Input selection aware monitoring for enhanced machine learning based positioning

Abstract

At a first apparatus, which may be a user equipment or a gNB: receiving, from a second apparatus, which may be a gNB, a network function or an LMF, a request to initiate obtaining at least one matching factor, which indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtaining the at least one matching factor based on the request; performing at least a machine learning model switching based on a monitoring decision request from the second apparatus, the request based at least on the obtained matching factor and at least one channel characteristic. The monitoring decision request may be further be based on at least one positioning accuracy parameter. The inputs may be transmission-reception-point measurements for a location management service or positioning determination.

Description

FIELD The present application relates to a method, apparatus, system and computer program for performing input selection aware monitoring for enhanced machine learning based positioning and in particular, but not exclusively to, a method, apparatus, system and computer program for performing transmission-reception selection aware monitoring for enhanced machine learning based positioning. BACKGROUND A communication system can be seen as a facility that enables communications between two or more entities such as terminals, and / or other nodes, or provides connected services to entities. A communication system can include communication networks and one or more compatible terminals (otherwise known as communication devices). Communications may carry, for example, voice, video, electronic mail (email), text message, multimedia data and / or content data and so on. Non-limiting examples of connected services provided by the communications system may comprise enhanced mobile broadband, ultra-reliable low latency communications, mission-critical communications, massive internet of things (loT), and multimedia services. In a communication system at least a part of communications between at least two entities occurs over a wireless link. Examples of networks in a communication system are public land mobile networks (PLMN), radio access networks such as terrestrial radio access networks or non-terrestrial radio access networks (e.g., satellite networks) and different wireless local networks, for example wireless local area networks (WLAN). Radio access networks can include cells and are therefore often referred to as cellular networks. A terminal may be referred to as user equipment (UE) or user device. A terminal is provided with an appropriate signal receiving and transmitting apparatus for enabling communications, for example enabling access to a communication network or communications directly with other terminals. The terminal may access a carrier provided by a base station, for example a base station of a radio access network, and transmit and / or receive communications on the carrier. A communication system and associated compatible terminals typically operate in accordance with a given standard or specification which sets out what various network entities of the communication system are permitted to do and how that should be achieved. Communication protocols and / or parameters which shall be used for communications are also typically defined. There are published studies which explore the benefits of augmenting the airinterface with features enabling improved support of Artificial Intelligence / Machine Learning (AI / ML) applications. For example, AI / ML support in location management functionality (LMF) of a 5G network is one field in which research is currently being performed. SUMMARY According to an aspect, there is provided a method for a first apparatus, the method comprising: receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtaining the at least one matching factor based on the request to initiate obtaining the at least one matching factor; performing at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic. The monitoring decision request may be further be based on at least one positioning accuracy parameter. Obtaining, in the first apparatus, at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase may comprise: identifying inputs used during the machine learning inference phase; identifying inputs used during the machine learning training phase; and generating the at least one matching factor based a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase. The inputs may be transmission-reception-point measurements for a location management service or positioning determination. The first apparatus may be one of: a user equipment; a gNB. The second apparatus may be one of: a gNB; a network function; and a LMF. According to a second aspect there is provided a method for a second apparatus, the method comprising: generating, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; obtaining, from the first apparatus, information based on the at least one matching factor; monitoring the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching. The expected accuracy may be further based on at least one positioning accuracy parameter. Monitoring the at least one matching factor may comprise obtaining at least one threshold matching factor corresponding to at least one of: at least one channel characteristic; and at least one positioning accuracy parameter. Obtaining at least one threshold matching factor may comprise obtaining at least one threshold matching factor based on a look-up table. Monitoring the at least one matching factor may comprise determining if the at least one matching factor is lower than the at least one threshold matching factor, wherein the monitoring decision request is generated based on the determination. The at least one matching factor may be based on a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase. The inputs may be transmission-reception-point measurements for a location management service or positioning determination. The method may further comprise: subscribing, with respect to a third apparatus, to a positioning analysis service; generating, for the third apparatus, machine learning functionality information; obtaining, from the third apparatus, at least one matching factor based rule for generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor. The first apparatus may be one of: a user equipment; a gNB. The second apparatus may be one of: a gNB; a network function; and a LMF. The third apparatus may be a network data analytics function. According to a third aspect, there is provided a first apparatus comprising means configured to: receive, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtain the at least one matching factor based on the request to initiate obtaining the at least one matching factor; perform at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic. The monitoring decision request may be further be based on at least one positioning accuracy parameter. The means configured to obtain, in the first apparatus, at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase may be configured to: identify inputs used during the machine learning inference phase; identify inputs used during the machine learning training phase; and generate the at least one matching factor based a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase. The inputs may be transmission-reception-point measurements for a location management service or positioning determination. The first apparatus may be one of: a user equipment; a gNB. The second apparatus may be one of: a gNB; a network function; and a LMF. According to a fourth aspect, there is provided a second apparatus comprising means configured to: generate, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; obtain, from the first apparatus, information based on the at least one matching factor; monitor the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; generate, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching. The expected accuracy may be further based on at least one positioning accuracy parameter. The means configured to monitor the at least one matching factor may be configured to obtain at least one threshold matching factor corresponding to at least one of: at least one channel characteristic; and at least one positioning accuracy parameter. The means configured to obtain at least one threshold matching factor may be configured to obtain at least one threshold matching factor based on a look-up table. The means configured to monitor the at least one matching factor may be configured to determine if the at least one matching factor is lower than the at least one threshold matching factor, wherein the monitoring decision request is generated based on the determination. The at least one matching factor may be based on a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase. The inputs may be transmission-reception-point measurements for a location management service or positioning determination. The means may further be configured to subscribe, with respect to a third apparatus, to a positioning analysis service; generate, for the third apparatus, machine learning functionality information; obtain, from the third apparatus, at least one matching factor based rule for generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor. The first apparatus may be one of: a user equipment; a gNB. The second apparatus may be one of: a gNB; a network function; and a LMF. The third apparatus may be a network data analytics function. According to a fifth aspect there is provided a first apparatus comprising: at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: receive, from the second apparatus, a request to initiate obtain at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtain the at least one matching factor based on the request to initiate obtaining the at least one matching factor; perform at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic. The monitoring decision request may be further be based on at least one positioning accuracy parameter. The apparatus caused to obtain, in the first apparatus, at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase may be caused to: identify inputs used during the machine learning inference phase; identify inputs used during the machine learning training phase; and generate the at least one matching factor based a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase. The inputs may be transmission-reception-point measurements for a location management service or positioning determination. The first apparatus may be one of: a user equipment; a gNB. The second apparatus may be one of: a gNB; a network function; and a LMF. According to a sixth aspect there is provided there is provided a second apparatus comprising: at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: generate, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; obtain, from the first apparatus, information based on the at least one matching factor; monitor the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; generate, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching. The expected accuracy may be further based on at least one positioning accuracy parameter. The apparatus caused to monitor the at least one matching factor may be caused to obtain at least one threshold matching factor corresponding to at least one of: at least one channel characteristic; and at least one positioning accuracy parameter. The apparatus caused to obtain at least one threshold matching factor may be caused to obtain at least one threshold matching factor based on a look-up table. The apparatus caused to monitor the at least one matching factor may be caused to determine if the at least one matching factor is lower than the at least one threshold matching factor, wherein the monitoring decision request is generated based on the determination. The at least one matching factor may be based on a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase. The inputs may be transmission-reception-point measurements for a location management service or positioning determination. The apparatus may be further caused to subscribe, with respect to a third apparatus, to a positioning analysis service; generate, for the third apparatus, machine learning functionality information; obtain, from the third apparatus, at least one matching factor based rule for generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor. The first apparatus may be one of: a user equipment; a gNB. The second apparatus may be one of: a gNB; a network function; and a LMF. The third apparatus may be a network data analytics function. According to a seventh aspect there is provided a first apparatus comprising: means for receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; means for obtaining the at least one matching factor based on the request to initiate obtaining the at least one matching factor; means for performing at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic.. According to an eighth aspect there is provided a second apparatus comprising: means for generating, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; means for obtaining, from the first apparatus, information based on the at least one matching factor; means for monitoring the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; means for generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching. According to a ninth aspect there is provided a computer program comprising instructions [or a computer readable medium comprising program instructions] for causing a first apparatus to perform at least the following: receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtaining the at least one matching factor based on the request to initiate obtaining the at least one matching factor; performing at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic. According to a tenth aspect there is provided a computer program comprising instructions [or a computer readable medium comprising program instructions] for causing a second apparatus to perform at least the following: generating, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; obtaining, from the first apparatus, information based on the at least one matching factor; monitoring the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching. According to an eleventh aspect there is provided a non-transitory computer readable medium comprising program instructions for causing a first apparatus to perform at least the following: receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtaining the at least one matching factor based on the request to initiate obtaining the at least one matching factor; performing at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic. According to a twelfth aspect there is provided a non-transitory computer readable medium comprising program instructions for causing a second apparatus to perform at least the following: generating, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; obtaining, from the first apparatus, information based on the at least one matching factor; monitoring the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching. According to a thirteenth aspect there is provided a first apparatus comprising: receiving circuitry configured to receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtaining circuitry configured to obtain the at least one matching factor based on the request to initiate obtaining the at least one matching factor; switching circuitry configured to perform at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic. According to a fourteenth aspect there is provided a second apparatus comprising: generating circuitry configured to generate for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; obtaining circuitry configured to obtain, from the first apparatus, information based on the at least one matching factor; monitoring circuitry configured to monitor the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; generating circuitry configured to generate, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching.. According to a fifteenth aspect there is provided a computer readable medium comprising program instructions for causing a first apparatus to perform at least the following: receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase; obtaining the at least one matching factor based on the request to initiate obtaining the at least one matching factor; performing at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic. According to a sixteenth aspect there is provided a computer readable medium comprising program instructions for causing a second apparatus to perform at least the following: generating, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase; obtaining, from the first apparatus, information based on the at least one matching factor; monitoring the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic; generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching. An apparatus comprising means for performing the actions of the method as described above. An apparatus configured to perform the actions of the method as described above. A computer program comprising program instructions for causing a computer to perform the method as described above. A chipset configured to implement the apparatus as described above. A computer program product stored on a medium may cause an apparatus to perform the method as described herein. According to an aspect, there is provided a non-transitory computer readable medium comprising program instructions for causing an apparatus to perform at least the method according to any of the preceding aspects. In the above, many different embodiments have been described. It should be appreciated that further embodiments may be provided by the combination of any two or more of the embodiments described above. DESCRIPTION OF FIGURES Embodiments will now be described, by way of example only, with reference to the accompanying Figures in which: Fig. 1 shows a representation of a network system according to some example embodiments; Fig. 2 shows a representation of a control apparatus according to some example embodiments; Fig. 3 shows a representation of an apparatus according to some example embodiments; Fig. 4 shows a schematic view of a functional framework for AI / ML for NR air interface; and Fig. 5 shows a schematic example scenario of UE and multiple TRPs in which some embodiments can be employed; Fig. 6 shows a graph of example errors produced in matched and unmatched inputs; Figs. 7 a to 7d shows example flow diagrams of high level operations for reactive and proactive matching factor methods according to some embodiments; Fig. 8 shows a graph of example errors produced in matched, partial matched and unmatched inputs; and Figs. 9 to 11 show flow diagrams of example matching factor determination and monitoring operations according to some embodiments. DETAILED DESCRIPTION In the following certain embodiments are explained with reference to apparatuses capable of communication with a communication system serving such apparatuses. Before explaining in detail the exemplifying embodiments, certain general principles of a communication system, for example a 5G communication system, that includes an access network (AN) and a core network, and apparatuses (e.g., terminals served by the communication system are briefly explained with reference to Figures 1,2 and 3 to assist in understanding the technology underlying the described examples. Figure 1 shows a schematic representation of a 5G wireless communication system (5GS). The 5GS may be comprised of a radio access network (RAN) (e.g., a 5G radio access network (5G-RAN) or a next generation radio access network (NG-RAN), a 5G core network (5GC), one or more application functions (Afs) and a more data network (DN). In some embodiments, an AF is a customer of the 5GC and is connected to a user plane function (UPF) of the 5GC via the DN and to network functions (NFs) of the 5GC via a network exposure function (NEF) of the 5GC. In some embodiments, the AF is a trusted application function and hence the trusted AF is implemented in the 5GC and connected to directly to other NFs of the 5GC. The AF may include functionality such as Location Management Function (LMF). The location management function (LMF), is central in the 5G positioning architecture. The LMF receives measurements and assistance information from the next generation radio access network (NG-RAN) and the mobile device, otherwise known as the user equipment (UE), via the access and mobility management function (AMF) over the NLs interface to compute the position of the UE. The LMF configures the UE using the LTE positioning protocol (LPP) via AMF. The NG RAN configures the UE using radio resource control (RRC) protocol over LTE-Uu and NR-Uu. The 5GC may comprise for instance the following network functions (NFs) (otherwise referred to as network entities): Network Slice Selection Function (NSSF); Network Exposure Function (NEF); Network Repository Function (NRF); Network Data Analytics Function (NWDAF), Policy Control Function (PCF); Unified Data Management (UDM); Authentication Server Function (AUSF); an Access and Mobility Management Function (AMF); and Session Management Function (SMF). The NFs of the 5GC may have a service-based architecture as described in TR 23.501 of the 3GPP standard. NF services that may be offered by the NFs of the 5GC and service-based interfaces for the NFs of the 5GC are described in 3GPP standard, and in particular in TR 23.501 and 23.502 of the 3GPP standard. Access to the 5GC by terminals may be done more generally via an access network, such as a 5G radio access network (5G-RAN). The 5G-RAN may comprise one or more base stations (e.g., gNodeBs (gNBs)). The gNBs of the 5G-RAN may include a gNB distributed unit connected to a gNB central unit, and remote radio heads connected to the gNB distributed units. In some embodiments, the one or more base stations of the 5G-RAN may be Evolved NodeB eNodeB (eNB). In some embodiments, the 5G-RAN may be a 3GPP radio access network (e.g. a RAN that operates using NR or LTE radio access technology as defined in the 3GPP standard). Figure 2 illustrates an example of an apparatus 200 that may implement one or more NFs of the 5GC, such as LMF illustrated in Figure 1. The apparatus 200 may comprise at least one random access memory (RAM) 211a, at least one read only memory (ROM) 211b, at least one processor 212, 213 and a network interface 214. The at least one processor 212, 213 may be coupled to the RAM 211a and the ROM 211b. The at least one processor 212, 213 may be configured to execute software code 215. The software code 215 may for example include instructions to perform actions or operations of one or more NFs of the 5GC. In some embodiments, the software code 215 may include instructions to perform one or more actions or operations related to federated learning (FL) or privacy budget control in accordance with aspects of the present disclosure. The software code 215 may be stored in the ROM 211b. The apparatus 200 may implement one or more NFs of the 5GC and may be interconnected with another apparatus 200 implementing one or more other NFs of the 5GC. In such embodiments, the apparatuses 200 may be part of a distributed computing system. In some embodiments, each NF of the 5GC may be implemented on a single apparatus 200. In such embodiments, the apparatus 200 may be a cloud computing system. Figure 3 illustrates an example of an apparatus 300 illustrated on Figure 1. The apparatus 300 may be any wireless communication device capable of sending and receiving radio signals. Non-limiting examples of an apparatus 300 comprise a terminal, a wireless communication device, user equipment, a mobile station (MS) or mobile device such as a mobile phone or what is known as a ’smart phone’, a computer provided with a wireless interface card or other wireless interface facility (e.g., USB dongle), a personal data assistant (PDA) or a tablet provided with wireless communication capabilities, a machine-type communications (MTC) device, an Internet of things (loT) communication device or any combinations of these or the like. The apparatus 300 may be configured to communicate with base stations (e.g., a eNB for a 4G network or a gNB for a 5G network) of an access network, such as the 5G-RAN and the 5GC via the base stations of the 5G-RAN using non-access stratum (NAS) signalling, for example, to communicate of data. The communications may include or carry one or more of voice, electronic mail (email), text message, multimedia, data, machine data and so on. The apparatus 300 may receive wireless signals (e.g., radio or cellular signals) over an air or radio interface 307 (generally referred to as a Uu interface) via appropriate apparatus 306 for receiving the wireless signals and may transmit wireless signals e.g., radio or cellular signals) via appropriate apparatus for transmitting the wireless signals. In Figure 3 the apparatus 306 includes one or more antennas (or an antenna array comprising a plurality of antennas) and a transceiver and is designated schematically by block 306. The apparatus 306 may be provided for example by means of a radio part and associated antenna arrangement comprising one or more antennas. The antenna arrangement may be arranged internally or externally to the mobile device. The apparatus 300 may include at least one processor 301, at least one memory ROM 302a, at least one RAM 302b and other possible components 303 for use in software and hardware aided execution of tasks it is designed to perform, including control of access to and communications with access networks, such as the 5G-RAN, and other apparatuses 300. The at least one processor 301 is coupled to the RAM 311a and the ROM 311b. The at least one processor 301 may be configured to execute an appropriate software code 308. The software code 308 may for example include instructions which when executed by the at least one processor 301 perform one or more actions or operations of present aspects. The software code 308 may be stored in the ROM 311b. The at least one processor 301, storage and other relevant control apparatus can be provided on an appropriate circuit board and / or in chipsets. This feature is denoted by reference 304. The terminal 300 may optionally have a user interface such as key pad 305, touch sensitive display screen or touch sensitive pad, combinations thereof or the like. Optionally one or more of a display, a speaker and a microphone may be provided depending on the type of the device. As discussed above in response to the requirements and complexities found in 5G positioning technologies and the persistent challenges in (non-line-of-sight) NLOS and complex (line-of-sight) LOS scenarios, Release 18 of 3GPP initiated an AI / ML study item to explore the potential of Al and machine learning in the context of positioning. This study, as provided in technical report TR 38.843 has identified two primary modalities through which AI / ML could be beneficial: Direct AI / ML Positioning where the AI / ML model directly outputs the location of the user equipment (UE) and which can employ channel observations as the AI / ML model’s input to ascertain the UE’s location; and AI / ML Assisted Positioning, where rather than directly determining the UE’s location, the AI / ML model aids conventional approaches by outputting measurements or by refining existing measurements. Furthermore the technical report discusses signalling and protocol aspects of Life Cycle Management (LCM) enabling functionality and model (if justified) selection, activation, deactivation, switching, and fallback. Additionally, the technical report discusses signalling / mechanism(s) for LCM to facilitate model training, inference, performance monitoring, data collection for both UE-sided and NW-sided models. A specific positioning accuracy enhancement disclosed was enabling method(s) to ensure consistency between training and inference regarding NW-side additional conditions (if identified) for inference at UE for relevant positioning cases. An example of a functional framework for Al / M L for a 5G / N R air interface is shown in Fig. 4. Figure 4 is a copy of Figure 4.4-1 from TR 38.843 v2.0.1. For example Fig. 4 shows a data collection function 400. The data collection function 400 is a function that provides input data to the model training function 402 (training data 401), management function (monitoring data 403), and inference function 406 (inference data 405). The training data 401 is data needed as input for the AI / ML Model Training function 402, the monitoring data 403 is data needed as input for the Management of AI / ML models or AI / ML functionalities and the inference data 405 is data needed as input for the AI / ML Inference function 406. In some examples the model training function 402 is a function that performs AI / ML model training, validation, and testing which may generate model performance metrics which can be used as part of the model testing procedure. The model training function 402 is also responsible for data preparation (e.g., data pre-processing and cleaning, formatting, and transformation) based on Training Data delivered by a Data Collection function 400. The model training function 402 can furthermore output the trained / Updated Model 411 to a Model Storage function 408 where available. Some examples feature a management function 404, which is a function that oversees the operation (e.g., selection / (de)activation / switching / fallback) and monitoring (e.g., performance) of AI / ML models or AI / ML functionalities. This function 404 can also be responsible for making decisions to ensure the proper inference operation based on data received from the Data Collection function 400 and the Inference function 406. For example, the management function 404 can be configured to provide a Management Instruction 409b to the inference function 406 as an input to manage the Inference function 406. This may include selection / (de)activation / switching of AI / ML models or AI / ML-based functionalities, fallback to non-AI / ML operation. The management function 404 can furthermore provide model transfer / delivery request 413 which can be used to request model(s) to the Model Storage function. The management function can furthermore provide performance feedback / retraining request 407 to the model training function 402 as information needed as input for the model training function 402 for model (re)training or updating purposes. An inference function 406 is a function that provides outputs from the process of applying AI / ML models or AI / ML functionalities, using the data that is provided by the Data Collection function 400 (i.e., Inference Data 405) as an input. The Inference function 406 is also responsible for data preparation (e.g., data pre-processing and cleaning, formatting, and transformation) based on Inference Data 405 delivered by a Data Collection function 400, if required. The Inference Output 409a is data used by the Management function 404 to monitor the performance of AI / ML models or AI / ML functionalities. The model storage function 408 can furthermore output model transfer / delivery in other words to deliver an AI / ML model to the Inference function 406. The data collection function 400 for example can be configured to generate channel signal information from multiple base stations within the communications environment. These base stations can be referred as transmission-reception-points (TRPs). For example, Fig. 5 shows an example communications environment 501 within which there are located 18 base stations (TRPs) 502 on a square lattice with spacing D, located D / 2 from the walls. The Fig. 5 is based on an example TRP layout and their indices for the evaluation of AI / ML based positioning as presented in Figure 6.4.1-5 of TR 38.843 v2.0.1. These TRPs 502 can be identified by a suitable unique coding or labelling (shown here in Fig. 5 by the labels 0 to 17). Additionally, Fig. 5 shows a location of the UE 500. In some situations, a selection of the number of inputs or TRP measurements can be used for the inference function. For example, the TRP measurements from the TRPs identified within the area 501 can be the inputs for the inference function. This selection can lead to dynamic TRP selections as inputs to the AI / ML models. The inference accuracy can vary considerably depending on the TRPs involved in the training (in terms of which TRPs are selected and also the order of selected TRPs for the AI / ML Model input). Additionally, if the LMF has a hard requirement in terms of positioning accuracy (for example in attempting to meet a QoS accuracy factor) then the LMF cannot provide a confident inference model accuracy in dynamic TRP selection. This is because the achievable positioning accuracy will largely vary depending on a matching or correlation between the considered TRP combination in the inference and the ones included in the training dataset. In other words, the matching or correlation between the selection inputs can affect the positioning accuracy of the model. As an example, the following scenario highlights this issue. The measurement campaign (or data collection) is performed in a real world scenario where N radio measurements (such as CSI or CIR measurements) are collected from an UE to the gNB at multiple TRPs, in M known locations (i.e., reference points). In the following example graphs as shown in Fig. 6 and Fig. 8, a site which has an “L-shape” is used as the environment, where the precise location of the M reference points (i.e., ground truth points) are known, and the precise location of the N = 4 TRPs is also known. In this example the site is densely populated with metallic surfaces / structures, leading to a harsh radio environment (in terms of multi-path propagation). Furthermore, the N TRPs are not always in LOS with the UE device to be localized (in other words detailing a non-friendly positioning scenario). In this example the UE device to be localized is placed on a reference point for a duration. The UE device transmits a 5G SRS signal. The N TRPs receive the transmitted signal and, thus, radio measurements per locator are collected. Then the UE device moves to the next reference point and the process is repeated. The measured data can be used to create our machine learning datasets including: 1-) radio measurements from the N TRPs; 2-) UE ground truth position (i.e., reference point location); and 3-) the precise location of the N TRPs. In this example a Neural Network (NN) is trained while considering all 4 TRPs and passing the radio measurements (to the ML model) from the 4 TRPs in the following order: [TRP 1, TRP 2, TRP 3, TRP 4], In inference, the pre-trained model is employed to determine a location estimate. The input to the inference function (the trained NN) can comprise the radio measurements from the same TRPs 1 / 2 / 3 / 4. With respect to Fig. 6 is shown an example positioning error graph where the Cumulative distribution function (CDF) against 2D positioning error is plotted 605 for a complete match where the inference model input is also 4 TRPs in the same order: [TRP 1, TRP 2, TRP 3, TRP 4] where the cumulative error is 1 m and a further plot 607 for a zero match (complete mismatch) where the inference model input is also 4 TRPs in the order: [TRP 3, TRP 4, TRP 2, TRP 1] and where the cumulative error is nearly 7m. In other words, the positioning accuracy degrades due to the input selection differing between the inference and training functions and specifically due to the change of the order of TRP radio measurements. The embodiments, as described in further detail herein, thus show apparatus and methods which aim to assess any mismatch between the inputs (or configuration related to TRPs involved in inference and training functions) and furthermore enable monitoring of the input selection accordingly to ensure a required positioning accuracy (i.e., to assess and monitor whether a required QoS can be met). This aim, in some embodiments, can be implemented by method and apparatus implementing related signaling enhancements which target monitoring without ground truth for ML based positioning. For example a metric referred as (TRP) matching factor can be defined. The TRP matching factor indicates a degree of matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase. This (TRP) matching factor can furthermore be employed in a suitable mapping between the matching between TRPs involved in training and inference (the matching factor) to the targeted positioning accuracy. In some embodiments the mapping can furthermore involve additional factors associated with the communications environment (such as at least one channel characteristic, LOS / NLOS). In some embodiments the objective is to use this metric to realize monitoring decisions including ML model switching. As such there can be performed at least a machine learning model switching based on a received monitoring decision request, where the monitoring decision request is based at least on: the obtained at least one matching factor (which can also be known as an input matching factor or TRP matching factor); and at least one channel characteristic. In some embodiments the monitoring based on the metric (the TRP matching factor) can be assisted by a suitable signaling enhancement. The signaling enhancement can for example include the network (for example the LMF or any suitable function) providing assistance towards the entity performing the AI / ML model inference function (for example either a UE or gNB) to select the best AIML model with a specific input (TRP) selection through a model switching procedure. With respect to Figs 7a and 7b are shown example flow diagrams from the viewpoint of the UE / gNB and LMF respectively of an overview of the example embodiments. Fig. 7a shows the operations of the UE or gNB aspects according to some embodiments. For example the UE or gNB (optionally) receiving, by 751, a request to initiate determining at least one matching factor. The request can, for example, comprise at least one of: indicating how to determine the at least one matching factor; and at least one threshold matching factor to compare against. As the operation 751 is optional, the definition of matching factor is made below in operation 753. Then is shown by 753 obtaining at least one matching factor, the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase. Furthermore, optionally, as shown by 755 is informing a second apparatus the obtained at least one matching factor. Otherwise, optionally, and shown by 757 is the operation of comparing the obtained at least one matching factor against a threshold matching factor. Then is shown by 759 performing at least a machine learning model switching based on a monitoring decision, the monitoring decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic (and optionally at least one positioning accuracy parameter). With respect to Fig. 7b is shown the LMF aspects of some embodiments. Thus Fig. 7b shows, the LMF (optionally) obtaining, by 761, a rule for determining and processing a matching factor. This rule can for example be provided by a further apparatus such as a NWDAF. Then is shown, by 763, generating, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates the degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase. Further is shown by 765, obtaining information based on the obtained at least one matching factor. For example, the information can be at least one matching factor, for a comparison with a determined threshold matching factor, or a comparison result from the comparison. Also optionally is shown determining, by 767, at least one monitoring decision and generating a decision request based at least on: the obtained at least one matching factor; and at least one channel characteristic (and optionally at least one positioning accuracy parameter). In the following discussion the interaction between the UE / gNB and the LMF according to some embodiments can be implemented according to the two cases which are considered and described in detail and shown with respect to Figs. 7c and 7d. As shown with respect to Fig. 7c is a flow diagram of the implementation of a reactive approach to monitoring based on the matching factor metric. This flow diagram is from the viewpoint of the first apparatus (for example the UE or gNB), The reactive approach has advantages at the entity in charge of model inference which could be the UE or the gNB in that the UE or gNB is not required to perform any monitoring operations. As such, the entity or first apparatus (for example the UE or gNB), is configured to obtain information which can assist or configure the matching factor calculation or determination. For example the information can comprise the indices identifying specific (TRP) inputs used in the training phase. This information can be found within a received request, such as shown by 701 in Fig. 7c, from a second apparatus, such as a LMF which requests the first apparatus or entity in charge of model inference (UE or gNB) to compute the (TRP) matching factor. The first apparatus or entity in charge of model inference (UE or gNB) can then, for each inference operation, obtain (or compute) the (TRP) matching factor as shown by 703 in Fig. 7c. The determined, computed or otherwise obtained (TRP) matching factor (or information based on the matching factor) can then be passed to the second apparatus, for example the LMF, as shown by 705 in Fig. 7c. Therefore, based on the received computed (TRP) matching factor, the second apparatus (for example the LMF) can estimate whether a targeted accuracy is probable to be reached or not, and then generate monitoring decisions accordingly. These decisions can be received, for example within a monitoring decision request can be received by the first apparatus, as shown by 707 in Fig. 7c. The determination of whether, based on the (TRP) matching factor, the targeted accuracy is probable and generating suitable monitoring decisions are described in further detail later. Having received the monitoring decision request, then the first apparatus can perform the monitoring decision request as shown by 709. The request can be for model switching as discussed in further detail herein. As shown in Fig. 7d is a flow diagram of an implementation of a proactive approach to monitoring. This flow diagram is similarly from the viewpoint of the first apparatus (for example the UE or gNB), The proactive approach has advantages as any signaling between the entity and any monitoring entity (for example the LMF, gNB or a suitable core network function) is reduced to situations where any quality issue is detected. As such, the entity or first apparatus (for example the UE or gNB), is configured to obtain, for example receive, information which can assist or configure the matching factor calculation or determination. For example, the information can comprise the indices identifying specific (TRP) inputs used in the training phase. Additionally, the information can comprise a (TRP) matching factor requirement or matching factor threshold which defines a threshold which defines whether the targeted accuracy is probable (within the wireless communications environment and as defined by a channel characteristic). This information can be found within a received request, such as shown by 711 in Fig. 7d, from a second apparatus, such as a LMF or gNB or suitable core network entity which requests the first apparatus or entity in charge of model inference (UE or gNB) to compute the (TRP) matching factor. The first apparatus or entity in charge of model inference (UE or gNB) can then, before determining an inference operation, obtain (or compute) the (TRP) matching factor as shown by 713 in Fig. 7d. The determined or computed (TRP) matching factor can then be compared to the received (TRP) matching factor requirement or matching factor threshold as shown by 715 in Fig. 7d. Where the received (TRP) matching factor requirement or matching factor threshold is above the determined or computed (TRP) matching factor then the inference operation can be stopped or paused and the second apparatus informed such as shown by 719 in Fig. 7d, which can then lead to the second apparatus generated and transmitting a suitable monitoring decision request. Where the received (TRP) matching factor requirement or matching factor threshold is below the determined or computed (TRP) matching factor then the inference operation can be performed. In such a manner the proactive approach enables the second apparatus or entity (for example the LMF) to indicate in advance the required TRP matching between training and inference to ensure minimum guaranteed level of positioning accuracy following targeted usage. For example, a guaranteed level of positioning accuracy could be 0.5 m for V2X applications. In the proactive example, the entity in charge of ML inference (UE or gNB) can then compare the determined current TRP matching factor to the required TRP matching factor (sent by the LMF) and based on this comparison inform the LMF when there is likely to be a shortfall in the accuracy. In some embodiments the first apparatus or entity is an UE and the second apparatus or entity the gNB or LMF. However it would be understood that in some embodiments the first apparatus is the gNB and the second apparatus the LMF or a suitable core network entity. 5 A determination of the (TRP) matching factor as discussed above can be implemented by determining a degree of matching or similarity between the inputs of the training phase and the inputs of the inference phase. In some embodiments this matching factor can be simple count or determination of the number of matching inputs. In other words, the matching factor can be generated 10 by comparing inputs of the training and inference phase and incrementing a count value where the same TRP index is found for the input of the training phase and the inference phase. In some embodiments the matching factor can be normalized relative to the number of inputs. Furthermore the matching factor can, in some embodiments, be 15 determined or obtained as a % matching factor value. For example, in the following table is shown a TRP matching factor computation example where the training input used [TRP1,TRP2,TRP3,TRP4] but the inference input can use these in the same order, different order or different TRPs:: Training Input Inference input Mate h factor count Normalize d Matching factor Matchin g factor % [TRP1 ,TRP2,TRP3,TRP 4] [TRP1,TRP2,TRP3,TRP 4] 4 4 / 4= 1 100% [TRP1,TRP2,TRP3,TRP 4] [TRP1,TRP2,TRP4,TRP 3] 2 2 / 4=0.5 50% [TRP1,TRP2,TRP3,TRP 4] [TRP2,TRP1,TRP4,TRP 3] 0 0 / 4=0 0% [TRP1 ,TRP2,TRP3,TRP 4] [TRP1,TRP2,TRP3,TRP 9] 3 3 / 4=0.75 75% [TRP1,TRP2,TRP3,TRP 4] [TRP1,TRP2,TRP7,TRP 8] 2 2 / 4=0.5 50% [TRP1,TRP2,TRP3,TRP 4] [TRP1,TRP6,TRP9,TRP 2] 1 1 / 4=0.25 25% [TRP1,TRP2,TRP3,TRP 4] [TRP5,TRP8,TRP7,TRP 1] 0 0 / 4=0 0% Although this example shows a factor generated based on 4 inputs, it is understood that the matching factor can be determined for any suitable number of inputs. In some embodiments the matching factor can be determined by any other suitable method. For example by determining the number of members of the intersection between the training TRP input set and the inference TRP input set and normalizing this number relative to the input set size. The monitoring based on the (TRP) matching factor determination is briefly discussed above and can be implemented by determining a threshold or minimum (TRP) matching factor against which the determined (TRP) matching factor is compared. In other words, a monitoring criteria could be expressing by comparing the calculated TRP matching factor to a pre-selected (or otherwise obtained) minimum or threshold value. Thus, for example: 100% means that the TRP combination of the inference is already included in training dataset; and 50% means that only half of the combination I indices of the inference model is included in the training dataset. The determination of the minimum or threshold matching between TRP indices in inference and training datasets can employ any suitable method and be generated based on one or more input such as targeted accuracy, communication environment factors such as channel characteristics or other information associated with the communication environment, such as whether the environment is significantly LOS or NLOS or the environment is a countryside or urban environment or the building type (low floor or high floor building, small size or big size building) in the environment. An example table, presented below, shows an example of rules which can be established by a LMF or other entity of apparatus to identify the minimum required or threshold (TRP) matching factor value in order to ensure the targeted accuracy (e.g., 1m) is met given the external factors (e.g., channel characteristics LOS or NLOS probability averaged over involved TRPs). Targeted accuracy Channel Characteristics Required (minimum or threshold) TRP matching factor 2m LOS 30% 2m NLOS 50% 1 m LOS 50% 0.5 m LOS 100% An example effect of a matching factor between 0% and 100% is shown in Fig. 8 where a “kicking-out” of a TRP during inference compared to the training inputs is shown. Thus, the training is the same as the example shown in Fig. 6, where the 4 TRPs while passing the radio measurement to the ML model are in the order: [TRP 1, TRP 2, TRP 3, TRP 4] (in other words the same pre-trained NN from the previous example is generated). Fig. 8 shows the 0% 607 and the 100% 605 matching factor results and furthermore shows an example inference positioning error plot 801 for the following input selections: the order of the TRPs stays the same as training; and zero pad the radio measurements of TRP 4. This results in the equivalent of using only 3 TRPs for inference: [TRP 1, TRP 2, TRP 3] or a matching factor of 75%. The simulation results shown in Fig. 8 highlights the effect / importance of the TRP matching factor on the positioning accuracy (i.e., the TRP matching factor can be directly inferred from the Fig. 8). The targeted accuracy input can be expressed as the maximum positioning error for the inference to be performed. In some examples the targeted accuracy input can be also expressed as a range, e.g., a minimum and / or maximum value. The calculation of the expected accuracy for a matching value can be determined based on the specific implementation. For example, the expected accuracy determination can be based on 90%-ile of accuracy calculated from training dataset or past positioning sessions. The monitoring (or as shown in Fig. 4, the management function 404) can then, based on the comparison between the matching factor and the minimum or threshold matching factor, control the operation of the AI / ML inference operation. For example to enable a model switching which includes the operations of a selection of a model, a (de)activation of a model, a switching or reordering of inputs of a model, or trigger data collection (to include the cases with other inputs / order) and model re-training, or a fallback operation where the use of the model is switched off. With respect to Fig. 9 these is shown in further detail signaling between a first apparatus, the UE / gNB 900 and a second apparatus LMF 902. In this approach, the entity in charge of ML model inference (UE or gNB) 900 receives a request as shown by 901 in Fig. 9, from the LMF 902 to compute the new monitoring metric (referred as the TRP matching factor). This request can also indicate to the entity UE or gNB 900 the reporting conditions of this new monitoring metric (e.g., each inference iteration, or each time lower than certain threshold value or a value aggregated over multiple inference iterations to decrease reporting overhead). Then as shown by 903 in Fig. 9, the first apparatus or entity (UE or gNB) 900 obtains or receives the required input model data (for example radio measurements) from various available TRPs and is configured to perform ML inference. Furthermore, as shown by 905 in Fig. 9, the first apparatus or entity (UE or gNB) 900 is configured to additionally determine or compute the TRP matching factor (for example using the method shown above) as informed by the request. As shown by 907 in Fig. 9 the first apparatus or entity (UE or gNB) 900 sends a computed monitoring metric (for example based on TRP matching factor value) to the second apparatus or LMF 902. In some embodiments this can be sent where reporting conditions are verified as indicated by LMF. The second apparatus or entity (LMF) 902 can therefore assess, as shown by 909 in Fig 9, the expected impact of the shared monitoring metric on the ML model performance. This can be, as discussed above, implementation related. For example a determined minimum or threshold matching factor is determined based on the required accuracy and other environmental or external factors and compared against the received matching factor from the UE or gNB 900. Then, as shown by 911 in Fig.9, based on assessed ML model performance, the LMF 902 can determine or realize monitoring decisions. For example, if the model accuracy is expected to be degraded due to low TRP matching factor values, then the LMF can decide to request UE to proceed with model switching or if no other model is available to proceed with model retraining (including additional TRP combinations) or fallback to non-ML methods. Finally, as shown by 913 in Fig. 9, the LMF informs the inference-entity (the UE or gNB 900) about the monitoring decision where the monitoring request is processed. The first apparatus, for example the UE or gNB can then perform model switching based on the monitoring decision request as shown by 915. As discussed above the model switching operation can include triggering model re-training with additional data. The advantage of this approach is that it is compatible with the case where model training and management is handled by the first apparatus or entity. The LMF in this scenario has the role of providing assistance by means of this new monitoring metric and the monitoring decisions are therefore made according to the targeted positioning accuracy / QoS. With respect to Fig. 10 there is shown in further detail signaling between a first apparatus, the UE / gNB 900 and a second apparatus LMF (or in some embodiments a gNB or other suitable core network entity) 902. In this approach, the first apparatus or entity (UE or gNB 900) is in charge of ML model inference. The first apparatus or entity (UE or gNB 900) as shown by 1001 in Fig. 10 is configured to generate and transmit to the second apparatus (LMF 902) an indication of inference approach (that it is about to implement an inference AI / ML phase). The LMF can then be configured to determine, as shown by 1003 in Fig. 10, to perform an analysis and select requirements for target positioning accuracy, for example determine a minimum or threshold matching factor in a manner as described above (for example in accordance with the targeted positioning accuracy and / or external factors such as channel characteristics). The LMF can then send a request to the first apparatus or entity (the UE or gNB), as shown by 1005 in Fig. 10, with theTRP matching factor threshold value with additional indications including configuration and reporting conditions. For example a number of samples over which the entity informs the LMF and if inference is allowed or not if the calculated monitoring metric is below the indicated threshold value. In some embodiments, the LMF requests the UE to report the difference between the UE’s determined matching factor and the required minimum or threshold matching factor indicated by LMF. For example whether the matching factor difference is -20% or +20%. The entity in charge of ML model inference (UE or gNB) 900 therefore receives this request, obtains or receives the required input model data (for example radio measurements) from various available TRPs and, as shown by 1007 in Fig. 10, the first apparatus or entity (UE or gNB) 900 is configured to additionally obtain (for example determine or compute or otherwise calculate) the TRP matching factor (for example using the method shown above) as informed by the request. As shown by 1009 in Fig. 10, the first entity or entity (UE or gNB) 900 is configured to check the TRP matching condition as requested by LMF before proceeding to the ML model inference. In other words, the obtained matching factor is compared to the (predetermined or otherwise obtained) minimum or threshold matching factor to determine whether the inference operation is to be continued (the determined matching factor being greater than the obtained minimum or threshold matching factor) or to implement an informing operation as shown by 1011 in Fig. 10. The informing operation as shown by 1011 in Fig. 10 (the determined matching factor being less than the obtained minimum or threshold matching factor) can cause the first apparatus or entity (UE or gNB) 900 to stop or pause the inference operation and inform the LMF that the matching factor requirement has not been met. This can be implemented by any suitable manner, for example in response to the earlier request. The second apparatus or entity (LMF) 902 can therefore assess, as shown by 1013 in Fig. 10, the expected impact on the ML model performance caused by the matching factor requirement not being met and based on assessed ML model performance, the LMF 902 can determine or realize monitoring decisions. For example, if the model accuracy is expected to be degraded due to low TRP matching factor values, then the LMF can decide to request UE to proceed with model switching or if no other model is available to proceed with model retraining (including additional TRP combinations) or fallback to non-ML methods. Following this, as shown by 1015 in Fig. 10, the LMF informs the inference-entity (the UE or gNB 900) about the monitoring decision where the monitoring request is processed. The first apparatus, for example the UE or gNB can then perform model switching based on the monitoring decision request as shown by 1017. In some embodiments, rather than informing the second apparatus (for example the gNB, LMF or suitable core network entity), the first apparatus (for example the UE or gNB) is configured to take the monitoring decision itself. As such, in these embodiments, there is no informing operation to an external apparatus nor is a monitoring process request (for model switching) received from an external apparatus or entity. In some embodiments the determination of the LMF rules can be implemented by a further or third apparatus or entity. An example of which is shown in Fig. 11, which shows a reactive example with additional signaling determining the matching factor based rules. It would be appreciated that determination of LMF rules by a further or third apparatus or entity can also be applied to the proactive example shown in Fig. 10. As such as shown by 1101 in Fig. 11 the LMF 902 is configured to subscribe to a new service (for example a V2X service) with respect to the generation and definition of UE positioning rules (for example defining a maximum error for UE positioning for the service). The request can be sent to a Network Data Analytics Function (NWDAF) 1100. Then as shown by 1103 in Fig. 11 the LMF is further configured to send ML functionality related information (including the labeled dataset) to the NWDAF 1100. The NWDAF 1100 can then be configured to analyse the received information and establish TRP matching factor based rules as shown by 1105 in Fig. 11. These TRP matching factor based rules can then as shown by 1107 in Fig. 11 be sent back to the LMF 902. The entity in charge of ML model inference (UE or gNB) 900 receives a request as shown by 901 in Fig. 11, from the LMF 902 to compute the new monitoring metric (referred as the TRP matching factor). This request can also indicate to the entity UE or gNB 900 the reporting conditions of this new monitoring metric (e.g., each inference iteration, or each time lower than certain threshold value or a value aggregated over multiple inference iterations to decrease reporting overhead). Then as shown by 903 in Fig. 11, the first apparatus or entity (UE or gNB) 900 obtains or receives the required input model data (for example radio measurements) from various available TRPs and is configured to perform ML inference. Furthermore, as shown by 905 in Fig. 11, the first apparatus or entity (UE or gNB) 900 is configured to additionally determine or compute the TRP matching factor (for example using the method shown above) as informed by the request. As shown by 907 in Fig. 11 the first apparatus or entity (UE or gNB) 900 sends a computed monitoring metric (for example based on TRP matching factor value) to the second apparatus or LMF 902. In some embodiments this can be sent where reporting conditions are verified as indicated by LMF. The second apparatus or entity (LMF) 902 can therefore assess, as shown by 909 in Fig. 11, the expected impact of the shared monitoring metric on the ML model performance. This can be, as discussed above, implementation related. For example a determined minimum or threshold matching factor is determined based on the required accuracy and other environmental or external factors and compared against the received matching factor from the UE or gNB 900. Then, as shown by 911 in Fig. 11, based on assessed ML model performance, the LMF 902 can determine or realize monitoring decisions. For example, if the model accuracy is expected to be degraded due to low TRP matching factor values, then the LMF can decide to request UE to proceed with model switching or if no other model is available to proceed with model retraining (including additional TRP combinations) or fallback to non-ML methods. Finally, as shown by 913 in Fig. 11, the LMF informs the first apparatus or entity, the inference-entity (the UE or gNB 900) about the monitoring decision where the monitoring request is processed. The first apparatus, for example the UE or gNB can then perform model switching based on the monitoring decision request as shown by 915. As discussed above a first apparatus or entity, for example the inference entity, can be implemented by a UE when the second apparatus or entity, for example the monitoring entity, can be implemented by a LMF. However the first apparatus can be implemented by a UE and the second apparatus be implemented by a gNB or suitable core network entity. In some embodiments the first apparatus can be implemented by a gNB and the second apparatus be implemented by a LMF or suitable core network entity. It is noted that whilst some embodiments have been described in relation to 5G systems, similar principles can be applied in relation to other networks and communication systems. Therefore, although certain embodiments were described above by way of example with reference to certain example architectures for radio access and core networks, radio access technologies and standards, embodiments may be applied to any other suitable forms of communication systems that implement other radio access technologies than those illustrated and described herein. It is also noted herein that while the above describes example embodiments, there are several variations and modifications which may be made to the disclosed solution without departing from the scope of the present invention. In general, the various embodiments, the apparatus described above may be implemented in hardware or special purpose circuitry, software, logic or any combination thereof. For example the apparatus can be implemented as a chipset. Some aspects of the disclosure may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device, although the disclosure is not limited thereto. While various aspects of the disclosure may be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof. As used in this application, the term “circuitry” may refer to one or more or all of the following: (a) hardware-only circuit implementations (such as implementations in only analog and / or digital circuitry) and (b) combinations of hardware circuits and software, such as (as applicable): (i) a combination of analog and / or digital hardware circuit(s) with software / firmware and (ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions) and (c) hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation. This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and / or firmware. The term circuitry also covers, for example and if applicable to the particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in server, a cellular network device, or other computing or network device. The embodiments of this disclosure may be implemented by computer software executable by a data processor of the mobile device, such as in the processor entity, or by hardware, or by a combination of software and hardware. Computer software or program, also called program product, including software routines, applets and / or macros, may be stored in any apparatus-readable data storage medium and they comprise program instructions to perform particular tasks. A computer program product may comprise one or more computer-executable components which, when the program is run, are configured to carry out embodiments. The one or more computer-executable components may be at least one software code or portions of it. Further in this regard it should be noted that any blocks of the logic flow as in the Figures may represent program steps, or interconnected logic circuits, blocks and functions, or a combination of program steps and logic circuits, blocks and functions. The software may be stored on such physical media as memory chips, or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, and optical media such as for example DVD and the data variants thereof, CD. The physical media is a non-transitory media. The memory may be of any type suitable to the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. The data processors may be of any type suitable to the local technical environment, and may comprise one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASIC), FPGA, gate level circuits and processors based on multi core processor architecture, as non-limiting examples. Embodiments of the disclosure may be practiced in various components such as integrated circuit modules. The design of integrated circuits is by and large a highly automated process. Complex and powerful software tools are available for converting a logic level design into a semiconductor circuit design ready to be etched and formed on a semiconductor substrate. The scope of protection sought for various embodiments of the disclosure is set out by the independent claims. The embodiments and features, if any, described in this specification that do not fall under the scope of the independent claims are to be interpreted as examples useful for understanding various embodiments of the disclosure. The foregoing description has provided by way of non-limiting examples a full and informative description of the exemplary embodiment of this disclosure. However, 5 various modifications and adaptations may become apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the appended claims. However, all such and similar modifications of the teachings of this disclosure will still fall within the scope of this invention as defined in the appended claims. Indeed, there is a further embodiment 10 comprising a combination of one or more embodiments with any of the other embodiments previously discussed.

Claims

1. A method for a first apparatus, the method comprising:receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase;obtaining the at least one matching factor based on the request to initiate obtaining the at least one matching factor;performing at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on:the obtained at least one matching factor; and at least one channel characteristic.

2. The method of claim 1, wherein the monitoring decision request is further based on at least one positioning accuracy parameter.

3. The method of any of claims 1 or 2, wherein obtaining, in the first apparatus, at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase comprises:identifying inputs used during the machine learning inference phase;identifying inputs used during the machine learning training phase; andgenerating the at least one matching factor based a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase.

4. The method of any of claims 1 to 3, wherein the inputs are transmissionreception-point measurements for a location management service or positioning determination.

5. The method of any of claims 1 to 4, wherein the first apparatus is one of:a user equipment;a gNB.

6. The method of any of claims 1 to 5, wherein the second apparatus is one of:a gNB;a network function; anda LMF.

7. A method for a second apparatus, the method comprising:generating, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase;obtaining, from the first apparatus, information based on the at least one matching factor;monitoring the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic;generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching.

8. The method of claim 7, wherein the expected accuracy is further based on at least one positioning accuracy parameter.

9. The method of any of claims 7 or 8, wherein monitoring the at least one matching factor comprises obtaining at least one threshold matching factor corresponding to at least one of: at least one channel characteristic; and at least one positioning accuracy parameter.

10. The method of claim 9, wherein obtaining at least one threshold matching factor comprises obtaining at least one threshold matching factor based on a look-up table.

11. The method of any of claims 9 or 10, wherein monitoring the at least one matching factor comprises determining if the at least one matching factor is lower than the at least one threshold matching factor, wherein the monitoring decision request is generated based on the determination.

12. The method of any of claims 7 to 11, wherein the at least one matching factor is based on a matching of the identified inputs used during the machine learning inference phase and identified inputs used during the machine learning training phase.

13. The method of any of claims 7 to 12, wherein the inputs are transmission-reception-point measurements for a location management service or positioning determination.

14. The method of any of claims 7 to 13, further comprising:subscribing, with respect to a third apparatus, to a positioning analysis service;generating, for the third apparatus, machine learning functionality information;obtaining, from the third apparatus, at least one matching factor based rule for generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor.

15. The method of any of claims 7 to 14, wherein the first apparatus is one of:a user equipment;a gNB.

16. The method of any of claims 7 to 15, wherein the second apparatus is one of:a gNB;a network function; anda LMF.

17. The method of any of claims 7 to 16, wherein the third apparatus is a network data analytics function.

18. A first apparatus, comprising: at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the first apparatus at least to:receive, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase;obtain the at least one matching factor based on the request to initiate obtaining the at least one matching factor;perform at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on:the obtained at least one matching factor; andat least one channel characteristic.

19. A second apparatus, comprising: at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the second apparatus at least to:generate, for a first apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase;obtain, from the first apparatus, information based on the at least one matching factor;monitor the at least one matching factor, to determine an expected accuracy based at least on: the at least one matching factor; and at least one channel characteristic;generate, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision request for performing, in the first apparatus, at least a machine learning model switching.

20. A computer program comprising instructions [or a computer readable medium comprising program instructions] for causing a first apparatus to perform at least the following:receiving, from the second apparatus, a request to initiate obtaining at least one matching factor, wherein the at least one matching factor indicates a degree matching between inputs used during a machine learning training phase and inputs used during a machine learning inference phase;obtaining the at least one matching factor based on the request to initiate obtaining the at least one matching factor;performing at least a machine learning model switching based on a monitoring decision request from a second apparatus, the monitoring decision request based at least on:the obtained at least one matching factor; andat least one channel characteristic.

21. A computer program comprising instructions [or a computer readable medium comprising program instructions] for causing a second apparatus to perform at least the following:generating, for a first apparatus, a request to initiate obtaining at least one 5 matching factor, wherein the at least one matching factor indicates a degree matching between inputs used by the first apparatus during a machine learning training phase and inputs used during a machine learning inference phase;obtaining, from the first apparatus, information based on the at least one matching factor;10 monitoring the at least one matching factor, to determine an expected accuracybased at least on: the at least one matching factor; and at least one channel characteristic;generating, for the first apparatus, at least one monitoring decision request based on the monitoring of the at least one matching factor, the at least one monitoring decision 15 request for performing, in the first apparatus, at least a machine learning model switching.36

Images