Machine learning aware relaxation of measurements reporting

The machine learning-aware approach for measurement relaxation in UE communication optimizes power consumption by adapting machine learning models to reduce measurement reporting, addressing the challenge of balancing reporting frequency and accuracy in wireless communication.

WO2024246826A1PCT designated stage expired Publication Date: 2024-12-05NOKIA TECHNOLOGIES OY

Patent Information

Application Number
PCT/IB2024/055298
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-06-01
Filing Date
2024-05-30
Publication Date
2024-12-05

AI Technical Summary

Technical Problem

In wireless communication, user equipment (UE) faces challenges in efficiently managing power consumption through measurement reporting, as frequent and extensive CSI reporting increases power consumption and signaling overhead, necessitating a method to reduce measurement reporting while maintaining performance accuracy.

Method used

A machine learning-aware approach that determines measurement relaxation, adapts the usage of machine learning models, and updates them to ensure compatibility with changes in input data, allowing for reduced measurement reporting while maintaining targeted accuracy.

Benefits of technology

This solution enables power savings for UE and reduces radio network resource consumption by optimizing measurement reporting, ensuring that the machine learning model's performance remains accurate and compatible with varying input conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure IB2024055298_05122024_PF_FP_ABST
    Figure IB2024055298_05122024_PF_FP_ABST
Patent Text Reader

Abstract

Disclosed is a method comprising: performing measurements and providing them as an input to a machine learning model, determining that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determining an adaptation for usage of the machine learning, determining if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapting the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and performing measurements and their reporting in accordance with the measurement relaxation.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] Machine Learning Aware Relaxation of Measurements Reporting

[0002] Field

[0003] The following example embodiments relate to wireless communication and determining how a user equipment is to perform and report measurements.

[0004] Background

[0005] In cellular communication, efficient usage of power is an important aspect for a user equipment (UE). One aspect that has potential for power saving in a UE is how often the UE is to perform and report measurements relating to the network, as well as how many measurements the UE is to perform. For example, the UE may report channel state information (CSI) according to a pre-determined framework. The time and frequency resources that may be used by the UE to report CSI may be controlled by an access node. CSI may comprise channel quality indicator (CQI), precoding matrix indicator (PMI), CSLRS resource indicator (CRI), SS / PBCH block resource indicator (SSBRI), layer indicator (LI), rank indicator (RI), layer 1 reference signal received power (Ll-RSRP) or layer 1 signal interference and noise ratio (Ll-SINR). Such a CSI measurement reporting may be performed according to reporting settings and resource settings. If the frequency of reporting, as well as the amount of features measured included in the reporting, are changed, that may impact the power consumption of the UE.

[0006] Brief Description

[0007] The scope of protection sought for various embodiments of the invention is set out by the independent claims. The exemplary 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 invention.

[0008] According to a first aspect there is provided an apparatus comprising means for performing: performing measurements and providing them as an input to a machine learning model, determining that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determining an adaptation for usage of the machine learning, determining if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapting the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and performing measurements and their reporting in accordance with the measurement relaxation.

[0009] In some example embodiments according to the first aspect, the means comprises at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, to cause the performance of the apparatus.

[0010] According to a second aspect there is provided an apparatus comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, are configured to cause the apparatus at least to: perform measurements and provide them as an input to a machine learning model, determine that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determine an adaptation for usage of the machine learning, determine if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapt the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and perform measurements and their reporting in accordance with the measurement relaxation.

[0011] According to a third aspect there is provided a method comprising: performing measurements and providing them as an input to a machine learning model, determining that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determining an adaptation for usage of the machine learning, determining if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapting the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and performing measurements and their reporting in accordance with the measurement relaxation.

[0012] In some example embodiment according to the third aspect the method is a computer implemented method. According to a fourth aspect there is provided a computer program comprising instructions which, when executed by an apparatus, cause the apparatus to perform at least the following: perform measurements and provide them as an input to a machine learning model, determine that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determine an adaptation for usage of the machine learning, determine if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapt the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and perform measurements and their reporting in accordance with the measurement relaxation.

[0013] According to a fifth aspect there is provided a computer program comprising instructions stored thereon for performing at least the following: performing measurements and providing them as an input to a machine learning model, determining that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determining an adaptation for usage of the machine learning, determining if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapting the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and performing measurements and their reporting in accordance with the measurement relaxation.

[0014] According to a sixth aspect there is provided a non-transitory computer readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to perform at least the following: perform measurements and provide them as an input to a machine learning model, determine that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determine an adaptation for usage of the machine learning, determine if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapt the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and perform measurements and their reporting in accordance with the measurement relaxation.

[0015] According to a seventh aspect there is provided a non-transitory computer readable medium comprising program instructions stored thereon for performing at least the following: performing measurements and providing them as an input to a machine learning model, determining that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determining an adaptation for usage of the machine learning, determining if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapting the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and performing measurements and their reporting in accordance with the measurement relaxation.

[0016] According to an eighth aspect there is provided a computer readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to perform at least the following: perform measurements and provide them as an input to a machine learning model, determine that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determine an adaptation for usage of the machine learning, determine if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapt the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and perform measurements and their reporting in accordance with the measurement relaxation.

[0017] According to a ninth aspect there is provided a computer readable medium comprising program instructions stored thereon for performing at least the following: performing measurements and providing them as an input to a machine learning model, determining that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model, determining an adaptation for usage of the machine learning, determining if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes, adapting the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and performing measurements and their reporting in accordance with the measurement relaxation.

[0018] According to a tenth aspect there is provided an apparatus comprising means for performing: determining that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receiving, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0019] In some example embodiments according to the tenth aspect, the means comprises at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, to cause the performance of the apparatus.

[0020] According to an eleventh aspect there is provided an apparatus comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, are configured to cause the apparatus at least to: determine that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receive, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0021] According to a twelfth aspect there is provided a method comprising: determining that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receiving, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0022] In some example embodiment according to the twelfth aspect the method is a computer implemented method.

[0023] According to a thirteenth aspect there is provided a computer program comprising instructions which, when executed by an apparatus, cause the apparatus to perform at least the following: determine that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receive, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0024] According to a fourteenth aspect there is provided a computer program comprising instructions stored thereon for performing at least the following: determining that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receiving, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0025] According to a fifteenth aspect there is provided a non-transitory computer readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to perform at least the following: determine that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receive, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0026] According to a sixteenth aspect there is provided a non-transitory computer readable medium comprising program instructions stored thereon for performing at least the following: determining that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receiving, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0027] According to a seventeenth aspect there is provided a computer readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to perform at least the following: determine that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receive, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

[0028] According to an eighteenth aspect there is provided a computer readable medium comprising program instructions stored thereon for performing at least the following: determining that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model, and receiving, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model. List of Drawings

[0029] In the following, the invention will be described in greater detail with reference to the embodiments and the accompanying drawings, in which

[0030] FIG. 1 illustrates an example embodiment of a radio access network.

[0031] FIG. 2 illustrates a signaling chart according to an example embodiment of L1 / L2 triggered mobility.

[0032] FIG. 3 illustrates an example embodiment of a machine learning model that is run in a network side.

[0033] FIG. 4 illustrates a flow chart according to an example embodiment of adapting usage of machine learning.

[0034] FIG. 5, FIG. 6A and FIG. 6B illustrate signalling charts according to example embodiments.

[0035] FIG. 7A and FIG. 7B illustrate example embodiment of adapting usage of machine learning.

[0036] FIG. 8 and FIG. 9 illustrate example embodiments of an apparatus.

[0037] Description of Embodiments

[0038] The following embodiments are exemplifying. Although the specification may refer to “an”, “one”, or “some” embodiment(s) in several locations of the text, this does not necessarily mean that each reference is made to the same embodiment(s), or that a particular feature only applies to a single embodiment. Single features of different embodiments may also be combined to provide other embodiments.

[0039] As used in this application, the term ‘circuitry’ refers to all of the following: (a) hardware-only circuit implementations, such as implementations in only analog and / or digital circuitry, and (b) combinations of circuits and software (and / or firmware), such as (as applicable): (i) a combination of processor(s) or (ii) portions of processor(s) / software including digital signal processor(s), software, and memory(ies) that work together to cause an apparatus to perform various functions, and (c) circuits, such as a microprocessor(s) or a portion of a microprocessor(s), that require software or firmware for operation, even if the software or firmware is not physically present. This definition of ‘circuitry’ applies to all uses of this term in this application. As a further example, as used in this application, the term ‘circuitry’ would also cover an implementation of merely a processor (or multiple processors) or a portion of a processor and its (or their) accompanying software and / or firmware. The term ‘circuitry’ would also cover, for example and if applicable to the particular element, a baseband integrated circuit or applications processor integrated circuit for a mobile phone or a similar integrated circuit in a server, a cellular network device, or another network device. The above-described embodiments of the circuitry may also be considered as embodiments that provide means for carrying out the embodiments of the methods or processes described in this document.

[0040] The techniques and methods described herein may be implemented by various means. For example, these techniques may be implemented in hardware (one or more devices), firmware (one or more devices), software (one or more modules), or combinations thereof. For a hardware implementation, the apparatus(es) of embodiments may be implemented within one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), graphics processing units (GPUs), processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described herein, or a combination thereof. For firmware or software, the implementation can be carried out through modules of at least one chipset (e.g. procedures, functions, and so on) that perform the functions described herein. The software codes may be stored in a memory unit and executed by processors. The memory unit may be implemented within the processor or externally to the processor. In the latter case, it can be communicatively coupled to the processor via any suitable means. Additionally, the components of the systems described herein may be rearranged and / or complemented by additional components in order to facilitate the achievements of the various aspects, etc., described with regard thereto, and they are not limited to the precise configurations set forth in the given figures, as will be appreciated by one skilled in the art.

[0041] Embodiments described herein may be implemented in a communication system, such as in at least one of the following: Long Term Evolution (LTE), LTE- Advanced, a fifth generation (5G) mobile or cellular communication system, 5G-Advanced and / or 6G. The embodiments are not, however, restricted to the systems given as an example but a person skilled in the art may apply the solution to other communication systems provided with necessary properties.

[0042] FIG. 1 depicts examples of simplified system architectures showing some elements and functional entities, all being logical units, whose implementation may differ from what is shown. The connections shown in FIG. 1 are logical connections; the actual physical connections may be different. It is apparent to a person skilled in the art that the system may comprise also other functions and structures than those shown in FIG. 1. The example of FIG. 1 shows a part of an example embodiment of a radio access network.

[0043] FIG. 1 shows terminal devices 100 and 102 configured to be in a wireless connection on one or more communication channels in a cell with an access node (such as (e / g)NodeB) 104 providing the cell. The terminal devices 100 and 102 may also be called as mobile device, or user equipment (UE), or user terminal, user device, etc. The access node 104 may also be referred to as a node, or a base station, or any other type of interfacing device including a relay station capable of operating in a wireless environment. The physical link from a terminal device to a (e / g)NodeB is called uplink (UL) or reverse link and the physical link from the (e / g)NodeB to the terminal device is called downlink (DL) or forward link. It should be appreciated that (e / g)NodeBs or their functionalities may be implemented by using any access node, host, server or access point etc. entity suitable for such a usage. It is to be noted that although one cell is discussed in this exemplary embodiment, for the sake of simplicity of explanation, multiple cells may be provided by one access node in some example embodiments.

[0044] A communication system may comprise more than one (e / g)NodeB in which case the (e / g)NodeBs may also be configured to communicate with one another over links, wired or wireless, designed for the purpose. These links may be used for signalling purposes. The (e / g)NodeB is a computing device configured to control the radio resources of communication system it is coupled to. The (e / g)NodeB includes or is coupled to transceivers. From the transceivers of the (e / g)NodeB, a connection is provided to an antenna unit that establishes bidirectional radio links to user devices. The antenna unit may comprise a plurality of antennas or antenna elements. The (e / g)NodeB is further connected to core network 110 (CN or next generation core NGC). Depending on the system, the counterpart on the CN side may be a serving gateway (S-GW, routing and forwarding user data packets), packet data network gateway (P-GW), for providing connectivity of terminal devices to external packet data networks, or mobile management entity (MME), etc.

[0045] The user equipment (UE) illustrates one type of an apparatus to which resources on the air interface are allocated and assigned, and thus any feature described herein with a UE may be implemented with a corresponding apparatus, such as a relay node. An example of such a relay node is a layer 3 relay (self-backhauling relay) towards the base station. Another example of such a relay node is a layer 2 relay. Such a relay node may comprise a UE part and a Distributed Unit (DU) part. A CU (centralized unit) may coordinate the DU operation via Fl AP -interface for example.

[0046] The UE may refer to a portable computing device that includes wireless mobile communication devices operating with or without a subscriber identification module (SIM), or an embedded SIM, eSIM. A UE may also be a device having capability to operate in Internet of Things (loT) network which is a scenario in which objects are provided with the ability to transfer data over a network without requiring human-to-human or human-to-computer interaction. The UE may also utilise cloud computing. The UE (or in some embodiments a layer 3 relay node) is configured to perform one or more of user equipment functionalities. It is to be noted that the UE may also be a vehicle or a household appliance capable of using cellular communication.

[0047] Additionally, although the apparatuses have been depicted as single entities, different units, processors and / or memory units (not all shown in FIG. 1) may be implemented.

[0048] 5G enables using multiple input - multiple output (MIMO) antennas, many more base stations or nodes than the LTE (a so-called small cell concept), including macro sites operating in cooperation with smaller stations and employing a variety of radio technologies depending on service needs, use cases and / or spectrum available. 5G mobile communications supports a wide range of use cases and related applications including video streaming, augmented reality, different ways of data sharing and various forms of machine type applications such as (massive) machine-type communications (mMTC), including vehicular safety, different sensors and real-time control. 5G is expected to have multiple radio interfaces, namely below 6GHz, cmWave and mmWave, and also being integratable with existing legacy radio access technologies, such as the LTE. Integration with the LTE may be implemented as a system, where macro coverage is provided by the LTE and 5G radio interface access comes from small cells by aggregation to the LTE. In other words, 5G is may support both inter-RAT operability (such as LTE-5G) and inter-RI operability (inter-radio interface operability, such as below 6GHz - cmWave, below 6GHz - cmWave - mmWave). One of the concepts considered to be used in 5G networks is network slicing in which multiple independent and dedicated virtual sub-networks (network instances) may be created within the same infrastructure to run services that have different requirements on latency, reliability, throughput and mobility.

[0049] The architecture in LTE networks is fully distributed in the radio and fully centralized in the core network. The low latency applications and services in 5G may require bringing the content close to the radio which may lead to local break out and multi-access edge computing (MEC). 5G enables analytics and knowledge generation to occur at the source of the data. MEC provides a distributed computing environment for application and service hosting. It also has the ability to store and process content in close proximity to cellular subscribers for faster response time. Edge computing covers a wide range of technologies such as wireless sensor networks, mobile data acquisition, mobile signature analysis, cooperative distributed peer-to- peer ad hoc networking and processing also classifiable as local cloud / fog computing and grid / mesh computing, dew computing, mobile edge computing, cloudlet, distributed data storage and retrieval, autonomic self-healing networks, remote cloud services, augmented and virtual reality, data caching, Internet of Things (massive connectivity and / or latency critical), critical communications (autonomous vehicles, traffic safety, real-time analytics, time-critical control, healthcare applications).

[0050] The communication system is also able to communicate with other networks, such as a public switched telephone network or the Internet 112, and / or utilise services provided by them. The communication network may also be able to support the usage of cloud services, for example at least part of core network operations may be carried out as a cloud service (this is depicted in FIG. 1 by “cloud” 114). The communication system may also comprise a central control entity, or a like, providing facilities for networks of different operators to cooperate for example in spectrum sharing. Edge cloud may be brought into radio access network (RAN) by utilizing network function virtualization (NFV) and software defined networking (SDN). Using edge cloud may mean access node operations to be carried out, at least partly, in a server, host or node operationally coupled to a remote radio head or base station comprising radio parts. It is also possible that node operations will be distributed among a plurality of servers, nodes or hosts. Application of cloudRAN architecture enables RAN real time functions being carried out at the RAN side (in a distributed unit, DU 104) and non-real time functions being carried out in a centralized manner (in a centralized unit, CU 108).

[0051] It should also be understood that the distribution of labour between core network operations and base station operations may differ from that of the LTE or even be non-existent. Some other technology that may be used includes for example Big Data and all-IP, which may change the way networks are being constructed and managed. 5G (or new radio, NR) networks are being designed to support multiple hierarchies, where MEC servers can be placed between the core and the base station or NodeB (gNB). It should be appreciated that MEC can be applied in 4G networks as well.

[0052] It is to be noted that the depicted system is an example of a part of a radio access system and the system may comprise a plurality of (e / g)NodeBs, UEs may have access to a plurality of radio cells and the system may comprise also other apparatuses, such as physical layer relay nodes or other network elements, etc. Additionally, in a geographical area of a radio communication system a plurality of different kinds of radio cells as well as a plurality of radio cells may be provided. Radio cells may be macro cells (or umbrella cells) which are large cells, usually having a diameter of up to tens of kilometers, or smaller cells such as micro-, femto- or picocells. The (e / g)NodeBs of FIG. 1 may provide any kind of these cells. A cellular radio system may be implemented as a multilayer network including several kinds of cells. In some exemplary embodiments, in multilayer networks, one access node provides one kind of a cell or cells, and thus a plurality of (e / g)NodeBs are required to provide such a network structure.

[0053] Regarding mobility of a UE connected to a cellular communication network, an example of beam-level mobility is a mechanism of layer 1 (Ll) / layer 2 (L2)-Triggered Mobility (LTM). LTM differs from layer 3 (L3) mobility procedures where the handover between two cells is decided by RRC layer, as with LTM the handover is performed by the MAC layer terminated in the Distributed Unit (DU). FIG. 2 illustrates a signaling chart of an example embodiment of the LTM. In this example embodiment, the signaling occurs between a UE 200, a first DU 202, a CU 204 and a second DU 206. This example embodiment illustrates the LTM from a serving cell in the first DU 202 to a target cell in the second DU 206, which corresponds to a inter-DU intra-CU scenario. The same diagram would apply as well in case of intra-DU intra-CU cell change in which the first DU 202 would be the same as the second DU 206. In this example embodiment, there is a preparation phase in which the LTM is performed between the signaling 210 and 245. First, the UE 200 transmits measurement report 210 to the first DU 202, which then transmits the measurement report 212 to the CU 204. The CU 204 then transmits a UE context setup request 214 to the first DU 202, which then responds by transmitting a UE context setup response 216. It is to be noted that the UE context setup response is transmitted from a DU to CU container.

[0054] Next, the CU 204 transmits a UE context setup request 220 to the second DU 206, which in response transmits a UR context setup response 225. After this, the CU 204 performs step 230 that generates an RRC reconfiguration that comprises determining a configuration for reporting measurements for LI cell change and determining configurations of a prepared cell. The CU 204 then transmits RRC reconfiguration signaling 240 to the UE 200, which then responds by transmitting RRC reconfiguration complete signaling 245. In other words, the network determines to configure potential target cells for LTM based on measurement report received from the UE 200.

[0055] The UE 200 then starts to report periodically the LI beam measurements of serving and candidate target cells as illustrated in signaling 250. Upon determining that there is a target candidate cell having a better radio link beam measurement than the serving cell, e.g., Ll- RSRP of target beam measurement > Ll-RSRP of serving beam measurement + Off for e.g., Time -to-Trigger (TTT) time, the serving cell, that is provided by the first DU 202, transmits a MAC Control Element (MAC CE) command 255, or a LI message, to trigger the cell change to the target candidate cell. The handover from serving cell to target cell, in other words the cell change 260, is performed by the UE. The LTM may have a benefit that, compared to a baseline handover and a conditional handover, the interruption during the handover execution may be reduced substantially as the UE does not need to perform higher layer (RRC, PDCP) reconfiguration and the UE may in some examples connect to the target cell without a random access procedure, in other words, the UE may perform RACHless to connect the target cell.

[0056] As mentioned previously, the CSI measurement reporting may be performed in accordance with reporting and resource settings. For example, each reporting setting CSI-ReportConfig may be associated with a single downlink bandwidth part (BWP), that may be indicated by higher layer parameter BWP-Id, given in the associated CSI-ResourceConfig for channel measurement and comprises parameter(s) for one CSI reporting band: codebook configuration comprising codebook subset restriction, time-domain behavior, frequency granularity for CQI and PMI, measurement restriction configurations, and the CSI-related quantities to be reported by the UE such as the layer indicator (LI), Ll-RSRP, Ll-SINR, CRI, and SSBRI (SSB Resource Indicator). The time domain behavior of the CSI-ReportConfig may be indicated by the higher layer parameter reportConfigType and may be set to 'aperiodic', 'semiPersistentOnPUCCH', 'semiPersistentOnPUSCH', or 'periodic'. For 'periodic' and 'semiPersistentOnPUCCH' / 'semiPersistentOnPUSCH' CSI reporting, the configured periodicity and slot offset applies in the numerology of the uplink (UL) BWP in which the CSI report is configured to be transmitted on. The higher layer parameter reportQuantity indicates the CSI-related, Ll-RSRP-related or Ll-SINR-related quantities to report.

[0057] As for resource settings, for example, each CSI resource setting CSI-ResourceConfig may comprise a configuration of a list of S>1 CSI Resource Sets that are given by higher layer parameter csi-RS-ResourceSetList, where the list is comprised of references to either or both of non-zero power (NZP) CSLRS resource set(s) and SS / PBCH block set(s) or the list is comprised of references to CSI-IM resource set(s). Each CSI Resource Setting may be located in the downlink (DL) BWP identified by the higher layer parameter BWP-id, and all CSI Resource Settings linked to a CSI Report Setting have the same DL BWP.

[0058] Measurements reporting reduction is of interest as it may allow achieving power saving for a UE as well as saving radio network resources for user data communication. For example, measurement reporting overhead consumes radio resources as well as induces UE energy consumption. Also, in L1 / L2 triggered mobility (LTM), periodic LI measurements are to be carried out thus increasing power consumption of the UE. For that, LI beam reporting increases the signaling overhead at the system as well as the UE power consumption. Reduction of measurement reporting can be understood as reducing the amount of measurements performed by the UE, in which case the UE can obtain power savings. Yet, reduction of measurement reporting may also be understood as reducing the reporting, although measurements as such are still being performed as before. Reducing the reporting as such may be beneficial, for example if a network is heavily loaded as fewer reporting may ease the loading of the network.

[0059] The measurements performed by the UE may be provided as an input to a machine learning (ML) model. This may be understood as usage of ML. Therefore, ML may be associated with a use case in which feature-based reduction of measurements is performed and this affects the measurements reported by the UE. Reduction of measurements, which may also be understood as relaxation of measurements, in other words, measurement relaxation, therefore affects the input provided to the ML model, which has an impact on the performance of the ML model. Therefore, it is desirable to have an approach for changing reported measurements when measurement reduction is performed sporadically e.g. based on correlation condition, considering that these received measurements are targeted to be used as an input to a predefined ML model. Also, it is desirable to have an approach for determining how the UE and network can coordinate this change to the input of the ML model input, as well as determining what the network can control, or limit, for the ML model hosted and run by the UE and ML-based functionality performed by the UE. Further, it is desirable to be able to determine how the reduced, or omitted, measurements are accounted for.

[0060] It is thus desirable to allow adaptation of the usage of the ML to account for varying measurement reporting. Such varying measurement reporting may be understood as relaxation of measurement reporting in terms of for example performing and reporting measurements, by a UE, less often, or not reporting for all cells, or a combination of both. Therefore, in case the ML model is executed at the network side, for example by an access node, or any other suitable network entity, the measurements performed by one or more UEs are received as inputs for the ML model. Thus, the ML model may be predefined and trained in advance to operate and expect a fixed set of input features corresponding to the input layer dimensions of the ML model. FIG. 3 illustrates an example embodiment of a ML model that is run in the network side. The ML model may be run by any suitable network entity, for example by an access node. In some other example embodiments, the ML model may also be run by a UE. In this example embodiment the ML model is trained to receive as input 300 three features that are measurements reported by a UE. Then, for each of the features, there are ten samples in this example embodiment, which means that in this example embodiment, the ML model 310 has an input layer 315 that has dimensions of 3x10. In case the set of these features is shortened or the number of samples per feature is reduced due to a measurement relaxation, which may be understood as relaxation of performing measurements by the UE, relaxation of measurement reporting as such, or a combination or both, then the ML model may not be run anymore. In case the inputs, that are no longer present, are replaced with dump values, then the ML model may be run, but such an approach may cause the prediction accuracy to become degraded and thus the performance does not achieve a target accuracy. In other words, the inputs received do not correspond to the dimensions of the input layer anymore, which affects the usage of ML as the ML model is not compatible with the changes caused to the input by the measurement relaxation.

[0061] In the example embodiment of FIG. 3 a UE transmits measurement reports regarding its serving cell and one or more neighbour cells such that there are for example 10 samples per cell. In other words, the temporal dimension is 10, and in this example embodiment there are 10 samples for each of the 3 cells for which the UE reports its measurements. Those samples may be used as the input 300 to the ML model 310. The input 300 may have different characteristics such as, type, sampling rate, dimensions and / or shape. In this example embodiment, the input 300 may be used for example if the ML model 310 is used for making a handover decision. Then, any measurement relaxations, such as ignoring some cells, or transmitting less samples per cell, or a combination of both, from the UE side to obtain power saving, and, additionally, or alternatively, air interface resources optimization, may result in a mismatch between the reported measurements and the expected input dimension for the ML model 310 at the network side. For that, a coordinated adaptation of the usage of the ML, which may comprise updating the ML model, between the network and the UE would be beneficial. FIG. 4 illustrates a flow chart according to an example embodiment regarding how to adapt usage of ML, by for example updating an ML model being used, so that it is compatible with measurement relaxations allowed for the measurements performed by the UE. In this example embodiment the adaptation may be performed before deployment of the ML model, in a so called a-priori mode, or after the deployment of the ML model by the network, in a so called a-posteriori mode, or by using a combination of both. It is to be noted that in the a-priori mode, the updating of the ML model may comprise replacing the ML model with another one. In the a-priori mode adapting the usage of ML may comprise having multiple ML models that can be used for updating the ML. This way there may be a dedicated ML model for different input dimensions and depending on the input dimensions associated with the measurement relaxation, model trained for those input dimensions may be used when adapting the usage of ML. The ML models may be identified by the identification (ID). In this example embodiment, the UE performs the measurement regarding one or more cells and reports them to the network as illustrated in step 400. It is then evaluated in step 405 if the received reported measurement reports comprise input intended for the ML model hosted by the network. If not, then in step 410 the network checks the measurement relaxation condition and allows the UE to run measurement relaxations. If yes, then in step 425 it is checked if the ML model is aware of measurement relaxation rules. If yes, the a-priori mode may be used and the network checks, in other words identifies, an ML model related measurement relaxations conditions, which is also then to be used to replace the current ML model being used, in step 430. Then, the identified ML model may be indicated to the UE and the UE may perform at least the ML inference associated with the identified ML model, and the measurement relaxation is allowed to be run by the UE as illustrated in step 420. Alternatively, if the a-priori mode may be used, it may be determined that the ML model that is currently being used already supports the measurement relaxation and the measurement relaxation is allowed to be run by the UE as illustrated in step 420. The current ML model may be able to support the measurement relaxation for example by adapting the input to be suitable for the ML model and thus the adaptation of the usage of ML comprises adapting the input of the ML model. In this a-priori mode, the identified ML model is trained before its deployment for real time inference. It is to be noted that there may be also other ML models trained such that one ML model is trained for each input that corresponds to one option of measurement relaxation. The measurement relaxation may be done differently and thus there may be different measurement relaxation options each causing the input to have different dimensions. The training may be achieved for example by using historical data that has been collected and then used to train the ML model hosted by an access node, such as a gNB, or any other suitable network entity. The historical data may thus be used to analyze the inputs, for example on sliding window basis, to identify possible schemes for relaxation of measurement reporting. Thereafter, the ML model training may be realized to account for different possible combinations of available measurements among the whole set of initial ones. In this a-priori mode, the network may then inform UE about the possible relaxation options supported by the trained and deployed ML model. This may work as a hard constraint on the UE which should limit its measurements relaxation to the options indicated by the network.

[0062] If on the other hand it is determined at step 425 that the ML model, that is currently being used, is not aware of rules for relaxing the measurement scheme, in other words is not compatible with relaxation of measurements, then the a-posteriori mode is used. In this mode, the network checks, in step 440, measurement relaxation condition and then indicates a request to adapt the usage of ML by updating the ML model in step 455. The updating may be achieved for example by re-training the ML model. Next, in step 455, it is determined if relaxation approval is obtained and if yes, then the measurement relaxation is allowed to be run by the UE as illustrated in step 420. If it is determined, in step 455, that the determined that relaxation approval is not obtained, then the UE discards the measurement relaxation, as illustrated in step 460, and does not perform the measurement relaxation.

[0063] In the a-posteriori mode, the ML model may be evolving based on the measurement relaxation request from UE. Re-training of the trained ML model deployed by the network may be triggered to account for measurement relaxation possibilities. Once the ML model is updated, then the measurement relaxation may be allowed at the UE side.

[0064] To enable adaptation of a usage of ML deployed by a network, in a-priori or in a-posteriori mode, there may be enhancements to signalling between a UE and the network. The signalling may be for example between the UE and an access node that provides the serving cell for the UE. FIG. 5 illustrates a signalling chart according to an example embodiment. In this example embodiment, the signalling is between a UE 500 and an access node providing a serving cell 505. In this example embodiment, a ML model is used, and it is trained and prepared by the network, using any suitable one or more network entities. In this example embodiment, first, there is signalling 510 between the UE 500 and the access node 505. The signalling 510 is regarding capabilities of the UE 500. This signalling indicates to the access node 505 the capabilities that the UE has in terms of performing computing. Thus, the capabilities may comprise indication of hardware capabilities, of software capabilities or of a combination of both. This allows the access node 505 to be able to determine the type of computing the UE 500 is capable of performing. Then, there is signalling 515 regarding identifying a configuration for the ML model that is used such that it is compatible with the capabilities of the UE 500. The configuration may comprise for example indicating number of layers and nodes used if the ML model is for example a neural network. The signalling 515 is transmitted to the UE 500 from the access node 505. The UE 500 then responds by transmitting signalling 520 acknowledging the identifying of the ML model and the configuration for the ML model.

[0065] Then, the UE 500 may transmit signalling 525 regarding UE ML relaxation capabilities to the access node 505. It is to be noted that the UE 500 can trigger the signaling 525 at any time during an RRC connection. The access node 505 then analyses possibilities for the measurement relaxation in terms of target accuracy of the ML model as illustrated in step 530. The possibilities may comprise for example analysing what measurements can be relaxed and how. After the analyses, the access node 505 transmits the signalling 535 indicating the measurement relaxations options and corresponding adaptation to the usage of ML by indicating update to the ML model to the UE 500. The update may comprise identifying one or more ML models and their configurations. Next, the UE 500 performs checking of measurement relaxations as illustrated in block 540. The UE 500 may decide which measurement relaxation to perform based on its capabilities and conditions and may use the tailored ML model for the selected option. As in this example embodiment, the ML inference is performed by the UE 500, in step 550 the UE runs the ML and then provides the signalling 555 regarding output of the ML to the access node 505 and informs the access node 505 about the used measurement relaxation option. In addition, or as an alternative, to the output the UE 500 may provide to the access node 505 also key performance indicators (KPIs). FIG. 6A illustrates another signalling chart according to an example embodiment. In this example embodiment a-posteriori mode is utilized. The signalling is between a UE 600 and an access node 605. In this example embodiment, an ML model is used, and it is trained and prepared by the network, using any suitable one or more network entities. In this example embodiment, first, there is signalling 610 between the UE 600 and the access node 605. The signalling 610 is regarding capabilities of the UE 600. Then, there is signalling 615 regarding identifying a configuration for the ML model. The signalling 615 is transmitted to the UE 600 from the access node 605. The UE 600 then responds by transmitting signalling 620 acknowledging the identifying of the configuration for the ML model. The signalling 610, 615 and 620 corresponds to the signalling 510, 515 and 520 described in the previous example embodiment.

[0066] In this example embodiment, the network shares with the UE 600 the trained ML model that takes as an input a complete set of input data. In this example embodiment, the UE 600 performs ML inference, as illustrated in step 630 and then provides the output of the ML model to the access node 605 using signalling 635. Additionally, or alternatively, KPIs may be provided. Then in step 640 the UE 600 checks for possibilities to relax measurements performed by the UE 600. This may be because during the inference, the UE 600 may observe a possibility to reduce, in other words, relax, measurements performed regarding the network. For example, the UE 600 may wish to perform less neighbour cell measurements and achieve power saving. Thus, the UE 600 may transmit signalling 645 to the access node 605 and request the measurement relaxation. Thereafter the access node 605 may check if the proposed measurement relaxation can be accepted and prepare the most suitable ML model for it as illustrated in step 650. Thus, if the relaxation is acceptable, the access node, or any suitable network entity, may adapt the ML model used in accordance with the measurement relaxation. Using signalling 655, the access node 605 then transmits to the UE 600 and indicates measurement relaxation approval or denial.

[0067] FIG. 6B illustrates a signaling chart according to another example embodiment, which is a variation of the example embodiment illustrated in FIG. 6A. Like in the example embodiment of FIG. 6A, there is signaling between the UE 600 and an access node 605. The signaling 610, 615, 620, 635 and 645 as well as steps 630 and 640 correspond to those of the example embodiment of FIG. 6A. However, in this example embodiment, the selection, training and inference operations of the ML model are performed by the UE 600. When the UE then considers realizing measurement relaxation possibilities, and transmits the request using signaling 645, the network is to check the proposed relaxation as illustrated in step 660. The checking may be performed by the access node 605, or by any other one or more network entities or by a combination of both. The checking and determining if the measurement relaxation can be performed by the UE may be based on for example the experiences, or decisions, or a combination of both, of other neighbouring cells.

[0068] The access node 605 then transmits either approval or denial of the measurement relaxation using signaling 665. If acceptance is transmitted, then in step 670 the ML model is adapted, by the UE 600, to correspond to the determined measurement relaxation scheme. Then, the UE 600 transmits to the access node 605 parameters of the adapted ML model properties. This is performed using signaling 675. Adaptation of the ML model may be performed to account for different ML model inputs options such as relaxed and not relaxed.

[0069] Fig. 7A and FIG. 7B illustrate example embodiments regarding adapting usage of ML by updating a ML model such that it is compatible with a relaxation of a measurement scheme. For a ML model, the inputs may be denoted as {X}WxM. For example, in the example embodiment of FIG. 3, the ML inputs refer to measurements performed by a UE regarding serving cell and one or more neighbour cells. In that example embodiment there are 10 samples per each cell and thus the dimensions are N = 3 and M = 10. The output of the ML model is handover decision denoted as Y, which may refer to for example the ID of either a target cell, or a beam. If the input measurements are relaxed, then just a subset of N X M are available to ML model inference. This may correspond to there being less cells, or beams, e.g. 2 instead of 3, or less samples e.g. 7 instead of 10.

[0070] FIG. 7A illustrates an example embodiment to adapting the usage of ML by updating the ML model and the adapting may be for example such that it would be compatible with the scenario described above. In this example embodiment, different measurement sets 702, 704 and 706 of that are defined as inputs 700. Then a dedicated ML model may be trained per one input measurement set. For example, ML model 712 may be dedicated to the measurement set 702, ML model 714 may be dedicated to the measurement set 704, and ML model 716 may be dedicated to the measurement set 706. Then the identification (ID) of the ML model 710 may be determined based on the measurement set that is going to be used and the ML model ID is then provided as the output 720. Thus, the different ML models may have different architectures and / or parameters, yet, their output is the same.

[0071] FIG. 7B illustrates another example embodiment of adapting the usage of ML by updating a ML model to correspond to relaxation of measurements performed and reported by a UE. In this example embodiment the size of an input vector 760 is kept constant. Thus, when an input 750 is received, that has a size different than expected, then an adapter function 755 may be applied to the input vector 750. This adapter function 755 replaces missing measurement values with dump values and can be considered as an update to the ML model used. There can be for example appropriate padding values that are selected by the adapter function 755. Such padding values may be determined based on for example mean / median value, K nearest neighbors (KNN), or even a separate ML model to predict the data for the padding values. However, for the sake of argument and without loss of generality, it may also be considered that the ML model inputs comprise RSRP values from cell 1 to cell N. As measurement relaxation is applied, then RSRP from cell N is no longer reported. The value for the cell N may then be replaced, for ML inference, by a dump value such as -150 dbm, which is not conventionally reported. However, the ML model should be trained for this replacement of data samples with dump values to let the model learn also implicitly the cases where the expected measurements are not reported. Thus, having an adapter function 755 allows to provide input 760 to the ML model 770 such that the X K2available features are accommodated to IV X M features (with K1< N and K2< M ) and then the output 755 can be obtained.

[0072] To describe a further example of measurement relaxation aware ML model adaptation, a use case of handover optimization may be considered. In this use case the target is to predict the next beam, or cell id, towards which a UE should be attached during its mobility. The ML model may then receive as its input consecutive RSRP measurements from serving and neighboring cells during a predefined period. As the output, the ML model may then provide target cell, or beam, ids. The ML model may be any suitable model, for example, long short- term memory (LSTM) model. In this example embodiment, the expected ML input data is a matrix of size 3 x 10 3 x 10 corresponding to RSRP measurements from serving and 2 neighboring cells during the last 10 consecutive time iterations. Then, the LSTM model is first selected and trained with a labelled data set including the original input data size 3 x 10. This model may be referred to as model id1. It is to be noted that if measurement relaxation is occurring and the UE is measuring 2 cells instead of the intended 3, the input data for ML model id1is now 2 x 10 , and in this case, the inference is not possible and the expected prediction of the target cell id cannot be obtained. Thus, another model id2may defined and it may also be LSTM but trained with labelled data set of input data size of 2 x 10. These two models ic^and id2can have the same architecture in terms of layers and number of hidden layers, however, their weights will be different.

[0073] It is to be noted that the example embodiments are applicable also when the measurements performed by a UE are other measurements than those related to the network, but could also be other measurements performed by the UE. Additionally, the example embodiments described above are applicable to other use cases in which the input that is provided to an ML model, is relaxed.

[0074] FIG. 8 illustrates an apparatus 800, which may be an apparatus such as, or comprised in, a user equipment, according to an example embodiment. The apparatus 800 comprises a processor 810. The processor 810 interprets computer program instructions and processes data. The processor 810 may comprise one or more programmable processors. The processor 810 may comprise programmable hardware with embedded firmware and may, alternatively or additionally, comprise one or more application specific integrated circuits, ASICs.

[0075] The processor 810 is coupled to a memory 820. The processor is configured to read and write data to and from the memory 820. The memory 820 may comprise one or more memory units. The memory units may be volatile or non-volatile. It is to be noted that in some example embodiments there may be one or more units of non-volatile memory and one or more units of volatile memory or, alternatively, one or more units of non-volatile memory, or, alternatively, one or more units of volatile memory. Volatile memory may be for example RAM, DRAM or SDRAM. Non-volatile memory may be for example ROM, PROM, EEPROM, flash memory, optical storage or magnetic storage. In general, memories may be referred to as non-transitory computer readable media. The memory 820 stores computer readable instructions that are execute by the processor 810. For example, non-volatile memory stores the computer readable instructions and the processor 810 executes the instructions using volatile memory for temporary storage of data and / or instructions.

[0076] The computer readable instructions may have been pre-stored to the memory 820 or, alternatively or additionally, they may be received, by the apparatus, via electromagnetic carrier signal and / or may be copied from a physical entity such as computer program product. Execution of the computer readable instructions causes the apparatus 800 to perform functionality described above.

[0077] In the context of this document, a “memory” or “computer-readable media” may be any non- transitory media or means that can contain, store, communicate, propagate or transport the instructions for use by or in connection with an instruction execution system, apparatus, or device, such as a computer.

[0078] The apparatus 800 further comprises, or is connected to, an input unit 830. The input unit 830 comprises one or more interfaces for receiving a user input. The one or more interfaces may comprise for example one or more motion and / or orientation sensors, one or more cameras, one or more accelerometers, one or more microphones, one or more buttons and one or more touch detection units. Further, the input unit 830 may comprise an interface to which external devices may connect to.

[0079] The apparatus 800 also comprises an output unit 840. The output unit comprises or is connected to one or more displays capable of rendering visual content such as a light emitting diode, LED, display, a liquid crystal display, LCD and a liquid crystal on silicon, LCoS, display. The output unit 840 further comprises one or more audio outputs. The one or more audio outputs may be for example loudspeakers or a set of headphones.

[0080] The apparatus 800 may further comprise a connectivity unit 850. The connectivity unit 850 enables wired and / or wireless connectivity to external networks. The connectivity unit 850 may comprise one or more antennas and one or more receivers that may be integrated to the apparatus 800 or the apparatus 800 may be connected to. The connectivity unit 850 may comprise an integrated circuit or a set of integrated circuits that provide the wireless communication capability for the apparatus 800. Alternatively, the wireless connectivity may be a hardwired application specific integrated circuit, ASIC.

[0081] It is to be noted that the apparatus 800 may further comprise various component not illustrated in the FIG. 8. The various components may be hardware component and / or software components.

[0082] FIG. 9 illustrates an example embodiment of an apparatus that may be an access node or be comprised in an access node such as a gNB. The apparatus may be, for example, a circuitry or a chipset applicable to an access node to realize the described embodiments. The apparatus 900 may be an electronic device comprising one or more electronic circuitries. The apparatus 900 may comprise a communication control circuitry 910 such as at least one processor, and at least one memory 920 including a computer program code (software) 922 wherein the at least one memory and the computer program code (software) 922 are configured, with the at least one processor, to cause the apparatus 900 to carry out any one of the example embodiments of the access node described above.

[0083] The memory 920 may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, flash memory, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. The memory may comprise a configuration database for storing configuration data. For example, the configuration database may store current neighbour cell list, and, in some example embodiments, structures of the frames used in the detected neighbour cells.

[0084] The apparatus 900 may further comprise a communication interface 930 comprising hardware and / or software for realizing communication connectivity according to one or more communication protocols. The communication interface 930 may provide the apparatus with radio communication capabilities to communicate in the cellular communication system. The communication interface may, for example, provide a radio interface to terminal devices. The apparatus 900 may further comprise another interface towards a core network such as the network coordinator apparatus and / or to the access nodes of the cellular communication system. The apparatus 900 may further comprise a scheduler 940 that is configured to allocate resources.

[0085] Even though the invention has been described above with reference to examples according to the accompanying drawings, it is clear that the invention is not restricted thereto but can be modified in several ways within the scope of the appended claims. Therefore, all words and expressions should be interpreted broadly and they are intended to illustrate, not to restrict, the embodiment. It will be obvious to a person skilled in the art that, as technology advances, the inventive concept can be implemented in various ways. Further, it is clear to a person skilled in the art that the described embodiments may, but are not required to, be combined with other embodiments in various ways.

Claims

Claims1. An apparatus comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, are configured to cause the apparatus at least to: perform measurements and provide them as an input to a machine learning model; determine that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model; determine an adaptation for usage of the machine learning; determine if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes; adapt the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation; and perform measurements and their reporting in accordance with the measurement relaxation.

2. An apparatus according to claim 1, wherein if the adaptation for the usage of the machine learning causes its performance not to be in accordance with the targeted accuracy, the apparatus is further caused to discard the measurement relaxation.

3. An apparatus according to claim 1 or 2, wherein updating the machine learning model comprises re-training the machine learning model, or replacing the machine learning model with another machine learning model, wherein the other machine learning model is selected from a group of one or more other machine learning models.

4. An apparatus according to claim 3, wherein the group comprises a plurality of machine learning models such that there is a dedicated machine learning model per each input characteristics associated with its respective option of the measurement relaxation.

5. An apparatus according to claim 4, wherein input characteristics comprises one or more of the following: input type, input sampling rate, input dimension, or input shape.

6. An apparatus according to any of claims 3 to 5, wherein the apparatus is further caused to receive the adaptation configurations for updating the machine learning model from the access node; or the apparatus is caused to determine the adaptation configuration for updating the machine learning model.

7. An apparatus according to any previous claim, wherein the apparatus is further caused to: transmit, to the access node, a request for using the measurement relaxation; and receive, from the access node, an approval or denial to use the measurement relaxation, wherein the measurement relaxation is in accordance with the targeted usage of machine learning.

8. An apparatus according to claim 7, wherein apparatus is further caused to perform one of the following: upon receiving the approval from the access node, perform the update and transmit a configuration of the other machine learning model to the access node; or upon receiving the approval from the access node, receiving also the other machine learning model from the access node.

9. An apparatus according to any previous claim, wherein adapting the machine learning comprises applying an adapter function to the input of the machine learning model.

10. An apparatus comprising at least one processor, and at least one memory storing instructions that, when executed by the at least one processor, are configured to cause the apparatus at least to: determine that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model; andreceive, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

11. An apparatus according to claim 10, wherein the apparatus is further caused to: receive information regarding the measurement relaxation from the user equipment; analyze measurement relaxation possibilities for a targeted accuracy of the usage of the machine learning; and indicate to the user equipment options regarding the measurement relaxation and one or more machine learning models that are compatible with their respective measurement relaxation options.

12. An apparatus according to claim 11, wherein receiving the information regarding the measurement relaxation comprise receiving, from the user equipment, a request for the measurement relaxation.

13. An apparatus according to claim 11 or 12, wherein the apparatus is further caused to evaluate the request and in response to the evaluation, transmit to the user equipment an approval or a denial regarding the measurement relaxation.

14. An apparatus according to any of claims 10 to 13, wherein the apparatus is further caused to determine that the updating of the machine learning model requires re-training the machine learning model, or replacing the machine learning model with another machine learning model .

15. An apparatus according to claim 14, wherein the apparatus is further caused to transmit adaptation configurations for updating to the machine learning model to the user equipment.

16. An apparatus according to any of claims 10 to 15, wherein the apparatus is further caused to determine that adapting the machine learning requires applying an adapter function to the input of the machine learning model.

17. A method comprising: performing measurements and providing them as an input to a machine learning model; determining that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model; determining an adaptation for usage of the machine learning; determining if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes; adapting the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation; and performing measurements and their reporting in accordance with the measurement relaxation.

18. A method comprising: determining that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model; and receiving, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

19. A computer program comprising instructions which, when executed by an apparatus, cause the apparatus to perform at least the following: perform measurements and provide them as an input to a machine learning model; determine that a measurement relaxation is applicable, and that the measurement relaxation causes changes to the input of the machine learning model; determine an adaptation for usage of the machine learning; determine if the adaptation for the usage of the machine learning causes performance to be in accordance with a targeted accuracy, and if yes; adapt the usage of the machine learning by updating the machine learning model, wherein the updating is compatible with the changes caused to the input of the machine learning model by the measurement relaxation; and perform measurements and their reporting in accordance with the measurement relaxation.

20. A computer program comprising instructions which, when executed by an apparatus, cause the apparatus to perform at least the following: determine that measurement relaxation is applicable for measurements performed by a user equipment, wherein the user equipment uses machine learning, and the usage of the machine learning comprises providing the measurements as an input for a machine learning model, and wherein the measurement relaxation causes changes to the input of the machine learning model; and receive, from the user equipment, measurement reporting in accordance with adapted usage of the machine learning, wherein the adapted usage of the machine learning comprises updating the machine learning model used by the user equipment to be compatible with the changes caused to the input of the machine learning model by the measurement relaxation, and wherein the received measurement reporting corresponds to the updated machine learning model.

Citation Information

Patent Citations

  • Information processing method and device, communication equipment and storage medium

    CN114467324A

  • Machine learning for handover

    US20220322195A1

  • Measurement relaxation

    WO2023066967A2

  • Wireless terminal, radio access network node, and methods therefor

    WO2023079946A1

Cited By

  • Method, apparatus, and system for beam management

    WO2026165756A1

  • Low-power operation method and apparatus for reducing power consumption of next-generation energy-efficient mobile communication system by using artificial intelligence

    WO2026177481A1