEAssistancelnformation message and transmits the UEAssistancelnformationmessage to the T-BS 106. The T-BS 106 then transmits 540A an RRCReconfiguration message that includes a third inference configuration to the UE 102. In some implementations, the T-BS 106 generates the third inference configuration based on the functionality applicability report or the second inference information. The UE 102 then applies the third inference configuration and transmits 542 an RRCReconfigurationComplete message to the T-BS 106 (this message may explicitly or implicitly confirm the application of the third inference configuration). After this, the UE 102 and the T-BS 106 are able to 544 activate / deactivate the AI / ML functionality with the third inference configuration.

[0184] In some implementations, the UE 102 activates the AI / ML functionality after it receives 534 the RRCReestablishment message or after it transmits 536 the RRCReestablishmentComplete message or after it receives 540A the RRCReconfiguration message or after it transmits 542 the RRCReconfigurationComplete message. In another implementation, the T-BS 106 includes a field in RRCReestablishment message or RRCReconfiguration message to request the UE 102 to activate the AI / ML functionality with the stored inference configuration.

[0185] While the examples illustrated in Fig. 5A to 5B refer to a radio link failure, it will be recognized that in other implementations, instead of radio link failure, the UE 102 initiates a RRC connection re-establishment procedure due to re-configuration with sync failure, integrity check failure, RRC reconfiguration failure, SCG change failure, or T316 expiration. An example set of reasons for initiating such a procedure which may apply in this context are set out in the document TS 38.331 at section 5.3.7.2.

[0186] Referring now to Fig. 5B, a scenario 500B is similar to 500A and similarly involves S-BS 104, T-BS 106, and UE 102 activating / deactivating AI / ML functionality and performing an RRC reestablishment procedure. However, unlike scenario 500A, the S-BS 104 may not transmit the first inference information and / or first inference configuration to the T-BS 106.

[0187] In response to the detection 516, the decision 518, or initiating the RRC connection re-establishment procedure, the UE 102 stores 522 the first inferenceinformation, the first inference configuration, and / or the first inference applicability reporting configuration.

[0188] After receiving the RRCReestablishmentRequest, the T-BS 106 transmits 530 the RETRIEVE UE CONTEXT REQUEST to the S-BS 104. The T-BS 106 then receives 532B an RETRIEVE UE CONTEXT RESPONSE from the S-BS 104. In the example of Fig. 5B, the RETRIEVE UE CONTEXT RESPONSE includes neither the first inference information nor the first inference configuration.

[0189] The T-BS 106 then transmits 546B an RRCReconfiguration message to the UE 102. In the RRCReconfiguration message, the T-BS 106 includes a field to request the stored inference information and / or stored inference configuration form the UE 102 (e.g., an Inference Infoconfig request field). After receiving the RRCReconfiguration, the UE 102 includes the second inference information and / or first inference configuration in an RRCReconfigurationComplete message and transmits 548B the RRCReconfigurationComplete message to the T-BS 106. The second inference information represents the currently applicable inference information for the UE 102; if there is no relevant change between the status of the UE 102 at this stage and during the functionality configuration procedure 550, the second inference information and the first inference information may be the same.

[0190] In some implementations, the UE 102 does not include the second inference information in the RRCReconfigurationComplete message. Instead, the UE 102 includes the second inference information in a functionality applicability report message and transmits 538 the functionality applicability report message to the T-BS 106. In response to the receiving RRCReconfigurationComplete and / or the functionality applicability report, the T-BS 106 includes first or second inference configuration in an RRCReconfiguration message and transmits 540B the RRCReconfiguration message to the UE 102. The UE 102 then applies the first or the second inference configuration and transmits 542 an RRCReconfigurationComplete message to the T-BS 106. After this, the UE 102 and the T-BS 106 are able to 514 activate / deactivate the AI / ML functionality with the first or the second inference configuration.

[0191] In some implementations, the UE 102 deactivates or suspends 552 the AI / ML functionality subsequent to initiating the RRC connection re-establishmentprocedure. In such examples, the T-BS 106 may not include the inference request field in the 546B RRCReconfiguration message but instead includes a field in the RRCReconfiguration message or the RRCReestablishment message to request the UE 102 to activate the stored AI / ML functionality which had been deactivated or suspended at step 552. In another implementation, the UE 102 activates the AI / ML functionality after it receives the RRCReestablishment message or after it transmits the RRCReestablishmentComplete message. In some examples, the T-BS 106 activates the AI / ML functionality after it receives the RRCReestablishmentComplete message. In one example, the UE 102 includes a field in the RRCReestablishmentComplete message to indicate that it has activated the AI / ML functionality with the stored inference configuration / information.

[0192] Referring now to Fig. 5C, a scenario 500C is similar to 500A and 500B and similarly involves a base station 104, target base station 106, and UE 102 activating / deactivating AI / ML functionality and perform a RRC connection reestablishment procedure. However, unlike scenario 500B, the UE 102 may release the first inference information and / or first inference configuration after it detecting a radio link failure or after it decides to perform an RRC connection re-establishment procedure.

[0193] In response to the decision 516, the detection 518, or initiating the RRC connection re-establishment procedure, the UE 102 releases 523 the first inference information, first inference configuration, and / or the first inference applicability reporting configuration. The UE 102552 deactivates the AI / ML functionality in response to the decision 516, the detection 518, initiating the RRC connection reestablishment procedure, or the releasing 523.

[0194] In response to the RETRIEVE UE CONTEXT REQUEST message, the T-BS 106 then receives 532C a RETRIEVE UE CONTEXT RESPONSE message from the S-BS 104. In this case, the RETRIEVE UE CONTEXT RESPONSE includes neither the first inference information nor the first inference configuration. The T-BS 106 then transmits 534 an RRCReestablishment message to the UE 102. After completing the RRC connection re-establishment procedure, the T-BS 106 and the UE 102 then perform an AI / ML functionality configuration procedure 551, which issimilar to the procedure 550. The inference configuration in 551 may the same as 550 or not.

[0195] Referring now to Fig. 5D, the procedure up to and including the T-BS 106 transmitting the RETRIEVE UE CONTEXT REQUEST at step 530 proceeds in line with the process of Fig. 5B. The T-BS 106 then receives 532D an RETRIEVE UE CONTEXT FAILURE message from the S-BS 104. In the example of Fig. 5D, the RETRIEVE UE CONTEXT FAILURE message may or may not include the first inference information nor the first inference configuration.

[0196] After receiving the RETRIEVE UE CONTEXT FAILURE message, the T-BS 106 transmits 535 an RRCSetup message to the UE 102. In response to the RRCSetup message, the UE 102 then releases 523 the first inference information and / or the first inference configuration. In one implementation, the UE 102 does not deactivate the AI / ML functionality before transmitting the RRCReestablishmentRequest message or in response to initiating the RRC connection re-establishment procedure. Instead, the UE 102 deactivates the AI / ML functionality after (e.g., in response to) receiving the RRCSetup message. In response to the RRCSetup message, the UE 102 transmits 537 an RRCSetupComplete message and enters 501 a connected state.

[0197] Referring now to Fig. 6A, a method 600A can be implemented in a suitable UE and includes for receiving a command to configure the AI / ML functionality. For clarity, the method 600A is discussed with specific reference to the BS 104 and the UE 102. The method 600A may carry out UE procedures of the method 300 shown in Figure 3.

[0198] At block 602, the UE 102 communicates with a BS 104 in a connected mode (e.g., event 302 of Fig. 3). At block 604, the UE 102 receives a UECapabilityEnquiry message from the BS 104 (e.g., event 304 of Fig. 3). Then, at block 606, the UE 102 includes an inference capability in a UECapabilitylnformation message and transmit it to the BS 104 (e.g., event 306 of Fig. 3). In one example, the inference capability includes at least an inference information. At block 608, the UE 102 receives an RRCReconfiguration message that includes a first inference configuration and / or an inference applicability reporting configuration (e.g., event 308 of Fig. 3). Then, at block 610, the UE 102 transmits an Applicability functionality reporting message thatincludes a first inference information to the BS 104 (e.g., event 310 of Fig. 3). In one implementation, the UE 102 transmits the Applicability functionality reporting while the said inference capability in 606 does not include an inference configuration. At block 612, the UE 102 receives an RRCReconfiguration message that includes a second inference configuration (e.g., event 312 of Fig. 3). Then, at block 614, the UE 102 is able to activate / deactivate the AI / ML functionality (e.g., event 314 of Fig. 3).

[0199] Referring now to Fig. 6B, a method 600B can be implemented in a base station in a RAN and includes for transmitting a command to configure the AI / ML functionality. For clarity, the method 600B is discussed with specific reference to the RAN 105, BS 104 and the UE 102. The method 600B may carry out BS 104 procedures of the method 300 shown in Figure 3.

[0200] At block 616, the BS 104 communicates with a UE 102 in a connected mode (e.g., event 302 of Fig. 3). At block 618, the BS 104 decides to configure UE an AL / ML functionality. In response to this decision, in block 620, the BS 104 transmits an UECapabilityEnquiry message to the BS 104 (e.g., event 304 of Fig. 3). Then, at block 622, the BS 104 receives an UECapabilitylnformation message that includes an inference capability (e.g., event 306 of Fig. 3). In one example, the inference capability includes at least an inference information. At block 624, the BS 104 transmit an RRCReconfiguration message that includes a first inference configuration and / or an inference applicability reporting configuration (e.g., event 308 of Fig. 3).

[0201] At block 626, the BS 104 receives an Applicability functionality reporting message that includes a first inference information (e.g., event 310 of Fig. 3). At block 628, the BS 104 transmits an RRCReconfiguration message that includes a second inference configuration to the UE 102 (e.g., event 312 of Fig. 3). Then, at block 614, the BS 104 is able to activate / deactivate the AI / ML functionality with the UE 102 (e.g., event 314 of Fig. 3).

[0202] Referring now to Fig. 7A, a method 700A can be implemented in a suitable UE and includes a process for handling an RRCRelease message while it is configured with the AI / ML functionality. For clarity, the method 700A is discussed with specific reference to the S-BS 104, T-BS 106 and the UE 102. The method 700A may implement UE procedures of the method 400A shown in Fig. 4A.

[0203] At block 750, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 720, the UE 102 receives an RRCRelease message. In some examples, the RRCRelease message includes Suspendconfig field. Then, at block 722, the UE 102 stores the first inference configuration, the first inference information and / or the first applicability reporting information and enters an inactive state. In one example, the first inference configuration and / or the first inference information (optionally in addition or in alternative the first applicability reporting information) is stored in UE Inactive AS context. At block 726, the UE 102 decides to perform an RRC resume procedure with a T-BS 106. Then, at block 728, the UE 102 transmits an RRCResumeRequest message to the T-BS 106. At block 734A, the UE 102 receives an RRCResume message that includes a second inference configuration. In one implementation, the second inference configuration is the same as the first inference configuration. The RRCResume message may include one or more fields (e.g., useStoredinference field) to request UE 102 to apply either the first inference configuration or instead to include the second inference configuration (thus requesting the UE 102 to apply the second inference configuration). Then, at block 736A, the UE 102 applies the first or second inference configuration as requested and transmits an RRCResumeComplete message to the T-BS 106. In one example, the UE 102 includes a second inference information in the RRCResumeComplete message. Then, at block 714, the UE 102 is able to activate / deactivate the AI / ML functionality with the first or second inference configuration.

[0204] At block 738, the UE 102 transmits an Applicability functionality reporting message that includes a first inference information to the T-BS 106. In one implementation, the UE 102 transmits the Applicability functionality reporting while the said inference capability in 812 does not include an inference configuration. At block 740, the UE 102 receives an RRCReconfiguration message that includes a third inference configuration. Then, at block 744, the UE 102 is able to activate / deactivate the AI / ML functionality with the third inference configuration.

[0205] Referring now to Fig. 7B, a method 700B can be implemented in a base station in a RAN and includes a process for transmitting a command to configure the AI / ML functionality. For clarity, the method 700B is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102. The method 700B may implement S-BS104 procedures of the method 400A shown in Fig. 4A.

[0206] At block 750, the S-BS 104 is configured with configured with the AI / ML functionality with the UE 102. At block 716, the S-BS 104 decides to request the UE 102 to enter an inactive state. Then, at block 718A, the S-BS 104 stores the first inference configuration and / or the first inference information of the UE 102. At block 720, the S-BS 104 transmits an RRCRelease message that includes a Suspendconfig field to the UE 102. At block 730, the S-BS 104 receives a RETRIEVE UE CONTEXT REQUEST from a T-BS 106. Then, at block 732A, the S-BS 104 transmits the first inference information and / or first inference configuration (and / or applicability reporting information) by a RETRIEVE UE CONTEXT RESPONSE to the T-BS 106.

[0207] Referring now to Fig. 7C, a method 700C can be implemented in a base station in a RAN and includes a process for transmitting a command to configure the AI / ML functionality. For clarity, the method 700C is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102. The method 700C may implement T-BS 106 procedures of the method 400A shown in Fig. 4A.

[0208] At block 728, the T-BS 106 receives an RRC Resume Request message. Then, at block 730, the T-BS 106 transmits a RETRIEVE UE CONTEXT REQUEST to the S-BS 104. At block 732A, the T-BS 106 receives the first inference information and / or first inference configuration (and / or first applicability reporting information) by a RETRIEVE UE CONTEXT RESPONSE from the S-BS 104. At block 736A, the T-BS 106 includes a second inference configuration in an RRCResume message and transmit it to the UE 102 (at this stage, the T-BS may alternatively include instructions to continue to use the first inference configuration). The at block 736A, the T-BS 106 receives a RRCResumeComplete message from the UE 102. At block 714, the T-BS 106 is able to activate / deactivate the AI / ML functionality with the second inference configuration with the UE 102.

[0209] At block 738, the T-BS 106 receives an Applicability functionality reporting message that includes a second inference information from the UE 102. In one implementation, the T-BS 106 receives the second inference information from the RRCResumeComplete message instead of the Applicability functionality. At block 740, the T-BS 106 transmits an RRCReconfiguration message that includes a thirdinference configuration to the UE 102. Then, at block 744, the UE 102 is able to activate / deactivate the AI / ML functionality with the third inference configuration.

[0210] Referring now to Fig. 7D, a method 700D can be implemented in a suitable UE and includes a process for handling RRCRelease message while it is configured with the AI / ML functionality. For clarity, the method 700D is discussed with specific reference to the S-BS 104, T-BS 106 and the UE 102. The method 700D may implement UE 102 procedures of the method 400B shown in Fig. 4B.

[0211] At block 750, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 720, the UE 102 receives an RRCRelease message. In some examples, the RRCRelease message includes Suspendconfig field. Then, at block 722, the UE 102 stores the first inference configuration and / or the first inference information and / or first applicability reporting information and enters an inactive state. In one example, the first inference configuration and the first inference information (and / or first applicability reporting information) is stored in UE Inactive AS context. At block 726, the UE 102 decides to perform an RRC resume procedure with a T-BS 106. Then, at block 728, the UE 102 transmits an RRCResumeRequest message to the T-BS 106. At block 734B, the UE 102 receives an RRCResume message that includes an inference configuration and inference information request field (e.g., an Inference Infoconfig request). In one example, the received RRCResume message may include one or more field (e.g., useStoredinference field) to request UE 102 to apply the first inference configuration. Then, at block 736B, the UE 102 transmits an RRCResumeComplete message that includes the second inference information and / or the first inference configuration to the T-BS 106. At block 746B, the UE 102 receives an RRCReconfiguration message that includes a second inference configuration. Alternatively, this may may include an indication to continue using the fist inference configuration is appropriate. At block 748, the UE 102 applies the second inference configuration and transmits an RRCReconfigurationComplete message. Then, at block 714, the UE 102 able to activate / deactivate the AI / ML functionality with the second inference configuration. In one implementation, the UE 102 includes the first inference information and / or the first inference configuration in an Applicability functionality reporting message instead of including one or both of them in the RRCResumeComplete message.

[0212] Referring now to Fig. 7E, a method 700E can be implemented in a base station in a RAN and includes for transmitting a command to configure the AI / ML functionality. For clarity, the method 700E is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102. The method 700E may implement T-BS 106 procedures of the method 400B shown in Fig. 4B.

[0213] At block 728, the T-BS 106 receives an RRC Resume Request message. Then, at block 730, the T-BS 106 transmits a RETRIEVE UE CONTEXT REQUEST to the S-BS 104. At block 732B, the T-BS 106 receives a RETRIEVE UE CONTEXT RESPONSE from the S-BS 104. In one implementation, there is neither the inference information nor the inference configuration in the RETRIEVE UE CONTEXT RESPONSE. In another implementation, there is an inference information and / or an inference configuration in the RETRIEVE UE CONTEXT RESPONSE. Then, at block 734B, the T-BS includes an inference configuration and inference information request field in an RRCResume message and transmits it to the UE 102. Then, at block 736B, the T-BS 106 receives an RRCResumeComplete message that includes a second inference information and / or the first inference configuration from the UE 102. At block 746B, the T-BS 106 transmits an RRCReconfiguration message that includes a second inference configuration to the UE 102. At block 748, the T-BS 106 receives an RRCReconfigurationComplete message from the UE 102. Then, at block 714, the T-BS 106 is able to activate / deactivate the AI / ML functionality with the second inference configuration with the UE 102.

[0214] Referring now to Fig. 7F, a method 700F can be implemented in a base station in a RAN and includes a process for transmitting commands to configure the AI / ML functionality. For clarity, the method 700F is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102. The method 700F illustrates how a T-BS 106 may implement the methods 400A or400B as appropriate.

[0215] At block 728, the T-BS 106 receives an RRC Resume Request message. Then, at block 730, the T-BS 106 transmits a RETRIEVE UE CONTEXT REQUEST to the S-BS 104. At block 732, the T-BS 106 receives a RETRIEVE UE CONTEXT RESPONSE from the S-BS 104. Then, at block 770, the T-BS 106 checks if there is any inference information and / or the inference configuration in the RETRIEVE UE CONTEXT RESPONSE. If included, at block 734A, the T-BS 106 includes a secondinference configuration in an RRCResume message and transmit it to the UE 102 (in line with the process of Figure 4A for example). Then, at block 736A, the T-BS 106 receives an RRCResumeComplete message. If the check at 770 shows that the inference configuration and / or inference information is not included, the method may proceed as in Figure 4B, for example. As such, at block 734B, the T-BS 106 includes an inference configuration and inference information request field in an RRCResume message and transmits it to the UE 102. Then, at block 736B, the T-BS 106 receives an RRCResumeComplete message that includes a second inference information and / or the first inference configuration from the UE 102. At block 446B, the T-BS 106 transmits an RRCReconfiguration message that includes a second inference configuration to the UE 102.

[0216] Referring now to Fig. 7G, a method 700G can be implemented in a suitable UE and includes a process for handling an RRCRelease message while it is configured with the AI / ML functionality. For clarity, the method 700G is discussed with specific reference to the S-BS 104 and the UE 102. This process may be implemented within the contexts of processes such as the process 400C of Fig. 4C, for example.

[0217] At block 750, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 720, the UE 102 receives an RRCRelease message. At block 723, the UE 102 releases the first inference information and the first inference configuration (and / or applicability reporting information). Then at block 725, the UE 102 enters an idle state or inactive state according to the RRCRelease message from the BS 104. In one example, the UE 102 stores the UE Inactive AS context and releases the first inference information and / or the first inference configuration. In another example, the UE 102 releases the the VarConditionalReconfig, the VarServingSecurityCellSetID, the first inference information and / or the first inference configuration.

[0218] Referring now to Fig. 7H, a method 700H can be implemented in a suitable UE and includes a process for handling an RRCRelease message while it is configured with the AI / ML functionality. For clarity, the method 700H is discussed with specific reference to the T-BS 106, the S-BS 104 and the UE 102. The method 700Hcan reflect UE behaviour within the process of processes 400D, 400E and 400F described in the context of Figs. 4D, 4E and 4F.

[0219] At block 750, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 720, the UE 102 receives an RRCRelease message. Then, at block 722, the UE 102 stores the first inference configuration and / or the first inference information and enters an inactive state. In one example, the first inference configuration and the first inference is stored in UE Inactive AS context. At block 726, the UE 102 decides to perform an RRC resume procedure with a T-BS 106. Then, at block 728, the UE 102 transmits an RRCResumeRequest message to the T-BS 106. At block 721 , the UE 102 receives an RRCReject or RRCSetup or an RRCRelease message without a suspendConfig field. Then, at block 723, the UE 102 releases the first inference information and the first inference configuration. In one implementation, the UE 102 receives an RRCRelease message with suspendConfig field after it transmitting the RRCResumeRequest message at block 810. In this example, the UE 102 does not release the first inference information and the first inference configuration after it receives the RRCRelease message.

[0220] Referring now to Fig. 8A, a method 800A can be implemented in a suitable UE and includes a process for performing a RRC re-establishment procedure while it is configured with the AI / ML functionality. For clarity, the method 800A is discussed with specific reference to the S-BS 104, T-BS 106 and the UE 102. The method 800A illustrates how a UE 102 may implement the method 500A of Fig. 5A.

[0221] At block 850, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 818, the UE 102 decides to perform a RRC re-establishment procedure with the T-BS 106 (e.g., due to radio link failure or handover failure). Then, at block 822, the UE 102 stores the first inference configuration and / or the first inference information (and / or the first applicability reporting information). At block 852, the UE 102 deactivate the AI / ML functionality with the first inference configuration. Then, at block 828, the UE 102 transmits an RRCReestablishmentRequest message to the T-BS 106. At block 834, the UE 102 receives an RRCReestablishment message. At block 836, the UE 102 transmits an RRCReestablishmentComplete message to the T-BS 106. Then, at block 846A, the UE 102 receives a RRCReconfiguration message that includes a second inferenceconfiguration. At block 848A, UE 102 applies the second inference configuration and transmits an RRCReestablishmentComplete message to the T-BS 106. Then, at block 814, the UE 102 is able to activate / deactivate the AI / ML functionality with the second inference configuration.

[0222] Referring now to Fig. 8B, a method 800B can be implemented in a base station in a RAN and includes a process for transmitting a command to configure the AI / ML functionality. For clarity, the method 800B is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102. The method 800A illustrates how a T-BS 106 may implement the method 500Aof Fig. 5A.

[0223] At block 828, the T-BS 106 receives an RRCReestablishmentRequest message. Then, at block 830, the T-BS 106 transmits a RETRIEVE UE CONTEXT REQUEST to the S-BS 104. At block 832A, the T-BS 106 receives the first inference information and / or first inference configuration (and / or first applicability reporting information) by a RETRIEVE UE CONTEXT RESPONSE from the S-BS 104. At block 834, the T-BS 106 transmits an RRCReestablishment message to the UE 102. At block 836, the T-BS 106 receives an RRCReestablishmentComplete message from the UE 102. Then, at block 846A, the T-BS 106 includes a second inference configuration in an RRCReconfiguration message and transmit it to the UE 102. Then, at block 848A, the T-BS 106 receives an RRCReconfigurationComplete message from the UE 102. At block 940, the T-BS 106 is able to activate / deactivate the AI / ML functionality with the second inference configuration with the UE 102.

[0224] Referring now to Fig. 8C, a method 800C can be implemented in a suitable UE and includes a process for performing a RRC re-establishment procedure while it is configured with the AI / ML functionality. For clarity, the method 800C is discussed with specific reference to the S-BS 104, T-BS 106 and the UE 102. The method 800C illustrates how a UE 102 may implement the method 500B of Fig. 5B.

[0225] At block 850, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 818, the UE 102 decides to perform a RRC re-establishment procedure with the T-BS 106 (e.g., due to radio link failure or handover failure) as in Fig. 9A. Then, at block 822, the UE 102 stores the first inference configuration and / or the first inference information (and / or the first applicability reporting information). At block 846B, the UE 102 receives an RRCReconfiguration message that includes aninference configuration and inference information request field. At block 848B, UE 102 includes the first inference configuration and / or second inference information in an RRCReconfigurationComplete message and transmits it to the T-BS 106. In one implementation, the UE 102 includes the first inference configuration and / or second inference information in an Applicability functionality reporting instead includes them in the RRCReconfigurationComplete message. Then, at block 388, the UE 102 transmits the Applicability functionality reporting to the T-BS 106. At block 840B, the UE 102 receives an RRCReconfiguration message that includes a second inference configuration. At block 842, UE 102 applies the first second inference configuration and transmits an RRCReestablishmentComplete message to the T-BS 106. Then, at block 814, the UE 102 is able to activate / deactivate the AI / ML functionality with the second inference configuration.

[0226] Referring now to Fig. 8D, a method 800D can be implemented in a base station in a RAN and includes a process for transmitting a command to configure the AI / ML functionality. For clarity, the method 800D is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102. The method 800D illustrates how a T-BS 106 may implement the method 500B of Fig. 5B.

[0227] Initially, the T-BS 106 performs the actions as specified in block 828 and 830 (in Fig. 8B). At block 832B, the T-BS 106 receives a RETRIEVE UE CONTEXT RESPONSE from the S-BS 104. The RETRIEVE UE CONTEXT RESPONSE in this does not include either the inference information or the inference configuration. At block 834, the T-BS 106 transmits an RRCReestablishment message to the UE 102. At block 836, the T-BS 106 receives an RRCReestablishmentComplete message from the UE 102. Then, at block 846B, the T-BS 106 includes an inference configuration and inference information request field in an RRCReconfiguration message and transmits it to the UE 102. Then, at block 848B, the T-BS 106 receives an RRCReconfigurationComplete message that includes the first inference configuration and / or second inference information from the UE 102. In one implementation, the T-BS 106 receives the second inference information and / or the first inference configuration from an Applicability functionality reporting instead of from the RRCReconfigurationComplete message (block 838). Then, at block 840B, the T-BS 106 includes a second inference configuration in an RRCReconfiguration message and transmits it to the UE 102. Then, at block 842, the T-BS 106 receivesan RRCReconfigurationComplete message from the UE 102. At block 814, the T-BS 106 is able to activate / deactivate the AI / ML functionality with the second inference configuration with the UE 102.

[0228] Referring now to Fig. 8E, a method 800E can be implemented in a base station in a RAN and includes a process for transmitting a command to configure the AI / ML functionality. For clarity, the method 800E is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102. The method of Fig. 8E illustrates how a T-BS 106 may operate to handle processes according to either the method of Fig. 5A or Fig. 5B.

[0229] At block 828, the T-BS 106 receives an RRCReestablishmentRequest message. Then, at block 830, the T-BS 106 transmits a RETRIEVE UE CONTEXT REQUEST to the S-BS 104. At block 832, the T-BS 106 receives a RETRIEVE UE CONTEXT RESPONSE from the S-BS 104. Then, at block 890, the T-BS 106 checks whether the RESPONSE includes inference information and / or inference configuration. If it is included, the T-BS 106 performs the actions as specified in blocks 836, 846, 846A, 848A and 814 (i.e. in line with the process of Fig. 5A and 8B). If it is not included, the T-BS 106 performs the actions as specified in blocks 834, 836, 846B, 848B, 838, 840B, 842 and 814 (i.e. in line with the process of Fig. 5B and 8D).

[0230] Referring now to Fig. 8F, a method 800F can be implemented in a suitable UE and includes a process for performing a RRC re-establishment procedure while it is configured with the AI / L functionality. For clarity, the method 800F is discussed with specific reference to the S-BS 104, T-BS 106 and the UE 102. The method 800F illustrates how a UE 102 may implement the method 500C of Fig. 5C.

[0231] At block 850, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 818, the UE 102 decides to perform a RRC re-establishment procedure with the T-BS 106 (e.g., due to radio link failure or handover failure). Then, at block 823, the UE 102 releases the first inference information and / or the first inference configuration in response to this decision. The UE 102 then deactivate the AI / ML functionality with the first inference configuration. Then, at block 828, the UE 102 transmits an RRCReestablishmentRequest message to the T-BS 106. At block 834, the UE 102 receives an RRCReestablishment message. At block 836, the UE102 transmits an RRCReestablishmentComplete message to the T-BS 106. 851 , the UE 102 is configured with the AI / ML functionality with the T-BS 106. The UE 102 may thus be provided with second inference configuration. At block 814, the T-BS 106 is able to activate / deactivate the AI / ML functionality with the second inference configuration with the UE 102.

[0232] Referring now to Fig. 8G, a method 800G can be implemented in a suitable UE and includes for perform a RRC re-establishment procedure while it is configured with the AI / ML functionality. For clarity, the method 800G is discussed with specific reference to the S-BS 104, T-BS 106 and the UE 102.

[0233] At block 850, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 854, the UE 102 decides to perform a CHO or conditional LTM procedure. Then, at block 860, the UE 102 checks the trigger event of this decision is RRC re-establishment procedure or not. Then, at block 823, the UE 102 releases the stored inference information and / or the stored inference configuration and / or inference applicability reporting configuration while the trigger event is RRC reestablishment procedure. If not, at block 864, the UE 102 keeps the stored inference configuration and stored inference information and / or inference applicability reporting configuration

[0234] Referring now to Fig. 8H, a method 900H can be implemented in a suitable UE and includes for perform a RRC re-establishment procedure while it is configured with the AI / ML functionality. For clarity, the method 800H is discussed with specific reference to the S-BS 104, T-BS 106 and the UE 102. The method 800H illustrates how a UE 102 may implement the method 500D of Fig. 5D.

[0235] At block 850, the UE 102 is configured with the AI / ML functionality with the S-BS 104. At block 818, the UE 102 decides to perform a RRC re-establishment procedure with the T-BS 106 (e.g., due to radio link failure or handover failure). Then, at block 822, the UE 102 stores the first inference configuration and / or the first inference information (and / or first applicability reporting information). At block 852, the UE 102 deactivates the AI / ML functionalities with the first inference configuration. Then, at block 828, the UE 102 transmits an RRCReestablishmentRequest message to the T-BS 106. At block 835, the UE 102 receives an RRCSetup message from the T-BS 106. Then, at block 823, the UE 102 releases the first inference informationand / or the first inference configuration and / or inference applicability reporting configuration.

[0236] Referring now to Fig. 8I, a method 800I can be implemented in a base station in a RAN and includes for transmitting a command to configure the AI / ML functionality. For clarity, the method 800I is discussed with specific reference to the RAN 105, T-BS 106, S-BS 104 and the UE 102.

[0237] At block 850, the S-BS 104 is configured with configured with the AI / ML functionality with the UE 102. At block 880, the S-BS 104 detects the UE 102 is out of sync (e.g., receives a RETRIEVE UE CONTEXT REQUEST from another BS or the guard timer to receive a response message expired). Then, at block 822, the S-BS 104 stores the first inference information and / or the first inference configuration of the UE 102. At block 814, the S-BS 104 dseactivates the AI / ML functionality with the first inference configuration with the UE 102.

[0238] The following additional considerations apply to the foregoing discussion.

[0239] In some implementations, “message” is used and can be replaced by “information element (IE)”. In some implementations, “IE” is used and can be replaced by “field”. In some implementations, “configuration” can be replaced by “configurations” or the configuration parameters.

[0240] A user device in which the techniques of this disclosure can be implemented (e.g., the UE 102) can be any suitable device capable of wireless communications such as a smartphone, a tablet computer, a laptop computer, a mobile gaming console, a point-of-sale (POS) terminal, a health monitoring device, a drone, a camera, a media-streaming dongle or another personal media device, a wearable device such as a smartwatch, a wireless hotspot, a femtocell, or a broadband router. Further, the user device in some cases may be embedded in an electronic system such as the head unit of a vehicle or an advanced driver assistance system (ADAS). Still further, the user device can operate as an internet-of-things (loT) device or a mobile-internet device (MID). Depending on the type, the user device can include one or more general-purpose processors, a computer-readable memory, a user interface, one or more network interfaces, one or more sensors, etc.

[0241] Certain embodiments are described in this disclosure as including logic or a number of components or modules. Modules may can be software modules (e.g., code stored on non-transitory machine-readable medium) or hardware modules. A hardware module is a tangible unit capable of performing certain operations and may be configured or arranged in a certain manner. A hardware module can comprise dedicated circuitry or logic that is permanently configured (e.g., as a special-purpose processor, such as a field programmable gate array (FPGA) or an Applicability -specific integrated circuit (ASIC)) to perform certain operations. A hardware module may also comprise programmable logic or circuitry (e.g., as encompassed within a general-purpose processor or other programmable processor) that is temporarily configured by software to perform certain operations. The decision to implement a hardware module in dedicated and permanently configured circuitry, or in temporarily configured circuitry (e.g., configured by software) may be driven by cost and time considerations.

[0242] When implemented in software, the techniques can be provided as part of the operating system, a library used by multiple Applicability s, a particular software Applicability , etc. The software can be executed by one or more general-purpose processors or one or more special-purpose processors.

[0243] Upon reading this disclosure, those of skill in the art will appreciate still additional alternative structural and functional designs for inference configuration / information through the disclosed principles herein. Thus, while particular embodiments and Applicability s have been illustrated and described, it is to be understood that the disclosed embodiments are not limited to the precise construction and components disclosed herein. Various modifications, changes and variations, which will be apparent to those of ordinary skill in the art, may be made in the arrangement, operation and details of the method and apparatus disclosed herein without departing from the spirit and scope defined in the appended claims.

Claims

CLAIMS1. A computer implemented method performed by a user equipment, UE, the method comprising:performing a functionality configuration procedure in communication with a source base station, the functionality configuration procedure resulting in at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with the source base station;receiving an RRCRelease message from the source base station; entering an inactive state;transmitting an RRCResumeRequest message to a target base station; receiving a reply to the RRCResumeRequest message from the target base station, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply,wherein the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

2. A method according to claim 1 , wherein inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

3. A method according to claim 1 or claim 2, wherein inference configuration comprises one or more configuration parameters to enable AI / ML such as channel state information, CSI, feedback, beam management, positioning, and mobility.

4. A method according to any one of the preceding claims, wherein inference applicability reporting configuration provides scheduling for the UE to report inference information.

5. A method according to any one of the preceding claims, further comprising,after receiving the RRCRelease message, storing the at least one configuration UE.

6. A method according to claim 5, wherein the reply to the RRCResumeRequest message indicates whether to configure AI / ML functionalities with the target base station using the at least one stored configuration.

7. A method according to claim 6, wherein the reply to the RRCResumeRequest message indicates to use an updated configuration set to configure AI / ML functionalities with the target base station.

8. A method according to claim 7, wherein the updated configuration comprises second inference configuration.

9. A method according to claim 5, further comprising:after receiving the reply to the RRCResumeRequest message from the target base station, transmitting to the target base station at least one of the at least one configuration.

10. A method according to claim 9, further comprising:after receiving the reply to the RRCResumeRequest message from the target base station, transmitting to the target base station second inference information.

11. A method according to claim 9 or claim 10, wherein a further message from the target base station indicates whether to configure AI / ML functionalities with the target base station using the at least one stored configuration.

12. A method according to claim 11, wherein the further message is an RRCReconfiguration message.

13. A method according to claim 11 or claim 12, wherein the further message indicates to use an updated configuration to configure AI / ML functionalities with the target base station.

14. A method according to any one of claims 1 to 4, further comprising: after receiving the RRCRelease message, releasing the at least one configuration at the UE.

15. A method according to claim 14, further comprising:after receiving the reply to the RRCResumeRequest message from the target base station, performing a functionality configuration procedure in communication with the target base station, the functionality configuration procedure resulting in one or more further configurations for configuring AI / ML functionalities with the target base station.

16. A method according to any one of claims 5 to 15, wherein the reply to the RRCResumeRequest message from the target base station is an RRCResume message.

17. A method according to any one of claims 1 to 4, further comprising, after receiving the RRCRelease message, storing the at least one configuration at the UE, andafter receiving the reply to the RRCResumeRequest message from the target base station, releasing the at least one configuration at the UE.

18. A method according to claim 17, wherein the reply to the RRCResumeRequest message from the target base station is an RRCSetup message.

19. A method according to any one of claims 1 to 4, further comprising, after receiving the RRCRelease message, storing the at least one configuration at the UE, andafter receiving the reply to the RRCResumeRequest message from the target base station, entering the inactive state.

20. A method according to claim 19, wherein the reply to the RRCResumeRequest message from the target base station is an RRCReject message.

21. A method according to any one of claims 1 to 4, further comprising,after receiving the RRCRelease message, storing the at least one configuration at the UE, andwherein the reply to the RRCResumeRequest message from the target base station is a RRCRelease message.

22. A method according to claim 21 , further comprising,after receiving the RRCRelease message, releasing the at least one configuration at the UE.

23. A computer implemented method performed by a user equipment, UE, the method comprising:performing a functionality configuration procedure in communication with a source base station, the functionality configuration procedure resulting in at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with the source base station;transmitting an RRCReestablishmentRequest message to a target base station;receiving a reply to the RRCReestablishmentRequest message from the target base station, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply,and wherein the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

24. A method according to claim 23, wherein the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE contextinformation, UE mobility information, or analytics requirements for the AI / ML procedures.

25. A method according to claim 23 or claim 24, wherein inference configuration comprises one or more configuration parameters to enable AI / ML for channel state information, CSI, feedback, beam management, positioning, and mobility.

26. A method according to any one of the preceding claim, wherein inference applicability reporting configuration provides scheduling for the UE to report inference information.

27. A method according to any one of claims 23 to 26, further comprising, prior to transmitting the RRCReestablishmentRequest message, storing the at least one configuration at the UE.

28. A method according to claim 24, further comprising, receiving a message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the at least one configuration.

29. A method according to claim 28, wherein the message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using at least one configuration indicates to use an updated configuration to configure AI / ML functionalities with the target base station.

30. A method according to claim 29, wherein the updated configuration comprises second inference configuration.

31. A method according to claim 29 or claim 30, wherein the message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the at least one configuration is an RRCReconfiguration message.

32. A method according to claim 27, further comprising:after receiving a reply to the RRCReestablishmentRequest message from the target base station, transmitting to the target base station at least one of the at least one configuration.

33. A method according to claim 32, further comprisingprior to the transmitting to the target base station of the at least one of the at least one configuration, receiving a message from the target base station requesting at least one of the at least one configuration.

34. A method according to claim 33, wherein the message from the target base station requesting at least one of the at least one configuration is an RRCReconfiguration message.

35. A method according to claims 32 to 34, further comprising:after transmitting to the target base station at least one of the at least one configuration, receiving a message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the at least one stored configuration.

36. A method according to claim 35, wherein the message from the target base station that indicates whether to configure AI / ML functionalities with the target base station using the at least one configuration is an RRCReconfiguration message.

37. A method according to any one of claims 23 to 26, further comprising, prior to transmitting the RRCReestablishmentRequest message, releasing the at least one configuration at the UE.

38. A method according to claim 37, further comprising:after receiving the reply to the RRCReestablishmentRequest message from the target base station, performing a functionality configuration procedure in communication with the target base station, the functionality configuration procedure resulting in at least one further configuration for configuring AI / ML functionalities with the target base station.

39. A method according to any one of claims 23 to 26, further comprising, prior to transmitting the RRCReestablishmentRequest message, storing the at least one configuration at the UE, andafter receiving a reply to the RRCReestablishmentRequest message from the target base station, releasing the at least one configuration at the UE.

40. A method according to claim 39, wherein the reply to the RRCReestablishmentRequest message from the target base station is an RRCSetup message.

41. A computer implemented method performed by a target base station, T-BS, the method comprising:receiving an RRCResumeRequest message from a user equipment, UE, wherein UE being associated with at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with a source base station;transmitting a reply to the RRCResumeRequest message to the UE, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply,and wherein the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

42. A method according to claim 41 , wherein the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

43. A method according to claim 41 or claim 42, wherein inference configuration comprises one or more configuration parameters to enable AI / ML such as channel state information, CSI, feedback, beam management, positioning, and mobility.

44. A method according to any one of claims 41 to 43, wherein inference applicability reporting configuration provides scheduling for the UE to report inference information.

45. A method according to any one of claims 41 to 44, further comprising, responsive to receiving the RRCResumeRequest message, transmitting a RETRIEVE UE CONTEXT REQUEST to the source base station.

46. A method according to claim 45, further comprising, after transmitting the RETRIEVE UE CONTEXT REQUEST to the source base station, receiving a response from the source base station.

47. A method according to claim 46, wherein the response to the RETRIEVE UE CONTEXT REQUEST is a RETRIEVE UE CONTEXT RESPONSE.

48. A method according to claim 46 or claim 47, wherein the response to the RETRIEVE UE CONTEXT REQUEST comprises the at least one configuration, the at least one configuration being stored at the UE.

49. A method according to claim 48, further comprising, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / ML functionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

50. A method according to claim 49, wherein the updated at least one configuration comprises a second inference configuration.

51. A method according to any one of claims 41 to 47, wherein, after receiving the RRCResumeRequest message, the method further comprises transmitting a request for the at least one configuration to the UE.

52. A method according to any one of claims 41 to 47 or 51 , further comprising receiving the at least one configuration from the UE.

53. A method according to claim 52, wherein the at least one configuration received from the UE comprises a first inference configuration.

54. A method according to claim 53, further comprising receiving first inference information and / or second inference information from the UE.

55. A method according to any one of claims 52 to 54, further comprising, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / ML functionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

56. A method according to claim 55, wherein the updated at least one configuration comprises a second inference configuration.

57. A method according to any one of claims 41 to 47, further comprising:after transmitting the reply to the RRCResumeRequest message to the UE, performing a functionality configuration procedure in communication with the UE.

58. A method according to any one of claims 1 to 7, wherein the reply to the RRCResumeRequest message to the UE causes the UE to release the at least one configuration.

59. A method according to claim 18, wherein the reply to the RRCResumeRequest message is an RRCSetup message.

60. A method according to any one of claims 41 to 47, wherein the reply to the RRCResumeRequest message is an RRCReject message.

61. A method according to any one of claims 41 to 47, wherein the reply to the RRCResumeRequest message is an RRCRelease message.

62. A computer implemented method performed by a target base station, T-BS, the method comprising:receiving an RRCReestablishmentRequest message from a user equipment, UE, wherein UE being associated with at least one configuration configuring one or more artificial intelligence / machine learning, AI / ML, functionalities with a source base station;transmitting a reply to the RRCReestablishmentRequest message to the UE, wherein further coordination of the one or more AI / ML functionalities is responsive to the reply,and wherein the at least one configuration comprises at least one of: first inference information, first inference configuration and first inference applicability reporting configuration.

63. A method according to claim 62, wherein the inference information comprises at least one of: identification information for an AI / ML model, inference type information indicating a type of inference used in the AI / ML functionalities, UE context information, UE mobility information, or analytics requirements for the AI / ML procedures.

64. A method according to claim 62 or claim 63, wherein inference configuration comprises one or more configuration parameters to enable AI / ML such as channel state information, CSI, feedback, beam management, positioning, and mobility.

65. A method according to any one of claims 62 to 64, wherein inference applicability reporting configuration provides scheduling for the UE to report inference information.

66. A method according to any one of claims 62 to 66, further comprising,responsive to receiving the RRCReestablishmentRequest message, transmitting a RETRIEVE UE CONTEXT REQUEST io the source base station.

67. A method according to claim 66, further comprising, after transmitting the RETRIEVE UE CONTEXT REQUEST to the source base station, receiving a response from the source base station.

68. A method according to claim 67, wherein the response to the RETRIEVE UE CONTEXT REQUEST S a RETRIEVE UE CONTEXT RESPONSE.

69. A method according to claim 67 or claim 68, wherein the response to the RETRIEVE UE CONTEXT REQUEST comprises the at least one configuration, the at least one configuration being stored at the UE.

70. A method according to claim 69, further comprising, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / ML functionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

71. A method according to claim 70, wherein the updated at least one configuration comprises a second inference configuration.

72. A method according to any one of claims 61 to 68 wherein, after receiving the RRCReestablishmentRequest message, the method further comprises transmitting a request for the at least one configuration to the UE.

73. A method according to any one of claims 61 to 68 or 72, further comprising receiving the at least one configuration from the UE.

74. A method according to claim 73, wherein the at least one configuration received from the UE comprises a first inference configuration.

75. A method according to claim 74, further comprising receiving first inference information and / or second inference information from the UE.

76. A method according to any one of claims 73 to 75, further comprising, after receiving the at least one configuration, transmitting to the UE an indication of whether to configure AI / ML functionalities with the target base station using the stored at least one configuration or an updated at least one configuration.

77. A method according to claim 76, wherein the updated at least one configuration comprises a second inference configuration.

78. A method according to any one of claims 61 to 68, further comprising:after transmitting the reply to the RRCReestablishmentRequest message to the UE, performing a functionality configuration procedure in communication with the UE.

79. A method according to any one of claims 61 to 68, wherein the reply to the RRCReestablishmentRequest message to the UE causes the UE to release the at least one configuration80. A method according to claim 79, wherein the reply to the RRCReestablishmentRequest message is an RRCSetup message.

81. A user equipment, UE, for performing the method of any one of claims 1 to 40.

82. A base station for performing the method of any one of claims 41 to 80.

Images