Devices and methods of communication
Terminal devices manage AI/ML model inferences by switching schemes, adjusting configurations, and providing availability information to handle undetectable or unreliable inputs, improving model robustness and reducing overhead in new radio systems.
Patent Information
- Application Number
- PCT/CN2024/085576
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-04-02
- Publication Date
- 2025-10-09
AI Technical Summary
AI/ML model inputs for terminal devices in new radio systems may be undetectable or unreliable, leading to measurement overhead and degradation of model inference performance.
Terminal devices implement model management strategies such as switching to non-AI schemes, adjusting model configurations, skipping unreliable measurements, and providing availability information to network devices to enhance robustness and reliability of model inference.
Enhances the robustness and reliability of AI/ML model inferences by detecting and handling issues with undetectable or unreliable inputs, facilitating efficient model management and reducing measurement overhead.
Smart Images

Figure CN2024085576_09102025_PF_FP_ABST
Abstract
Description
DEVICES AND METHODS OF COMMUNICATIONTECHNICAL FIELD
[0001] Embodiments of the present disclosure generally relate to the field of telecommunication, and in particular, to devices and methods of communication for management of an artificial intelligence (AI) / machine learning (ML) model.BACKGROUND
[0002] AI / ML technology will be used as a proactive approach for improving handover performance in a new radio (NR) system. For example, for a use case of a spatial domain prediction, a terminal device may only perform measurements on a set A and use measurement results of the set A to predict measurement results of a set B. Thus, measurement overhead may be reduced. However, the measurement results of the set A may be not always available or reliable.SUMMARY
[0003] In general, embodiments of the present disclosure provide methods, devices and computer storage media of communication for model management.
[0004] In a first aspect, there is provided a terminal device. The terminal device comprises a processor. The processor is configured to cause the terminal device to: determine that a problem occurs in a model inference of a first model based on at least one of the following: a model input for the first model is undetectable or is undetectable for a first time duration or for a first number of model inference cycles, the model input for the first model is unreliable or is unreliable for a second time duration or for a second number of model inference cycles, or a requirement on a processing time of the model inference is unfulfilled; and perform a model management comprising at least one of the following: switching to a non-AI scheme, switching from a first configuration of the first model to a second configuration of the first model, switching from the first model to a second model, or skipping a set of undetectable or unreliable measurement results in the model input or skipping the set of undetectable or unreliable measurement results for a third time duration.
[0005] In a second aspect, there is provided a terminal device. The terminal device comprises a processor. The processor is configured to cause the terminal device to: receive, from a network device, a request for availability information of a model input for a model; and transmit, to the network device, the availability information of the model input comprising at least one of the following: an indication of whether measurement results in the model input are available, identity information associated with a set of undetectable or unreliable measurement results in the model input, or the measurement results in the model input.
[0006] In a third aspect, there is provided a terminal device. The terminal device comprises a processor. The processor is configured to cause the terminal device to: receive, from a network device, a list of model configurations for a model, a model configuration in the list of model configurations comprising information of a model input and information of a model output for the model and an applicable condition associated with the model input and the model output; and in accordance with a determination that the applicable condition is fulfilled, perform a model inference based on the model input and the model output.
[0007] In a fourth aspect, there is provided a method of communication. The method comprises: determining, at a terminal device, that a problem occurs in a model inference of a first model based on at least one of the following: a model input for the first model is undetectable or is undetectable for a first time duration or for a first number of model inference cycles, the model input for the first model is unreliable or is unreliable for a second time duration or for a second number of model inference cycles, or a requirement on a processing time of the model inference is unfulfilled; and performing a model management comprising at least one of the following: switching to a non-AI scheme, switching from a first configuration of the first model to a second configuration of the first model, switching from the first model to a second model, or skipping a set of undetectable or unreliable measurement results in the model input or skipping the set of undetectable or unreliable measurement results for a third time duration.
[0008] In a fifth aspect, there is provided a method of communication. The method comprises: receiving, at a terminal device and from a network device, a request for availability information of a model input for a model; and transmitting, to the network device, the availability information of the model input comprising at least one of the following: an indication of whether measurement results in the model input are available, identity information associated with a set of undetectable or unreliable measurement results in the model input, or the measurement results in the model input.
[0009] In a sixth aspect, there is provided a method of communication. The method comprises: receiving, at a terminal device and from a network device, a list of model configurations for a model, a model configuration in the list of model configurations comprising information of a model input and information of a model output for the model and an applicable condition associated with the model input and the model output; and in accordance with a determination that the applicable condition is fulfilled, performing a model inference based on the model input and the model output.
[0010] In a seventh aspect, there is provided a computer readable medium having instructions stored thereon. The instructions, when executed on at least one processor, cause the at least one processor to perform the method according to any of the fourth to sixth aspects of the present disclosure.
[0011] Other features of the present disclosure will become easily comprehensible through the following description.BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Through the more detailed description of some embodiments of the present disclosure in the accompanying drawings, the above and other objects, features and advantages of the present disclosure will become more apparent, wherein:
[0013] FIG. 1 illustrates an example communication network in which some embodiments of the present disclosure can be implemented;
[0014] FIG. 2 illustrates a schematic diagram of an example spatial domain prediction in which some embodiments of the present disclosure can be implemented;
[0015] FIG. 3 illustrates a signaling chart illustrating an example process of communication in accordance with some embodiments of the present disclosure;
[0016] FIG. 4 illustrates a signaling chart illustrating another example process of communication in accordance with some embodiments of the present disclosure;
[0017] FIG. 5 illustrates a signaling chart illustrating still another example process of communication in accordance with some embodiments of the present disclosure;
[0018] FIG. 6 illustrates a flowchart of an example method of communication implemented at a terminal device in accordance with some embodiments of the present disclosure;
[0019] FIG. 7 illustrates a flowchart of another example method of communication implemented at a terminal device in accordance with some embodiments of the present disclosure;
[0020] FIG. 8 illustrates a flowchart of still another example method of communication implemented at a terminal device in accordance with some embodiments of the present disclosure; and
[0021] FIG. 9 is a simplified block diagram of a device that is suitable for implementing embodiments of the present disclosure.
[0022] Throughout the drawings, the same or similar reference numerals represent the same or similar element.DETAILED DESCRIPTION
[0023] Principle of the present disclosure will now be described with reference to some embodiments. It is to be understood that these embodiments are described only for the purpose of illustration and help those skilled in the art to understand and implement the present disclosure, without suggesting any limitations as to the scope of the disclosure. The disclosure described herein can be implemented in various manners other than the ones described below.
[0024] In the following description and claims, unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skills in the art to which this disclosure belongs.
[0025] As used herein, the term ‘terminal device’ refers to any device having wireless or wired communication capabilities. Examples of the terminal device include, but not limited to, user equipment (UE) , personal computers, desktops, mobile phones, cellular phones, smart phones, personal digital assistants (PDAs) , portable computers, tablets, wearable devices, Internet of things (IoT) devices, ultra-reliable and low latency communications (URLLC) devices, Internet of everything (IoE) devices, machine type communication (MTC) devices, device on vehicle for V2X communication where X means pedestrian, vehicle, or infrastructure / network, devices for integrated access and backhaul (IAB) , small data transmission (SDT) , mobility, multicast and broadcast services (MBS) , positioning, dynamic / flexible duplex in commercial networks, reduced capability (RedCap) , Space borne vehicles or air borne vehicles in non-terrestrial networks (NTN) including Satellites and high altitude platforms (HAPs) encompassing unmanned aircraft systems (UAS) , extended reality (XR) devices including different types of realities such as augmented reality (AR) , mixed reality (MR) and virtual reality (VR) , the unmanned aerial vehicle (UAV) commonly known as a drone which is an aircraft without any human pilot, devices on high speed train (HST) , or image capture devices such as digital cameras, sensors, gaming devices, music storage and playback appliances, or Internet appliances enabling wireless or wired Internet access and browsing and the like. The ‘terminal device’ can further has ‘multicast / broadcast’ feature, to support public safety and mission critical, V2X applications, transparent IPv4 / IPv6 multicast delivery, IPTV, smart TV, radio services, software delivery over wireless, group communications and IoT applications. It may also incorporate one or multiple subscriber identity module (SIM) as known as multi-SIM. The term ‘terminal device’ can be used interchangeably with a UE, a mobile station, a subscriber station, a mobile terminal, a user terminal or a wireless device.
[0026] The term ‘network device’ refers to a device which is capable of providing or hosting a cell or coverage where terminal devices can communicate. Examples of a network device include, but not limited to, a Node B (NodeB or NB) , an evolved NodeB (eNodeB or eNB) , a next generation NodeB (gNB) , a transmission reception point (TRP) , a remote radio unit (RRU) , a radio head (RH) , a remote radio head (RRH) , an IAB node, a low power node such as a femto node, a pico node, a reconfigurable intelligent surface (RIS) , Network-controlled Repeaters, and the like.
[0027] The terminal device or the network device may have artificial intelligence (AI) or machine learning capability. It generally includes a model which has been trained from numerous collected data for a specific function, and can be used to predict some information.
[0028] The terminal or the network device may work on several frequency ranges, e.g. FR1 (410 MHz to 7125 MHz) , FR2 (24.25GHz to 71GHz) , frequency band larger than 100GHz as well as Tera Hertz (THz) . It can further work on licensed / unlicensed / shared spectrum. The terminal device may have more than one connections with the network devices under MR-DC application scenario. The terminal device or the network device can work on full duplex, flexible duplex and cross division duplex modes.
[0029] The network device may have the function of network energy saving, self-organizing networks (SON) / minimization of drive tests (MDT) . The terminal may have the function of power saving.
[0030] The embodiments of the present disclosure may be performed in test equipment, e.g. signal generator, signal analyzer, spectrum analyzer, network analyzer, test terminal device, test network device, channel emulator.
[0031] In one embodiment, the terminal device may be connected with a first network device and a second network device. One of the first network device and the second network device may be a master node and the other one may be a secondary node. The first network device and the second network device may use different radio access technologies (RATs) . In one embodiment, the first network device may be a first RAT device and the second network device may be a second RAT device. In one embodiment, the first RAT device is eNB and the second RAT device is gNB. Information related with different RATs may be transmitted to the terminal device from at least one of the first network device or the second network device. In one embodiment, first information may be transmitted to the terminal device from the first network device and second information may be transmitted to the terminal device from the second network device directly or via the first network device. In one embodiment, information related with configuration for the terminal device configured by the second network device may be transmitted from the second network device via the first network device. Information related with reconfiguration for the terminal device configured by the second network device may be transmitted to the terminal device from the second network device directly or via the first network device.
[0032] As used herein, the singular forms ‘a’ , ‘an’ and ‘the’ are intended to include the plural forms as well, unless the context clearly indicates otherwise. The term ‘includes’ and its variants are to be read as open terms that mean ‘includes, but is not limited to. ’ The term ‘based on’ is to be read as ‘at least in part based on. ’ The term ‘one embodiment’ and ‘an embodiment’ are to be read as ‘at least one embodiment. ’ The term ‘another embodiment’ is to be read as ‘at least one other embodiment. ’ The terms ‘first, ’ ‘second, ’ and the like may refer to different or same objects. Other definitions, explicit and implicit, may be included below.
[0033] In some examples, values, procedures, or apparatus are referred to as ‘best, ’ ‘lowest, ’ ‘highest, ’ ‘minimum, ’ ‘maximum, ’ or the like. It will be appreciated that such descriptions are intended to indicate that a selection among many used functional alternatives can be made, and such selections need not be better, smaller, higher, or otherwise preferable to other selections.
[0034] In the context of the present disclosure, the term ‘AI / ML model’ herein refers to a model used for mobility management. The term ‘AI / ML model’ may be interchangeably used with ‘ML model’ or ‘AI model’ or ‘model’ .
[0035] In the context of the present disclosure, the term ‘spatial domain prediction’ herein may be interchangeably used with ‘intra-frequency prediction’ , ‘inter-frequency prediction’ , ‘inter-cell prediction’ or ‘inter-beam prediction’ . The term ‘measurement results’ herein may refer to any suitable measurement metrics, such as reference signal received power (RSRP) , reference signal received quality (RSRQ) , or signal-to-interference-plus-noise ratio (SINR) . The term ‘signal strength or quality’ herein may refer to any suitable measured values, such as RSRP, RSRQ, or SINR. The term ‘measurement results’ herein may refer to measurement results of cell (serving cell or neighbor cell) or beam or frequency measurements. The term ‘model input’ or ‘model output’ may indicate a measurement result associated with at least one of a cell identity (ID) , a beam ID, a layer 1 (L1) or layer 2 (L2) triggered mobility (LTM) candidate ID, a LTM candidate physical cell identity (PCI) , a LTM synchronization signal block (SSB) configuration, or a SSB index.
[0036] The term ‘similarity’ or ‘dissimilarity’ or ‘diversity’ herein may be a value that is used to represent / reflect a difference between a data distribution and another data distribution. The value may comprise at least one of correlation coefficient, cosine similarity, Kullback-Leibler divergence, Euclidean distance, Manhattan distance, Jaccard similarity coefficient, etc.
[0037] In the context of the present disclosure, the term ‘undetectable’ herein may be interchangeably used with ‘unavailable’ or ‘unidentified’ , and the term ‘unreliable’ herein may be interchangeably used with ‘problematic’ or “with low quality” .
[0038] As mentioned above, the measurement results of the set A as a model input may be not always available or reliable, which may result in degradation of a model inference.
[0039] Embodiments of the present disclosure provide solutions of model management so as to overcome the above and other potential issues. In one aspect, a terminal device may determine whether a problem occurs in a model inference of a first model based on at least one of the following: a model input for the first model is undetectable or is undetectable for a first time duration or for a first number of model inference cycles, the model input for the first model is unreliable or is unreliable for a second time duration or for a second number of model inference cycles, or a requirement on a processing time of the model inference is unfulfilled. Upon determination of the problem occurs in the model inference, the terminal device may perform a model management. In this way, a problem in a model inference may be detected and handled, and robust on the model inference may be facilitated.
[0040] In another aspect, a network device may transmit, to a terminal device, a request for availability information of a model input for a model. The terminal device may transmit, to the network device, the availability information of the model input comprising at least one of the following: an indication of whether measurement results in the model input are available, identity information associated with a set of undetectable or unreliable measurement results in the model input, or the measurement results in the model input. In this way, information of a model input of a network (NW) -side model may be provided to NW, and a management of the NW-side model may be facilitated.
[0041] In still another aspect, a terminal device may receive a list of model configurations for a model from a network device. A model configuration in the list of model configurations may comprise information of a model input and a model output for the model and an applicable condition associated with the model input and the model output. If the applicable condition is fulfilled, the terminal device may perform a model inference based on the model input and the model output. In this way, a model inference may be adjusted with an applicable condition, and thus robust on the model inference may be facilitated.
[0042] Principles and implementations of the present disclosure will be described in detail below with reference to the figures.
[0043] EXAMPLE OF COMMUNICATION NETWORK
[0044] FIG. 1 illustrates a schematic diagram of an example communication network 100 in which some embodiments of the present disclosure can be implemented. As shown in FIG. 1, the communication network 100 may include a terminal device 110 and a network device 120. In some embodiments, the network device 120 may provide one or more serving cells (not shown) to serve the terminal device 110.
[0045] The terminal device 110 may have a plurality of beams (not shown) , and the network device 120 may have a plurality of beams (not shown) . A channel (or called as a sub-channel in this case) may be formed between one of the plurality of beams of the terminal device 110 and one of the plurality of beams of the network device 120. The terminal device 110 may transmit information to the network device 120 or receive information from the network device 120 via one or more sub-channels.
[0046] It is to be understood that the number of devices in FIG. 1 is given for the purpose of illustration without suggesting any limitations to the present disclosure. The communication network 100 may include any suitable number of network devices and / or terminal devices and / or other network elements adapted for implementing implementations of the present disclosure.
[0047] As shown in FIG. 1, the terminal device 110 and the network device 120 may communicate with each other via a channel such as a wireless communication channel. The communications in the communication network 100 may conform to any suitable standards including, but not limited to, global system for mobile communications (GSM) , long term evolution (LTE) , LTE-evolution, LTE-advanced (LTE-A) , new radio (NR) , wideband code division multiple access (WCDMA) , code division multiple access (CDMA) , GSM EDGE radio access network (GERAN) , machine type communication (MTC) and the like. The embodiments of the present disclosure may be performed according to any generation communication protocols either currently known or to be developed in the future. Examples of the communication protocols include, but not limited to, the first generation (1G) , the second generation (2G) , 2.5G, 2.75G, the third generation (3G) , the fourth generation (4G) , 4.5G, the fifth generation (5G) communication protocols, 5.5G, 5G-advanced networks, or the sixth generation (6G) networks.
[0048] FIG. 2 illustrates a schematic diagram 200 of an example spatial domain prediction in which some embodiments of the present disclosure can be implemented. As shown in a scenario 210 of FIG. 2, quality of beams in a beam set {a1, a3, b1, b3} may be measured to predict quality of beams in a beam set {a2, a4, b2, b4} . For example, a model input is measured quality of the beams a1, a3, b1 and b3, and a model output is predicted quality of the beams a2, a4, b2 and b4.
[0049] As shown in a scenario 220 of FIG. 2, quality of cells in a cell set {cell 1, cell 3} may be measured to predict quality of cells in a cell set {cell 2, cell 4} . For example, a model input is measured quality of the cell 1 and cell 3, and a model output is predicted quality of the cell 2 and cell 4. These scenarios 210 and 220 may be called as a spatial domain prediction.
[0050] To perform the spatial domain prediction (i.e., perform a model inference) , a terminal device is required to be able to derive the model input as required by the model inference. However, based on real measurement environment, sometimes the terminal device may not derive all inputs for which a model wishes. For example, the terminal device may not detect or identify a specific frequency configured by NW due to e.g., weak signal strength or NW operations. Thus, a cell / beam measurement result corresponding to the specific frequency may not be used as the model input. In some other scenarios, quality of the model input sometime is not always good enough, which results in degradation of the model inference.
[0051] In view of this, embodiments of the present disclosure provide solutions of model management so as to enhance robust of a model inference. For illustration, the solutions will be detailed below with reference to FIGs. 3 to 5.
[0052] EXAMPLE IMPLEMENTATION OF MONITORING OF MODEL INFERENCE
[0053] Embodiments of the present disclosure provide a solution of detecting and handling a problem in a model inference of a UE-side model. The solution will be described in connection with FIG. 3.
[0054] FIG. 3 illustrates a signaling chart illustrating an example process 300 of communication in accordance with embodiments of the present disclosure. For the purpose of discussion, the process 300 will be described with reference to FIG. 1. The process 300 may involve the terminal device 110 and the network device 120 as illustrated in FIG. 1. It is to be understood that the steps and the order of the steps in FIG. 3 are merely for illustration, and not for limitation. For example, the order of the steps may be changed. Some of the steps may be omitted or any other suitable additional steps may be added. It is assumed that one or more models are deployed at the terminal device 110. The terminal device 110 is served by the network device 120.
[0055] As shown in FIG. 3, the network device 120 may transmit 310 a configuration of a model management to the terminal device 110. In some embodiments, the configuration may indicate information for determining a problem in a model inference. In some embodiments, the configuration may indicate information of the model management upon the problem occurs in the model inference. For example, the configuration may indicate any of time durations, number thresholds, or numbers of model inference cycles for the model management, e.g., any of first to fifth time durations, a first or second threshold value, a first or second range, a reference distribution, any of first to fifth number thresholds, or first or second number of model inference cycles that will be described later. In some alternative embodiments, any of these time durations, number thresholds, or numbers of model inference cycles may be predefined. In some alternative embodiments, the model management may be depended on UE implementation without any configurations. It is to be understood that the configuration may comprise any other suitable information, and the present disclosure does not limit this aspect.
[0056] Continuing to refer to FIG. 3, the terminal device 110 may determine 320 whether a problem occurs in a model inference of a model (for convenience, also referred to as a first model herein) .
[0057] With reference to FIG. 3, in some embodiments, the terminal device 110 may determine 321 whether the problem occurs in the model inference based on detectability of a model input for the first model. In some embodiments, if the model input for the first model is undetectable, the terminal device 110 may determine that the problem occurs in the model inference.
[0058] In the context of the present disclosure, the term ‘undetectable model input’ means that the terminal device 110 is unable to identify or detect a cell or beam corresponding to a configured model input. The term ‘detectable model input’ means that the terminal device 110 is able to identify or detect a cell or beam corresponding to a configured model input. The term ‘undetectable model input’ may be interchangeably used with ‘undetectable measurement result (s) ’ , and the term ‘detectable model input’ may be interchangeably used with ‘detectable measurement result (s) ’ .
[0059] In some embodiments, if number of detectable measurement results in the model input is smaller than or equal to a number threshold (also referred to as the first number threshold herein) , the terminal device 110 may determine that the model input is undetectable. In some embodiments, if number of undetectable measurement results in the model input is greater than or equal to a number threshold (also referred to as the second number threshold herein) , the terminal device 110 may determine that the model input is undetectable. In some embodiments, if one (i.e., any one) of measurement results in the model input is undetectable, the terminal device 110 may determine that the model input is undetectable. In some embodiments, if all the measurement results in the model input are undetectable, the terminal device 110 may determine that the model input is undetectable. It is to be understood that any combinations of the above embodiments may also be feasible for determination of the detectability of the model input.
[0060] In some embodiments, if the model input for the first model is undetectable for a time duration (also referred to as the first time duration herein) , the terminal device 110 may determine that the problem occurs in the model inference. In some embodiments, if the model input for the first model is undetectable for number of model inference cycles (also referred to as the first number of model inference cycles herein) , the terminal device 110 may determine that the problem occurs in the model inference. In this way, the problem in the model inference may be surely determined based on detectability of the model input.
[0061] With reference to FIG. 3, in some embodiments, the terminal device 110 may determine 322 whether the problem occurs in the model inference based on reliability of the model input for the first model. In some embodiments, if the model input for the first model is unreliable, the terminal device 110 may determine that the problem occurs in the model inference.
[0062] In the context of the present disclosure, the term ‘unreliable model input’ means that the terminal device 110 is able to identify or detect a cell or beam corresponding to a configured model input, but quality of the cell or beam is lower than or equal to a threshold value (also referred to as a first threshold value herein) . The term ‘reliable model input’ means that the terminal device 110 is able to identify or detect a cell or beam corresponding to a configured model input, and quality of the cell or beam is higher than or equal to a threshold value (also referred to as a second threshold value herein) . The term ‘unreliable model input’ may be interchangeably used with ‘unreliable measurement result (s) ’ or ‘first measurement result (s) ’ , and the term ‘reliable model input’ may be interchangeably used with ‘reliable measurement result (s) ’ or ‘second measurement result (s) ’ .
[0063] In some embodiments, if number of the first measurement results (i.e., with low quality) in the model input is greater than or equal to a number threshold (also referred to as a third number threshold herein) , the terminal device 110 may determine that the model input is unreliable. In some embodiments, if number of second measurement results (i.e., with high quality) in the model input is smaller than or equal to a number threshold (also referred to as a fourth number threshold herein) , the terminal device 110 may determine that the model input is unreliable. In some embodiments, if one (i.e., any one) of measurement results in the model input has a value lower than or equal to the first threshold value, the terminal device 110 may determine that the model input is unreliable. In some embodiments, if all the measurement results in the model input have values lower than or equal to the first threshold value, the terminal device 110 may determine that the model input is unreliable.
[0064] In some embodiments, if a statistical value or parameter of the measurement results in the model input is in a range (also referred to as a first range herein) , the terminal device 110 may determine that the model input is unreliable. For example, if an average value of the measurement results in the model input is lower than or equal to an average threshold, the terminal device 110 may determine that the model input is unreliable. In another example, if a variance of the measurement results in the model input is greater than or equal to a variance threshold, the terminal device 110 may determine that the model input is unreliable.
[0065] In some embodiments, if a cumulative distribution function (CDF) value of the number of the first measurement results in the model input is in a range (also referred to as a second range herein) , the terminal device 110 may determine that the model input is unreliable. For example, if the CDF value of the number of the first measurement results in the model input is lower than or equal to a first CDF threshold, the terminal device 110 may determine that the model input is unreliable. In another example, if the CDF value of the number of the first measurement results in the model input is greater than or equal to a second CDF threshold, the terminal device 110 may determine that the model input is unreliable.
[0066] In some embodiments, if a data distribution of the measurement results in the model input is deviated from the reference distribution, the terminal device 110 may determine that the model input is unreliable. For example, if similarity between the data distribution and the reference distribution is lower than or equal to a similarity threshold (i.e., the similarity is not good) , the terminal device 110 may determine that the model input is unreliable. It is to be understood that any other metrics of the data distribution may also be adopted. It is also to be understood that any combinations of the above embodiments may also be feasible for determination of the reliability of the model input.
[0067] In some embodiments, if the model input for the first model is unreliable for a time duration (also referred to as the second time duration herein) , the terminal device 110 may determine that the problem occurs in the model inference. In some embodiments, if the model input for the first model is unreliable for number of model inference cycles (also referred to as the second number of model inference cycles herein) , the terminal device 110 may determine that the problem occurs in the model inference. In this way, the problem in the model inference may be surely determined based on reliability of the model input.
[0068] With reference to FIG. 3, in some embodiments, the terminal device 110 may determine 323 whether the problem occurs in the model inference based on a requirement on a processing time of the model inference. In some embodiments, if the requirement on the processing time of the model inference is unfulfilled, the terminal device 110 may determine that the problem occurs in the model inference.
[0069] In some embodiments, for each model inference cycle, the requirement on the processing time of the model inference may be defined based on a time duration (also referred to as a fourth time duration herein) required for processing the model inference by the first model and a time duration (also referred to as a fifth time duration herein) required for detecting the model input.
[0070] For example, the requirement on the processing time of the model inference may be defined as shown in an equation (1) below. Tr = Tp + Ts + C (1)
[0071] where Tr denotes the requirement on the processing time of the model inference, Tp denotes the time duration required for processing the model inference by the first model, Ts denotes the time duration required for detecting the model input, and C denotes a time duration required for other purposes. It is to be noted that C is optional.
[0072] In other words, Tp may refer to a time duration the first model itself will take for processing the model inference. Ts means that the terminal device 110 may extent a time duration of the model inference by Ts if the model input is not available / detectable before. Otherwise, Ts = 0. In some embodiments, Ts may be equal to a SSB measurement timing configuration (SMTC) periodicity corresponding the cell / beam of the model input. In some embodiments, Ts may be determined based on an equation (2) or (3) below. Ts = MAX (TSMTC, TMGRP) (2) Ts = MAX (TSMTC, TMGRP, TDRX) (3)
[0073] where Ts denotes the time duration required for detecting the model input, TSMTC denotes a SMTC periodicity, TMGRP denotes a measurement gap repetition period (MGRP) periodicity, and TDRX denotes a discontinuous reception (DRX) periodicity.
[0074] In some embodiments, the terminal device 110 may determine the requirement on the processing time of the model inference based on the fourth time duration and the fifth time duration, e.g., based on any of the equations (1) to (3) or any other suitable ways. If the model inference is unfinished within the requirement on the processing time of the model inference, the terminal device 110 may determine that the requirement on the processing time of the model inference is unfulfilled. In this case, the terminal device 110 may determine that the problem occurs in the model inference.
[0075] Continuing to refer to FIG. 3, upon determination that the problem occurs in the model inference, the terminal device 110 may perform 330 a model management.
[0076] In some embodiments, the terminal device 110 may switch from a AI scheme to a non-AI scheme. In other words, the terminal device 110 may stop using the AI / ML model and fallback to the non-AI scheme, e.g., performing real measurements based on a measurement configuration.
[0077] In some embodiments, the terminal device 110 may switch from a current configuration (also referred to as a first configuration herein) of the first model to another configuration (also referred to as a second configuration herein) of the first model. In some embodiments, the terminal device 110 may select, as the second configuration, a model configuration for which a model input is detectable and reliable. For example, the terminal device 110 may determine whether a model input for a model configuration is detectable or reliable as described in the step 320. If the model input for the model configuration is detectable or reliable, the terminal device 110 may switch to the model configuration.
[0078] In some embodiments, the terminal device 110 may switch from the first model to another model (also referred to as a second model herein) . In some embodiments, if multiple models are deployed, the terminal device 110 may select, as the second model, a model for which a model input is detectable and reliable. For example, the terminal device 110 may determine whether a model input for a model is detectable or reliable as described in the step 320. If the model input for the model is detectable or reliable, the terminal device 110 may switch to the model.
[0079] In some embodiments, the terminal device 110 may skip a set of undetectable or unreliable measurement results in the model input. In some embodiments, the terminal device 110 may skip the set of undetectable or unreliable measurement results for a time duration (also referred to a third time duration herein) . After the time duration passes, the terminal device 110 may stop skipping the set of measurement results that is determined as being undetectable or unreliable previously, and may re-evaluate whether the problem occurs in the model inference.
[0080] It is to be understood that any combinations of the above model management actions may also be feasible.
[0081] Continuing to refer to FIG. 3, upon determination that the problem occurs in the model inference, the terminal device 110 may transmit 340 assistance information (also referred to as first information herein) to the network device 120.
[0082] In some embodiments, the first information may comprise information of the problem. In some embodiments, the information of the problem may comprise information of the model input corresponding to the set of undetectable or unreliable measurement results. For example, the information of the problem may comprise one or more cell or beam IDs corresponding to the set of undetectable or unreliable measurement results. In some embodiments, the information of the problem may indicate that the requirement on the processing time of the model inference is unfulfilled.
[0083] In some embodiments, the first information may comprise a request for modifying a configuration of the model input. In some embodiments, the terminal device 110 may request the network device 120 to modify the model input, e.g., by providing a preference on the model input to the network device 120.
[0084] In some embodiments, the first information may comprise information of the model management that is performed or to be performed by the terminal device 110. For example, the information of the model management may comprise an indication that the set of undetectable or unreliable measurement results is skipped or to be skipped for the third time duration. In some embodiments, the information of the model management may indicate the switching to the non-AI scheme. In some embodiments, the information of the model management may indicate the model configuration switching. In some embodiments, the information of the model management may indicate the model switching. It is to be understood that any combinations of the above information may also be feasible.
[0085] Continuing to refer to FIG. 3, upon determination that the model input for the first model is undetectable or is undetectable for the first time duration or for the first number of model inference cycles, the terminal device 110 may perform 350 the model inference based on a processed model input. In other words, the terminal device 110 may process the model input to optimize or adjust the model inference.
[0086] In some embodiments, upon determination that number of undetectable measurement results in the model input is smaller than or equal to a number threshold (also referred to as a fifth number threshold herein) , the terminal device 110 may process the model input.
[0087] In some embodiments, the terminal device 110 may process the model input by setting a value of an undetectable measurement result to be a specific value (also referred to a first value herein) . In some embodiments, the terminal device 110 may determine the first value based on a minimum required received level in a cell. For example, the terminal device 110 may set the value of the undetectable measurement result to be the minimum required received level in the cell, e.g., Qrxlevmin (e.g., -156 dBm) configured in system information block 1 (SIB1) for cell selection.
[0088] In some embodiments, the terminal device 110 may determine the first value based on a configured or predefined value. For example, the terminal device 110 may set the value of the undetectable measurement result to be a value that is configured by the network device 120 or is predefined (e.g., -156 dBm) .
[0089] In some embodiments, the terminal device 110 may determine the first value based on an average value of a set of detectable measurement results (e.g., all the detectable measurement results) in the model input.
[0090] In some embodiments, the terminal device 110 may determine the first value based on a measurement result corresponding to selected information of the model input, e.g., selected cell or beam ID or frequency. For example, the terminal device 110 may set the value of the undetectable measurement result to be a measurement result of a serving cell or other selected cell.
[0091] Then the terminal device 110 may continue performing the model inference after setting the value of the undetectable measurement result to be the first value. In some embodiments, if the number of undetectable measurement results in the model input is greater than the fifth number threshold, the terminal device 110 may not set the undetectable measurement result to be the first value. In this case, the terminal device 110 may consider that the problem occurs in the model inference and perform the model management as described above.
[0092] For illustration, an example procedure is described as below.
[0093] Upon performing the model inference, UE shall:
[0094] Consider the model input as undetectable input if the input cannot be detected;
[0095] Consider the model input as low quality input as following:
[0096] if the statistical parameter / value of input is below (e.g., average) / greater (e.g., variance) than a threshold, considered problematic input.
[0097] if the CDF of the number of problematic input is below / greater than a threshold, considered problematic input.
[0098] if the data distribution of input is deviated from a reference distribution (e.g., similarity is not good) , considered as problematic input.
[0099] 1 > if the undetectable input number is greater than a threshold_num (for a duration / timer / period or multiple times inference) , or,
[0100] 1> if the low quality input number is greater than a threshold_num (for a duration / timer / period or multiple times inference) ,
[0101] 2> stop using AI model and fallback to non-AI method, i.e., performing legacy measurement based on measConfig;
[0102] 2> switch AI model configuration, e.g., modify the model configuration to the one of which the model input is not problematic;
[0103] 2> switch AI model, e.g., if multiple models are deployed, select the model of which the model input is not problematic;
[0104] 2> skip the undetectable input for a duration, which means the UE continue performing the model inference without this input;
[0105] 2> provide assistance information to NW, including, undetectable input information, e.g., the cell / beam ID corresponding the input, or request NW to modify the input, e.g., provide the preference input to NW, or skip the undetectable input for a duration.
[0106] In this example procedure, an information element (IE) ‘threshold_num’ denotes a number threshold, and an IE ‘measConfig’ denotes a measurement configuration.
[0107] So far, detection and handling of a problem in a model inference of a UE-side model are described. With the process 300, robust on the model inference of the UE-side model may be facilitated.
[0108] EXAMPLE IMPLEMENTATION OF AVAILABILITY OF MODEL INPUT OF NW-SIDE MODEL
[0109] Embodiments of the present disclosure provide a solution of providing information of a model input of a NW-side model. The solution will be described in connection with FIG. 4.
[0110] FIG. 4 illustrates a signaling chart illustrating another example process 400 of communication in accordance with embodiments of the present disclosure. For the purpose of discussion, the process 400 will be described with reference to FIG. 1. The process 400 may involve the terminal device 110 and the network device 120 as illustrated in FIG. 1. It is to be understood that the steps and the order of the steps in FIG. 4 are merely for illustration, and not for limitation. For example, the order of the steps may be changed. Some of the steps may be omitted or any other suitable additional steps may be added. It is assumed that one or more models are deployed at the network device 120. The terminal device 110 is served by the network device 120.
[0111] As shown in FIG. 4, the network device 120 may transmit 410, to the terminal device 110, a request for availability information of a model input for a model. In some embodiments, the request may comprise information of the model input of the model, e.g., cell or beam ID (s) .
[0112] With reference to FIG. 4, upon reception of the request, the terminal device 110 may transmit 420 the availability information of the model input of the model to the network device 120.
[0113] In some embodiments, the availability information of the model input may comprise an indication of whether measurement results in the model input are available. In other words, the terminal device 110 may directly send an indication to indicate that whether all measurement results as the model input are available. For example, the indication may be 1 bit. If all measurement results as the model input are available, a value of the bit is set to be 1. If not all measurement results as the model input are available, a value of the bit is set to be 0. It is to be understood that any other suitable bit values may also be feasible.
[0114] In some embodiments, the availability information of the model input may comprise identity information associated with a set of undetectable or unreliable measurement results in the model input, e.g., undetectable or unreliable cell or beam ID (s) .
[0115] In some embodiments, the availability information of the model input may comprise all the measurement results in the model input, e.g., the measurement result of each cell or beam associated with the model input.
[0116] With the process 400, information of a model input of a NW-side model may be provided to NW, and NW implementation may be optimized.
[0117] EXAMPLE IMPLEMENTATION OF CONFIGURATION OF MODEL INFERENCE
[0118] Embodiments of the present disclosure provide a solution of configuring a model inference. The solution will be described in connection with FIG. 5.
[0119] FIG. 5 illustrates a signaling chart illustrating still another example process 500 of communication in accordance with embodiments of the present disclosure. For the purpose of discussion, the process 500 will be described with reference to FIG. 1. The process 500 may involve the terminal device 110 and the network device 120 as illustrated in FIG. 1. It is to be understood that the steps and the order of the steps in FIG. 5 are merely for illustration, and not for limitation. For example, the order of the steps may be changed. Some of the steps may be omitted or any other suitable additional steps may be added. It is assumed that one or more models are deployed at the terminal device 110. The terminal device 110 is served by the network device 120.
[0120] As shown in FIG. 5, the terminal device 110 may transmit 510, to the network device 120, assistance information (also referred to as second information herein) for model configurations. In some embodiments, the terminal device 110 may transmit the second information via a UE assistance information (UAI) message or any other suitable RRC messages. In some embodiments, the terminal device 110 may transmit the second information via a medium access control (MAC) control element (CE) .
[0121] In some embodiments, the second information may comprise a preference on an applicable condition for each model configuration. For example, the second information may comprise a preference on an applicable range for a model inference, e.g., a cell ID, a timing advance (TA) list, a radio access network-based notification area (RNA) list or a cell list.
[0122] In some embodiments, the second information may comprise a preference on the model input or output for each cell. In some embodiments, the second information may comprise a preference on a model input or output for a serving cell. For example, the second information may comprise an expected cell or beam ID as the model input or output.
[0123] In some embodiments, the second information may comprise a preference on maximum number of the model inputs or outputs. In some embodiments, the second information may comprise a preference on maximum number of applicable cells. In some embodiments, the second information may comprise a preference on minimum number of model inputs or outputs. In some embodiments, the second information may comprise a preference on minimum number of applicable cells.
[0124] In some embodiments, once the second information is available at the terminal device 110, the terminal device 110 may report the second information to the network device 120 for better NW implementation.
[0125] Continuing to refer to FIG. 5, the network device 120 may transmit 520 a list of model configurations for a model to the terminal device 110. In some embodiments, the network device 120 may generate the list of model configurations based on the second information reported by the terminal device 110. Alternatively, the network device 120 may generate the list of model configurations based on NW implementation.
[0126] In some embodiments, the list of model configuration may comprise information of the model input for the model, information of the model output for the model, and the applicable condition associated with the model input and output. In other words, the model input and / or output may be different under different scenarios or conditions.
[0127] In some embodiments, the applicable condition may be associated with the terminal device 110. For example, the applicable condition may be associated with at least one of the following: signal strength of the terminal device 110, a moving speed of the terminal device 110, a moving direction of the terminal device 110, or a location of the terminal device 110.
[0128] For illustration, an example configuration may be described as shown in Table 1.
[0129] Table 1
[0130] In this example, the network device 120 provide multiple model configurations. When the terminal device 110 is in different scenarios / conditions, different model configurations may be applied. For example, when RSRP of the terminal device 110 is greater than threshold_1, a model configuration comprising a model input {cell 1, cell 2} and a model output {cell 4} may be applied. It is to be understood that Table 1 is merely an example, and does not limit the present disclosure. Any other suitable forms may also be feasible.
[0131] In some embodiments, the applicable condition may be associated with a range where the model inference is allowed. For example, the applicable condition may be associated with at least one of the following: a TA list, a RNA list, a cell list, or a cell ID. In some embodiments, when the terminal device 110 camps on a cell which belongs to the TA list, the model is allowed to be applied. In some embodiments, when the terminal device 110 camps on a cell which belongs to the RNA list, the model is allowed to be applied. In some embodiments, when the terminal device 110 camps on a cell which belongs to the cell list, the model is allowed to be applied. In some embodiments, when the terminal device 110 camps on a cell which matches the cell ID, the model is allowed to be applied.
[0132] For illustration, an example cell-specific configuration list may be described as shown in Table 2.
[0133] Table 2
[0134] In this example, the network device 120 provide multiple model configurations. When the terminal device 110 camps on different cells, different model configurations may be applied. For example, when the terminal device 110 camps on cell 1, a model configuration comprising a model input {cell 1, cell 2, cell 3} and a model output {cell 4, cell 5} may be applied. It is to be understood that Table 2 is merely an example, and does not limit the present disclosure. Any other suitable forms may also be feasible.
[0135] In some embodiments, the model configuration may be associated with a measurement ID, the information of the model input may be included in a measurement object associated with the measurement ID, and the information of the model output may be included in a report configuration associated with the measurement ID. In other words, NW may configure a measurement ID (i.e., IE ‘measID’ ) for the model inference, and a measurement object and a report configuration linked to the measurement ID provide a related configuration for performing the model inference. In this way, a framework of radio resource management (RRM) measurement configuration (i.e., IE ‘measConfig’ ) may be reused.
[0136] In some embodiments, the applicable condition associated with the model input and output may be included in the report configuration. In some embodiments, the applicable condition associated with the model input and output may be included in the measurement object.
[0137] In some embodiments, the information of the model input may comprise a set of frequencies, i.e., one or multiple frequencies corresponding to measurement results used as the model input. In some embodiments, the information of the model input may comprise a set of cell IDs, i.e., one or multiple cell IDs corresponding to measurement results used as the model input. In some embodiments, the information of the model input may comprise a set of beam IDs, i.e., one or multiple beam IDs corresponding to measurement results used as the model input. It is to be understood that a beam ID may be a reference signal (e.g., channel status information reference signal (CSI-RS) ) resource ID.
[0138] In some embodiments, the information of the model output may comprise a set of frequencies, i.e., one or multiple frequencies corresponding to measurement results used as the model output. In some embodiments, the information of the model input may comprise a set of cell IDs, i.e., one or multiple cell IDs corresponding to measurement results used as the model output. In some embodiments, the information of the model input may comprise a set of beam IDs, i.e., one or multiple beam IDs corresponding to measurement results used as the model output.
[0139] In some embodiments, upon reception of a measurement configuration specific to a model, the terminal device 110 may set a model input of the model based on a measurement object associated with the measurement configuration, and set a model output of the model based on a report configuration associated with the measurement configuration.
[0140] Continuing to refer to FIG. 5, the terminal device 110 may perform 530 a model inference based on a model input and a model output when an applicable condition associated with the model input and the model output is fulfilled.
[0141] With the process 500, a model inference may be adjusted with an applicable condition, and thus robust on the model inference may be facilitated.
[0142] It is to be understood that operations or processes described above in connection with FIGs. 3 to 5 may be performed separately or in any suitable combination.
[0143] EXAMPLE IMPLEMENTATION OF METHODS
[0144] Corresponding to the above process, embodiments of the present disclosure provide methods of communication implemented at a terminal device. These methods will be described below with reference to FIGs. 6 to 8.
[0145] FIG. 6 illustrates a flowchart of an example method 600 of communication implemented at a terminal device in accordance with some embodiments of the present disclosure. For example, the method 600 may be performed at the terminal device 110 as shown in FIG. 1. For the purpose of discussion, in the following, the method 600 will be described with reference to FIG. 1. It is to be understood that the method 600 may include additional blocks not shown and / or may omit some blocks as shown, and the scope of the present disclosure is not limited in this regard.
[0146] At block 610, the terminal device 110 may determine that a problem occurs in a model inference of a first model. In some embodiments, the terminal device 110 may determine that the problem occurs in the model inference based on at least one of the following: a model input for the first model is undetectable or is undetectable for a first time duration or for a first number of model inference cycles, the model input for the first model is unreliable or is unreliable for a second time duration or for a second number of model inference cycles, or a requirement on a processing time of the model inference is unfulfilled.
[0147] In some embodiments, the terminal device 110 may determine that the model input is undetectable based on at least one of the following: number of detectable measurement results in the model input is smaller than or equal to a first number threshold; number of undetectable measurement results in the model input is greater than or equal to a second number threshold; one of measurement results in the model input is undetectable; or all the measurement results in the model input are undetectable.
[0148] In some embodiments, the terminal device 110 may determine that the model input is unreliable based on at least one of the following: number of first measurement results in the model input is greater than or equal to a third number threshold, values of the first measurement results being lower than or equal to a first threshold value; number of second measurement results in the model input is smaller than or equal to a fourth number threshold, values of the second measurement results being higher than or equal to a second threshold value; one of measurement results in the model input has a value lower than or equal to the first threshold value; all the measurement results in the model input have values lower than or equal to the first threshold value; a statistical value of the measurement results in the model input is in a first range; a CDF value of the number of first measurement results in the model input is in a second range; or a data distribution of the measurement results in the model input is deviated from a reference distribution.
[0149] In some embodiments, the terminal device 110 may determine the requirement on the processing time of the model inference based on a fourth time duration required for processing the model inference by the first model and a fifth time duration required for detecting the model input. If the model inference is unfinished within the requirement on the processing time of the model inference, the terminal device 110 may determine that the requirement is unfulfilled.
[0150] At block 620, the terminal device 110 may perform a model management. In some embodiments, the model management may comprise at least one of the following: switching to a non-AI scheme, switching from a first configuration of the first model to a second configuration of the first model, switching from the first model to a second model, or skipping a set of undetectable or unreliable measurement results in the model input or skipping the set of undetectable or unreliable measurement results for a third time duration.
[0151] In some embodiments, if the problem occurs, the terminal device 110 may transmit first information to the network device 120. In some embodiments, the first information may comprise at least one of the following: information of the problem, a request for modifying a configuration of the model input, or information of the model management. In some embodiments, the information of the problem may comprise information of the model input corresponding to the set of undetectable or unreliable measurement results. In some embodiments, the information of the model management may comprise an indication of the skipping of the set of undetectable or unreliable measurement results for the third time duration.
[0152] In some embodiments, if the model input for the first model is undetectable or is undetectable for the first time duration or for the first number of model inference cycles, the terminal device 110 may process the model input by setting a value of an undetectable measurement result to be a first value, and perform the model inference based on the model input processed. In some embodiments, the terminal device 110 may determine the first value based on one of the following: a minimum required received level in a cell, a configured or predefined value, an average value of a set of detectable measurement results in the model input, or a measurement result corresponding to selected information of the model input.
[0153] With the method 600, a problem in a model inference may be detected and handled, and robust on the model inference may be facilitated.
[0154] FIG. 7 illustrates a flowchart of another example method 700 of communication implemented at a terminal device in accordance with some embodiments of the present disclosure. For example, the method 700 may be performed at the terminal device 110 as shown in FIG. 1. For the purpose of discussion, in the following, the method 700 will be described with reference to FIG. 1. It is to be understood that the method 700 may include additional blocks not shown and / or may omit some blocks as shown, and the scope of the present disclosure is not limited in this regard.
[0155] At block 710, the terminal device 110 may receive, from the network device 120, a request for availability information of a model input for a model.
[0156] At block 720, the terminal device 110 may transmit the availability information of the model input to the network device 120. In some embodiments, the availability information of the model input may comprise at least one of the following: an indication of whether measurement results in the model input are available, identity information associated with a set of undetectable or unreliable measurement results in the model input, or the measurement results in the model input.
[0157] With the method 700, information of a model input of a NW-side model may be provided to NW, and a management of the NW-side model may be facilitated.
[0158] FIG. 8 illustrates a flowchart of still another example method 800 of communication implemented at a terminal device in accordance with some embodiments of the present disclosure. For example, the method 800 may be performed at the terminal device 110 as shown in FIG. 1. For the purpose of discussion, in the following, the method 800 will be described with reference to FIG. 1. It is to be understood that the method 800 may include additional blocks not shown and / or may omit some blocks as shown, and the scope of the present disclosure is not limited in this regard.
[0159] At block 810, the terminal device 110 may receive, from the network device 120, a list of model configurations for a model. In some embodiments, the model configuration in the list of model configurations may comprise information of a model input and information of a model output for the model and an applicable condition associated with the model input and the model output.
[0160] In some embodiments, the applicable condition may be associated with at least one of the following: signal strength of the terminal device, a moving speed of the terminal device, a moving direction of the terminal device, a location of the terminal device, a TA list, a RNA list, a cell list, or a cell ID.
[0161] In some embodiments, the model configuration may be associated with a measurement ID, the information of the model input may be included in a measurement object associated with the measurement ID, and the information of the model output may be included in a report configuration associated with the measurement ID.
[0162] In some embodiments, the information of the model input or the information of the model output may comprise one of the following: a set of frequencies, a set of cell identities, or a set of beam identities.
[0163] At block 820, the terminal device 110 may determine that the applicable condition is fulfilled.
[0164] At block 830, the terminal device 110 may perform a model inference based on the model input and the model output.
[0165] In some embodiments, the terminal device 110 may transmit, to the network device 120, second information comprising at least one of the following: a preference on the applicable condition for the model configuration in the list of model configuration; a preference on the model input or output for a serving cell; a preference on the model input or output for a cell in a cell list; a preference on maximum number of the model inputs or outputs; a preference on maximum number of applicable cells; a preference on minimum number of model inputs or outputs; or a preference on minimum number of applicable cells.
[0166] It is to be understood that operations of the methods 600, 700 and 800 correspond to that described in connection with FIGs. 3 to 5, and thus other details are not repeated here for conciseness.
[0167] EXAMPLE IMPLEMENTATION OF DEVICES
[0168] FIG. 9 is a simplified block diagram of a device 900 that is suitable for implementing embodiments of the present disclosure. The device 900 can be considered as a further example implementation of the terminal device 110 or the network device 120 as shown in FIG. 1. Accordingly, the device 900 can be implemented at or as at least a part of the terminal device 110 or the network device 120.
[0169] As shown, the device 900 includes a processor 910, a memory 920 coupled to the processor 910, a suitable transceiver 940 coupled to the processor 910, and a communication interface coupled to the transceiver 940. The memory 910 stores at least a part of a program 930. The transceiver 940 may be for bidirectional communications or a unidirectional communication based on requirements. The transceiver 940 may include at least one of a transmitter 942 or a receiver 944. The transmitter 942 and the receiver 944 may be functional modules or physical entities. The transceiver 940 has at least one antenna to facilitate communication, though in practice an access node mentioned in this application may have several ones. The communication interface may represent any interface that is necessary for communication with other network elements, such as X2 / Xn interface for bidirectional communications between eNBs / gNBs, S1 / NG interface for communication between a mobility management entity (MME) / access and mobility management function (AMF) / SGW / UPF and the eNB / gNB, Un interface for communication between the eNB / gNB and a relay node (RN) , or Uu interface for communication between the eNB / gNB and a terminal device.
[0170] The program 930 is assumed to include program instructions that, when executed by the associated processor 910, enable the device 900 to operate in accordance with the embodiments of the present disclosure, as discussed herein with reference to FIGs. 1 to 8. The embodiments herein may be implemented by computer software executable by the processor 910 of the device 900, or by hardware, or by a combination of software and hardware. The processor 910 may be configured to implement various embodiments of the present disclosure. Furthermore, a combination of the processor 910 and memory 920 may form processing means 950 adapted to implement various embodiments of the present disclosure.
[0171] The memory 920 may be of any type suitable to the local technical network and may be implemented using any suitable data storage technology, such as a non-transitory computer readable storage medium, semiconductor based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory, as non-limiting examples. While only one memory 920 is shown in the device 900, there may be several physically distinct memory modules in the device 900. The processor 910 may be of any type suitable to the local technical network, and may include one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs) and processors based on multicore processor architecture, as non-limiting examples. The device 900 may have multiple processors, such as an application specific integrated circuit chip that is slaved in time to a clock which synchronizes the main processor.
[0172] In some embodiments, a terminal device comprises a circuitry configured to: determine that a problem occurs in a model inference of a first model based on at least one of the following: a model input for the first model is undetectable or is undetectable for a first time duration or for a first number of model inference cycles, the model input for the first model is unreliable or is unreliable for a second time duration or for a second number of model inference cycles, or a requirement on a processing time of the model inference is unfulfilled; and perform a model management comprising at least one of the following: switching to a non-AI scheme, switching from a first configuration of the first model to a second configuration of the first model, switching from the first model to a second model, or skipping a set of undetectable or unreliable measurement results in the model input or skipping the set of undetectable or unreliable measurement results for a third time duration.
[0173] In some embodiments, a terminal device comprises a circuitry configured to: receive, from a network device, a request for availability information of a model input for a model; and transmit, to the network device, the availability information of the model input comprising at least one of the following: an indication of whether measurement results in the model input are available, identity information associated with a set of undetectable or unreliable measurement results in the model input, or the measurement results in the model input.
[0174] In some embodiments, a terminal device comprises a circuitry configured to: receive, from a network device, a list of model configurations for a model, a model configuration in the list of model configurations comprising information of a model input and information of a model output for the model and an applicable condition associated with the model input and the model output; and in accordance with a determination that the applicable condition is fulfilled, perform a model inference based on the model input and the model output.
[0175] The term ‘circuitry’ used herein may refer to hardware circuits and / or combinations of hardware circuits and software. For example, the circuitry may be a combination of analog and / or digital hardware circuits with software / firmware. As a further example, the circuitry may be any portions of hardware processors with software including digital signal processor (s) , software, and memory (ies) that work together to cause an apparatus, such as a terminal device or a network device, to perform various functions. In a still further example, the circuitry may be hardware circuits and or processors, such as a microprocessor or a portion of a microprocessor, that requires software / firmware for operation, but the software may not be present when it is not needed for operation. As used herein, the term circuitry also covers an implementation of merely a hardware circuit or processor (s) or a portion of a hardware circuit or processor (s) and its (or their) accompanying software and / or firmware.
[0176] Generally, various embodiments of the present disclosure may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. Some aspects 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. While various aspects of embodiments of the present disclosure are illustrated and described as block diagrams, flowcharts, or using some other pictorial representation, it will be appreciated that the 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.
[0177] The present disclosure also provides at least one computer program product tangibly stored on a non-transitory computer readable storage medium. The computer program product includes computer-executable instructions, such as those included in program modules, being executed in a device on a target real or virtual processor, to carry out the process or method as described above with reference to FIGs. 1 to 8. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, or the like that perform particular tasks or implement particular abstract data types. The functionality of the program modules may be combined or split between program modules as desired in various embodiments. Machine-executable instructions for program modules may be executed within a local or distributed device. In a distributed device, program modules may be located in both local and remote storage media.
[0178] Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may execute entirely on a machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
[0179] The above program code may be embodied on a machine readable medium, which may be any tangible medium that may contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine readable medium may be a machine readable signal medium or a machine readable storage medium. A machine readable medium may include but not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine readable storage medium would include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM) , a read-only memory (ROM) , an erasable programmable read-only memory (EPROM or Flash memory) , an optical fiber, a portable compact disc read-only memory (CD-ROM) , an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0180] Further, while operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are contained in the above discussions, these should not be construed as limitations on the scope of the present disclosure, but rather as descriptions of features that may be specific to particular embodiments. Certain features that are described in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment may also be implemented in multiple embodiments separately or in any suitable sub-combination.
[0181] Although the present disclosure has been described in language specific to structural features and / or methodological acts, it is to be understood that the present disclosure defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.
Claims
1.A terminal device, comprising:a processor configured to cause the terminal device to:determine that a problem occurs in a model inference of a first model based on at least one of the following:a model input for the first model is undetectable or is undetectable for a first time duration or for a first number of model inference cycles,the model input for the first model is unreliable or is unreliable for a second time duration or for a second number of model inference cycles, ora requirement on a processing time of the model inference is unfulfilled; andperform a model management comprising at least one of the following:switching to a non-artificial intelligence (AI) scheme,switching from a first configuration of the first model to a second configuration of the first model,switching from the first model to a second model, orskipping a set of undetectable or unreliable measurement results in the model input or skipping the set of undetectable or unreliable measurement results for a third time duration.2.The terminal device of claim 1, wherein the terminal device is further caused to:in accordance with a determination that the problem occurs, transmit, to a network device, first information comprising at least one of the following:information of the problem,a request for modifying a configuration of the model input, orinformation of the model management.3.The terminal device of claim 2, wherein the information of the problem comprises information of the model input corresponding to the set of undetectable or unreliable measurement results, orwherein the information of the model management comprises an indication of the skipping of the set of undetectable or unreliable measurement results for the third time duration.4.The terminal device of claim 1, wherein the terminal device is further caused to:determine that the model input is undetectable based on at least one of the following:number of detectable measurement results in the model input is smaller than or equal to a first number threshold;number of undetectable measurement results in the model input is greater than or equal to a second number threshold;one of measurement results in the model input is undetectable; orall the measurement results in the model input are undetectable.5.The terminal device of claim 1, wherein the terminal device is further caused to:determine that the model input is unreliable based on at least one of the following:number of first measurement results in the model input is greater than or equal to a third number threshold, values of the first measurement results being lower than or equal to a first threshold value;number of second measurement results in the model input is smaller than or equal to a fourth number threshold, values of the second measurement results being higher than or equal to a second threshold value;one of measurement results in the model input has a value lower than or equal to the first threshold value;all the measurement results in the model input have values lower than or equal to the first threshold value;a statistical value of the measurement results in the model input is in a first range;a cumulative distribution function (CDF) value of the number of first measurement results in the model input is in a second range; ora data distribution of the measurement results in the model input is deviated from a reference distribution.6.The terminal device of claim 1, wherein the terminal device is further caused to:determine the requirement on the processing time of the model inference based on a fourth time duration required for processing the model inference by the first model and a fifth time duration required for detecting the model input; andin accordance with a determination that the model inference is unfinished within the requirement on the processing time of the model inference, determine that the requirement is unfulfilled.7.The terminal device of claim 1, wherein the terminal device is further caused to:in accordance with a determination that the model input for the first model is undetectable or is undetectable for the first time duration or for the first number of model inference cycles, process the model input by setting a value of an undetectable measurement result to be a first value; andperform the model inference based on the model input processed.8.The terminal device of claim 7, wherein the terminal device is further caused to:determine the first value based on one of the following:a minimum required received level in a cell,a configured or predefined value,an average value of a set of detectable measurement results in the model input, ora measurement result corresponding to selected information of the model input.9.A terminal device, comprising:a processor configured to cause the terminal device to:receive, from a network device, a request for availability information of a model input for a model; andtransmit, to the network device, the availability information of the model input comprising at least one of the following:an indication of whether measurement results in the model input are available,identity information associated with a set of undetectable or unreliable measurement results in the model input, orthe measurement results in the model input.10.A terminal device, comprising:a processor configured to cause the terminal device to:receive, from a network device, a list of model configurations for a model, a model configuration in the list of model configurations comprising information of a model input and information of a model output for the model and an applicable condition associated with the model input and the model output; andin accordance with a determination that the applicable condition is fulfilled, perform a model inference based on the model input and the model output.11.The terminal device of claim 10, wherein the applicable condition is associated with at least one of the following:signal strength of the terminal device,a moving speed of the terminal device,a moving direction of the terminal device,a location of the terminal device,a timing advance (TA) list,a radio access network-based notification area (RNA) list,a cell list, ora cell identity (ID) .12.The terminal device of claim 10, wherein the model configuration is associated with a measurement identity (ID) , the information of the model input is included in a measurement object associated with the measurement ID, and the information of the model output is included in a report configuration associated with the measurement ID.13.The terminal device of claim 10 or 12, wherein the information of the model input or the information of the model output comprises one of the following:a set of frequencies,a set of cell identities, ora set of beam identities.14.The terminal device of claim 10, wherein the terminal device is further caused to:transmit, to the network device, second information comprising at least one of the following:a preference on the applicable condition for the model configuration in the list of model configuration;a preference on the model input or output for a serving cell;a preference on the model input or output for a cell in a cell list;a preference on maximum number of the model inputs or outputs;a preference on maximum number of applicable cells;a preference on minimum number of model inputs or outputs; ora preference on minimum number of applicable cells.15.A method of communication, comprising:determining, at a terminal device, that a problem occurs in a model inference of a first model based on at least one of the following:a model input for the first model is undetectable or is undetectable for a first time duration or for a first number of model inference cycles,the model input for the first model is unreliable or is unreliable for a second time duration or for a second number of model inference cycles, ora requirement on a processing time of the model inference is unfulfilled; andperforming a model management comprising at least one of the following:switching to a non-artificial intelligence (AI) scheme,switching from a first configuration of the first model to a second configuration of the first model,switching from the first model to a second model, orskipping a set of undetectable or unreliable measurement results in the model input or skipping the set of undetectable or unreliable measurement results for a third time duration.16.A method of communication, comprising:receiving, at a terminal device and from a network device, a request for availability information of a model input for a model; andtransmitting, to the network device, the availability information of the model input comprising at least one of the following:an indication of whether measurement results in the model input are available,identity information associated with a set of undetectable or unreliable measurement results in the model input, orthe measurement results in the model input.17.A method of communication, comprising:receiving, at a terminal device and from a network device, a list of model configurations for a model, a model configuration in the list of model configurations comprising information of a model input and information of a model output for the model and an applicable condition associated with the model input and the model output; andin accordance with a determination that the applicable condition is fulfilled, performing a model inference based on the model input and the model output.
Citation Information
Patent Citations
Conditional artificial intelligence, machine learning model, and parameter set configurations
US20230412470A1
Devices and methods for operating machine learning model performance evaluation
WO2023110108A1
Incorporating conditions into data-collection & ai / ML operations
WO2023151989A1
Deployment and update of machine learning models in a radio access network
WO2024028370A1