Communication method and communication device

Abstract

The embodiment of the invention provides a communication method and a communication device. The method comprises: when a first difference value is less than or equal to a first threshold value, sending first data to a network device, the first difference value being a difference value between the first data and a first anchor point, the first anchor point comprising one or more reference data, the first threshold value being a predefined or configured threshold value corresponding to the first anchor point; and performing communication according to the first data. In the method, the user equipment UE can determine the data quality and report the high-quality data to the base station BS, so that rapid and accurate training is realized.

Description

[0001] This application is related to and claims priority from U.S. Provisional Patent Application Serial No. 63 / 507,848, filed on June 13, 2023, entitled “AI MODEL TRAINING DATA SET PREPARATION AND CLEANING.”

[0002] The entire disclosure of the above application is incorporated herein by reference. TECHNICAL FIELD

[0003] Embodiments of the present application relate to the field of communications, and more particularly to a communication method and a communication apparatus. BACKGROUND

[0004] AI-based algorithms are introduced in modern wireless communications to solve some wireless problems, such as channel estimation, scheduling, channel state information (CSI) compression (from user equipment to base station), multiple-in multiple-out (MIMO) beamforming, positioning, etc. As a data-driven method, AI-based algorithms inevitably have the problem of low generalization ability: if the test data sample is an outlier of the training data set, the neural network will not be able to make accurate inference on the test data sample. Therefore, artificial intelligence (AI) models need a large amount of high-quality data to train their models.

[0005] For the current data collection scheme, the data of user equipment (UE) is collected by the base station (BS) through wireless communication. The quality and importance of the UE data are not evaluated on the UE side, and the UE reports data when available. This can cause various problems, such as a large amount of air interface overhead for UE data reporting. Poor data quality (e.g., bad data) can cause AI models to be biased, inaccurate, and unreliable, resulting in greater training latency.

[0006] Therefore, how to improve the quality of UE reported data and reduce air interface overhead is a technical problem to be solved. SUMMARY

[0007] Embodiments of the present application provide a communication method and a communication apparatus. In the technical solution of the present application, the UE can determine the data quality and report high-quality data to the BS, thereby realizing fast and accurate training.

[0008] According to a first aspect, embodiments of this application provide a communication method, comprising: sending first data to a network device when a first difference is less than or equal to a first threshold, wherein the first difference is the difference between the first data and a first anchor point, the first anchor point includes one or more reference data, and the first threshold is a predefined or configured threshold corresponding to the first anchor point; and communicating based on the first data.

[0009] In the communication method provided in this application, by configuring the anchor point, the UE can determine the data quality and report high-quality data to the BS, thereby achieving fast and accurate training.

[0010] The first data includes monitoring or measured data from user equipment. Furthermore, the first data is monitoring or measured data related to the AI ​​model. In this embodiment, the network device can be a base station (BS).

[0011] The first anchor point can be one of N anchor points configured by the BS for the UE, where N≥1. The configuration signal can be radio resource control (RRC), medium access control-control element (MAC-CE), or downlink control information (DCI), and can be broadcast, multicast, or unicast.

[0012] The BS (Browser Controller) configures thresholds associated with anchor points for reporting. In some application scenarios, the BS configures the threshold for each anchor point individually, or predefines thresholds for anchor points. In other application scenarios, the protocol predefines thresholds for multiple anchor points. For example, the protocol predefines a single threshold for all anchor points. Thresholds can be reconfigured over time.

[0013] In some possible application scenarios, the N anchor points, including the first anchor point, can also be configured by the BS through UE-specific signaling or group common signaling.

[0014] In a possible implementation, the first difference is the minimum or average of the K second differences, the first anchor point includes K reference data, the j-th second difference among the K second differences is the difference between the first data and the j-th reference data among the K reference data, K≥1 and 1≤j≤K.

[0015] The first anchor point is a set of reference data, such as reference coefficients ( Reference data () The set can be a vector, such as a one-dimensional array, where the size of the vector is r, and r is predefined or configured. The size of the set is K. , j = 1, 2, … K), where K is predefined or configured. The first anchor point includes K reference coefficients.

[0016] The UE calculates the difference between the first data (e.g., coefficients (c )) and the reference data c j (j = 1, 2, … K) in the first anchor point. The difference can be calculated by the following equation: 、 and . is the difference between the first data and the reference data c in the first anchor point. is the first data. is the jth reference data in the first anchor point. represents an inner product. represents a norm, which is a method of measuring the size of a vector, matrix, tensor, or function. f represents other custom functions. 1≤j≤K and 1≤i≤r. The difference between the first data and the reference data can also be calculated by a dot product, a Euclidean distance, or a DNN-based algorithm, etc. The specific calculation should not be understood as a limitation of the present application.

[0017] The first difference can be calculated by any of the following equations: and . is the difference between the first data and the first anchor point. may be the minimum value of the difference between the first data and the K reference data in the first anchor point. may be the average value of the difference between the first data and the K reference data in the first anchor point.

[0018] Alternatively, the difference between the first data and the first anchor point can also be obtained by mutual information, Hilbert-Schmidt independence criterion (HSIC) metric, Kullback-Leibler (KL) scattering, graph edit distance, Wasserstein distance, Jensen-Shannon divergence (JSD) distance, DNN-based algorithm, etc.

[0019] In the communication method provided by the present application, through the configuration of the anchor point, the UE can determine the data quality and report the high-quality data to the BS, so as to realize fast and accurate training.

[0020] In a possible implementation, the method further includes: sending, to the network device, an index of a second anchor point, the second anchor point being one of the N anchor points, wherein a third difference value corresponding to the second anchor point is a minimum third difference value in N third difference values, an nth third difference value in the N third difference values being a difference value between the first data and an nth anchor point, N≥1 and 1≤n≤N.

[0021] For a UE, one or more anchor points (the number of anchor points being N) are configured, where an anchor point is a set of reference data. The UE calculates a difference between the first data and the anchor point n (n=1 to N) Then the index k of the nearest anchor point is found. The data with the nearest anchor point index k is referred to as the data associated with the anchor point k in the embodiments of the present application.

[0022] The nearest anchor point of the first data is the second anchor point, so the first data is the data associated with the second anchor point. The UE can report its associated anchor point index to the BS. Then, if the BS finds that the data associated with the anchor point k is needed, the BS instructs the UE to report the data associated with the anchor point k and also indicates the resource for feedback.

[0023] In the communication method provided in the present application, the BS can instruct the UE to report data according to the index of the associated anchor point reported by the UE, and can collect more corresponding data to train or fine-tune its model.

[0024] In a possible implementation, each of the N anchor points corresponds to a priority, and sending the first data to the network device includes: sending the first data in a priority order of a second anchor point in the N anchor points.

[0025] The BS can configure a priority for each anchor point. The UE sends the first data in a priority order of the nearest anchor point of the first data. For example, the BS observes that there are enough data samples associated with anchor point-1 (for example, data collected in an outdoor environment), but much fewer data samples associated with anchor point-2 (for example, data collected in an indoor environment). The BS configures a higher priority for anchor point-2 to collect more corresponding data to train or fine-tune its model.

[0026] In the communication method provided in the present application, the BS can collect different data by setting the priority of the anchor point to train its model, thereby ensuring the generalization performance of the model.

[0027] In a possible implementation, the index value of the anchor point is a priority corresponding to the anchor point.

[0028] The anchor index value can indicate the priority of the anchor. For example, the smaller the index value, the higher the priority. A higher priority of an anchor means that the first data associated with the anchor should have a higher priority, for example, be reported first.

[0029] In the communication method provided in the present application, the BS can collect different data to train its model by setting the priority of the anchor, thereby ensuring the generalization performance of the model.

[0030] In a possible implementation, the N anchors are configured by radio resource control (RRC), medium access control-control element (MAC-CE), or downlink control information (DCI) signals from the network device.

[0031] In a possible implementation, the first data is sent through a transport block at the medium access control (MAC) layer or the physical (PHY) layer of the user equipment.

[0032] For data with a higher priority, the priority is higher in the medium access control (MAC) or physical (PHY) layer multiplexing. Multiple data can be multiplexed in a transport block (TB) in the MAC layer or the PHY layer, and the highest priority is the first to be included in the TB.

[0033] In a possible implementation, the first data includes monitoring data or measured data of the user equipment.

[0034] In a possible implementation, the first data includes any one or more of the following: perception data, measured data, channel data, neuron data of an artificial intelligence (AI) model, and potential output data of the AI model.

[0035] In a possible implementation, the first data is a coefficient of a predefined or configured orthogonal basis The first anchor includes K reference coefficients c k of the orthogonal basis, k = 1, 2, …, K, 1 ≤ k ≤ K.

[0036] According to a second aspect, the present application provides a communication apparatus comprising: a sending module configured to send first data to a network device when a first difference is less than or equal to a first threshold, the first difference being a difference between the first data and a first anchor point, the first anchor point comprising one or more reference data, the first threshold being a predefined or configured threshold corresponding to the first anchor point; and a processing module configured to communicate according to the first data.

[0037] In a possible implementation, the first difference is a minimum of K second differences or an average of the K second differences, the first anchor point comprises K reference data, a jth second difference of the K second differences is a difference between the first data and a jth reference data of the K reference data, K≥1 and 1≤j≤K.

[0038] In a possible implementation, the sending module is further configured to send an index of a second anchor point to the network device, the second anchor point being one of N anchor points, wherein a third difference corresponding to the second anchor point is a minimum of N third differences, a nth third difference of the N third differences being a difference between the first data and an nth anchor point, N≥1 and 1≤n≤N.

[0039] In a possible implementation, any of the N anchor points corresponds to a priority, and the sending module is further configured to send the first data according to an order of the priority of the second anchor point of the N anchor points.

[0040] In a possible implementation, the index value of the anchor point is a priority corresponding to the anchor point.

[0041] In a possible implementation, the N anchor points are configured by a radio resource control (RRC), a medium access control-control element (MAC-CE) or a downlink control information (DCI) signal from the network device.

[0042] In a possible implementation, the first data is sent through a transport block at a medium access control (MAC) layer or a physical (PHY) layer of the user equipment.

[0043] In a possible implementation, the first data comprises monitoring data or measured data of the user equipment.

[0044] In a possible implementation, the first data includes any one or more of perception data, measured data, channel data, neuron data of an artificial intelligence (AI) model, and potential output data of the AI model.

[0045] In a possible implementation, the first data is a coefficient of a predefined or configured orthogonal basis , the first anchor point includes K reference coefficients c k , k = 1, 2, …, K, and 1 ≤ k ≤ K.

[0046] According to a third aspect, a communication apparatus is provided, including a processor and a memory. The processor is connected to the memory. The memory is configured to store instructions, and the processor is configured to execute the instructions. When the processor executes the instructions stored in the memory, the processor is caused to perform the method in any possible implementation of the first aspect.

[0047] According to a fourth aspect, a communication system is provided, including the communication apparatus in any possible implementation of the second aspect, and a network device.

[0048] According to a fifth aspect, a computer-readable storage medium is provided, including instructions. When the instructions are run on a processor, the processor is caused to perform the method in any possible implementation of the first aspect.

[0049] According to a sixth aspect, a computer program product is provided, including computer program code. When the computer program code is run on a computer, the computer is caused to perform the method in any possible implementation of the first aspect.

[0050] It should be noted that all or part of the above computer program code can be stored in a first storage medium. The first storage medium can be packaged with the processor or packaged separately from the processor.

[0051] According to a seventh aspect, a chip system is provided, including a memory and a processor. The memory is configured to store a computer program, and the processor is configured to call and run the computer program from the memory, so that an electronic device in which the chip system is disposed performs the method in any possible implementation of the first aspect. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 is a schematic diagram of a communication system according to an embodiment of the present application.

[0053] Figure 2 is a schematic diagram of a communication system 100 according to an embodiment of the present application.

[0054] Figure 3is a schematic diagram of an ED 110 and base stations 170a, 170b, and / or 170c, which are embodiments of the application.

[0055] Figure 4 is a schematic diagram of units or modules in a device, which is an embodiment of the application.

[0056] Figure 5 is a schematic diagram of an AI-based communication device.

[0057] Figure 6 is a schematic diagram of a device 500 receiving reference data samples from a device 600, which is an embodiment of the application.

[0058] Figure 7 is a schematic diagram of reference data samples consisting of multiple groups, which is an embodiment of the application.

[0059] Figure 8 is a schematic diagram of DNN-based approximation, which is an embodiment of the application.

[0060] Figure 9 is a flowchart of a communication method, which is an embodiment of the application.

[0061] Figure 10 is a flowchart of a communication method, which is an embodiment of the application.

[0062] Figure 11 is a schematic diagram of projecting a high-dimensional signal to a low-dimensional signal, which is an embodiment of the application.

[0063] Figure 12 is a flowchart of a communication method, which is an embodiment of the application.

[0064] Figure 13 is a schematic diagram of a matrix U determined by an embodiment of the application.

[0065] Figure 14 is a schematic diagram of a first sampling matrix P1, which is an embodiment of the application.

[0066] Figure 15 is a schematic diagram of a sampling matrix compressing a matrix U, which is an embodiment of the application.

[0067] Figure 16 is a schematic diagram of a scoring distance on a low-frequency spectral space, which is an embodiment of the application.

[0068] Figure 17 is a flowchart of an embodiment of a communication method, which is an embodiment of the application.

[0069] Figure 18 is a flowchart of an embodiment of a communication method, which is an embodiment of the application.

[0070] Figure 19 is a schematic diagram of UE data reporting when one anchor point is configured, which is an embodiment of the present application.

[0071] Figure 20 is a schematic diagram of UE data reporting when multiple anchor points are configured, which is an embodiment of the present application.

[0072] Figure 21 is a flowchart of an embodiment of a communication method, which is an embodiment of the present application.

[0073] Figure 22 is a schematic diagram of UE data reporting according to a priority index, which is an embodiment of the present application.

[0074] Figure 23 is a schematic block diagram of a communication apparatus, which is an embodiment of the present application.

[0075] Figure 24 is a schematic block diagram of another communication apparatus, which is an embodiment of the present application. DETAILED DESCRIPTION

[0076] The technical solutions in the present application will be described below with reference to the drawings.

[0077] The technical solutions in the present application will be described below with reference to the drawings. Obviously, the described embodiments are only some, but not all, of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present application.

[0078] The present application will propose aspects, embodiments or features around systems including multiple devices, components, modules, etc. It should be understood and recognized that each system can include additional devices, components, modules, etc., and / or can not include all the devices, components, modules, etc. discussed in connection with the drawings. In addition, combinations of these options can also be used.

[0079] In addition, in the embodiments of the present application, the word "exemplary" and the phrase "as an example" are used to indicate illustration or description. Any embodiment or design scheme described as "exemplary" in the present application should not be interpreted as superior to or more advantageous than other embodiments or design schemes. Rather, the word "example" is used to present this concept in a specific way.

[0080] Unless specifically emphasized, the phrases "in some embodiments", "in some possible implementations", and the like, do not necessarily refer to the same embodiments, although they can. The terms "including", "containing", "having" and variations thereof, mean "including but not limited to".

[0081] In this application, "at least one" means one or more, "multiple" means two or more. "And / or" describes the association of associated objects, indicating that there can be three relationships. For example, A and / or B can mean only A, both A and B, only B, where A and B can be singular or plural. The character " / " generally indicates that the front and rear associated objects are "or" relationship.

[0082] The application scenarios described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art can know that, with the evolution of system architecture and the appearance of new application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0083] The technical solutions of the embodiments of the present application can be applied in various communication systems, for example, a Global System for Mobile Communications (GSM), a Code Division Multiple Access (CDMA) system, a Wideband Code Division Multiple Access (WCDMA) system, a general packet radio service (GPRS) system, a Long Term Evolution (LTE) system, an LTE frequency division duplex (FDD) system, an LTE time division duplex (TDD) system, a Universal Mobile Telecommunications System (UMTS), a Worldwide Interoperability for Microwave Access (WiMAX) communication system, a wireless local area network (WLAN), a fifth generation (5G) wireless communication system, a new radio (NR) wireless communication system, a sixth generation (6G) wireless communication system, or other evolved communication systems.

[0084] In order to better describe the solutions of the embodiments of the present application, the concepts and terms that may be involved in the present application will be described below.

[0085] (1) Data collection

[0086] Data is a very important component of artificial intelligence (AI) / machine learning (ML) technology. Data collection refers to the process of collecting data by a network node, a management entity, or a UE for AI / ML model training, data analysis, and inference.

[0087] (2) AI / ML model training AI / ML model training refers to the process of training an AI / ML model by learning the input / output relationship in a data-driven manner and obtaining a trained AI / ML model for inference.

[0088] (3) AI / ML model inference A process that uses a trained AI / ML model to produce a set of outputs from a set of inputs.

[0089] (4) AI / ML model validation As a sub-process of training, validation is used to evaluate the quality of an AI / ML model using a different dataset than the one used for model training. Validation can help select model parameters that achieve generalization beyond the dataset used for model training. The trained model parameters can be further adjusted through the validation process.

[0090] (5) AI / ML model testing Similar to validation, testing is also a sub-process of training, which is used to evaluate the performance of the final AI / ML model using a different dataset than the one used for model training and validation. Unlike AI / ML model validation, testing does not assume subsequent adjustments to the model.

[0091] (6) Online training Online training refers to an AI / ML training process in which the model used for inference is continuously trained (in near) real-time as new training samples arrive.

[0092] (7) Offline training Offline training is an AI / ML training process in which the model is trained on a collected dataset, where the trained model is subsequently used or transferred for inference.

[0093] (8) AI / ML model transfer / migration AI / ML model transfer / migration is a general term referring to the transfer of an AI / ML model from one entity to another in any manner. The transfer of an AI / ML model over the air includes parameters of a model structure known to the receiving end, as well as a new model with parameters. The transfer can include a complete model or a partial model.

[0094] (9) Life cycle management (LCM) When training and / or inferring AI / ML models on a device, the whole AI / ML process needs to be monitored and managed to guarantee the performance gain obtained by AI / ML techniques. For example, due to the randomness of wireless channels and the mobility of UEs, the propagation environment of wireless signals can change frequently. However, it is difficult for AI / ML models to maintain optimal performance in all scenarios, and even in some scenarios, the performance can drop sharply. Therefore, the life cycle management (LCM) of AI / ML models is crucial for the sustainable operation of AI / ML in the NR air interface. Life cycle management covers the whole process of applying AI / ML techniques on one or more nodes. Specifically, life cycle management includes at least one of the following sub-processes: data collection, model training, model identification, model registration, model deployment, model configuration, model inference, model selection, model activation, deactivation, model switching, model rollback, model monitoring, model updating, model transmission / delivery, and UE capability reporting. Model monitoring can be based on inference accuracy, including metrics related to intermediate key performance indicators (KPIs), or based on system performance, including metrics related to system performance KPIs, such as accuracy and relevance, overhead, complexity (computational and memory cost), latency (timeliness of monitoring results, from model failure to recovery), and power consumption. In addition, due to environmental changes after deployment, data distribution may change, so models based on input or output data distribution should also be considered.

[0095] (10) Supervised learning The goal of a supervised learning algorithm is to train a model that maps feature vectors (inputs) to labels (outputs) from training data that includes example feature-label pairs. Supervised learning can analyze training data and produce an inference function that can be used to map inference data. Supervised learning can be further divided into two types: classification and regression. Classification is used when the output of the AI / ML model is categorical data (i.e., with two or more categories). Regression is used when the output of the AI / ML model is a real number or continuous value.

[0096] (11) Unsupervised learning Unlike supervised learning, where AI / ML models learn to map inputs to target outputs, unsupervised methods learn a compact representation of input data without labeled data, which can be used for data exploration or analysis or to generate new data. One typical unsupervised learning is clustering, which explores the hidden structure of input data and provides a classification result for the data.

[0097] (12) Reinforcement learning Reinforcement learning is used to solve sequential decision-making problems. Reinforcement learning is a process of training an agent’s actions from inputs (states) and feedback signals (rewards) in an environment. In reinforcement learning, an agent interacts with an environment by performing actions to maximize cumulative rewards. Each time an agent performs an action, the current state in the environment can transition to a new state, and the new state resulting from the action will bring an associated reward. Subsequently, the agent can perform the next action based on the received reward and the new state in the environment. During the training phase, the agent interacts with the environment to accumulate experience. Since the cost of directly interacting with a real system is high, the environment is usually simulated by a simulator. During the inference phase, the agent can use the optimal decision rule learned from the training phase to achieve the maximum cumulative reward.

[0098] (13) FEDERATED LEARNING Federal learning (FL) is a machine learning technique for training an AI / ML model by a central node (e.g., a server) and multiple decentralized edge nodes (e.g., UEs, next generation NodeBs, “gNBs”). According to the wireless FL technique, the server can provide a set of model parameters (e.g., weights, biases, gradients) describing a global AI / ML model to the edge nodes. The edge nodes can initialize a local AI / ML model using the received global AI / ML model parameters. Then, the edge nodes can train the local AI / ML model using local data samples, resulting in trained local AI / ML models. Subsequently, the edge nodes can provide a set of AI / ML model parameters describing the local AI / ML models to the server. Upon receiving multiple sets of AI / ML model parameters describing respective local AI / ML models at multiple edge nodes from the multiple edge nodes, the server can aggregate the local AI / ML model parameters reported from the multiple UEs and update the global AI / ML model according to the aggregation. The progress of subsequent iterations is very similar to the first iteration. The server can send the aggregated global model to the multiple edge nodes. The above process can be iterated multiple times until the global AI / ML model is finalized, e.g., the AI / ML model converges or a training stopping condition is satisfied. Notably, the wireless FL technique does not involve the exchange of local data samples. In fact, the local data samples remain at the respective edge nodes.

[0099] AI-based algorithms have been introduced into modern wireless communications to solve some wireless problems, such as channel estimation, scheduling, channel state information (CSI) compression (from user equipment to base station), multiple-in multiple-out (MIMO) beamforming, positioning, etc. AI algorithms are a data-driven method that adjusts some predefined architecture through a set of data samples called training dataset. Recent AI trains DNN (including CNN, RNN, Transformer, etc.) architecture by setting neurons using the SGD algorithm.

[0100] AI technology (including ML technology) in communications includes AI-based communication in the physical layer and / or AI-based communication in the MAC layer. For the physical layer, AI communication can aim to optimize component design and / or improve algorithm performance. For the MAC layer, AI / ML-based communication can aim to utilize AI / ML capabilities to learn, predict, and / or make decisions to solve complex optimization problems using possibly better strategies and / or optimal solutions, e.g., to optimize functions in the MAC layer, such as intelligent TRP management, intelligent beam management, intelligent channel resource allocation, intelligent power control, intelligent spectrum utilization, intelligent modulation and coding scheme (MCS), intelligent hybrid automatic repeat request (HARQ) strategy, and / or intelligent transmit / receive (Tx / Rx) mode adaptation, etc.

[0101] An AI architecture can include multiple nodes, which can be organized in two modes, centralized and distributed, both of which can be deployed in an access network, core network, or edge computing system or third-party network. Centralized training and computing architecture is limited by the large amount of communication overhead that can exist and strict user data privacy. Distributed training and computing architecture can include several frameworks, such as distributed machine learning and federated learning. In some embodiments, the AI architecture can include an intelligent controller that can be executed as a single or multiple intelligent agents according to joint or separate optimization. New protocols and signaling mechanisms are needed so that the corresponding interface links can be personalized by custom parameters to meet specific needs, while the AI technology personalized by the custom parameters maximizes the reduction of signaling overhead and maximizes the spectral efficiency of the entire system.

[0102] New protocols and signaling mechanisms are provided for operating within and switching between different modes of operation, including between AI and non-AI modes, but also for measurements and feedback to accommodate different possible measurements and information that can require feedback, depending on the implementation.

[0103] It is now quite common for neural network models to become larger and deeper, which can easily require more computational resources than just one or two computers. Most neural network models will be trained on a powerful computing cloud. A user with a desired neural network architecture, raw training dataset, and training objectives can not have enough local computational resources to train their model locally. To access the powerful computing cloud, the user must send all specifications of their neural network architecture, training dataset, and training objectives to the network cloud completely. It is required that the user must trust the cloud and fully authorize the cloud to manipulate their intellectual property (neural network architecture, training dataset, and training objectives).

[0104] As a data-driven approach, AI-based algorithms inevitably have the problem of low generalization ability: if the test data sample is an outlier of the training dataset, the neural network will not be able to make accurate inferences on the test data sample. Even if the AI model is trained on a large dataset, it can not have the knowledge required to run effectively in other environments, especially in wireless communications where channel information changes rapidly.

[0105] In this application, the AI model is exemplified by a DNN, i.e. a deep neural network or network. The specific AI model should not be understood as limiting the application.

[0106] Figure 1 is a schematic diagram of a communication system, which is an embodiment of the present application.

[0107] Reference is made to Figure 1As a non-limiting illustrative example, a simplified schematic diagram of a communication system is provided. The communication system 100 includes a wireless access network 120. The wireless access network 120 can be a next generation (e.g., sixth generation (6G) or beyond) wireless access network, or a legacy (e.g., 5G, 4G, 3G, or 2G) wireless access network. One or more communication electric devices (EDs) 110a-110j (generally referred to as 110) can be interconnected to each other or to one or more network nodes (170a, 170b, generally referred to as 170) in the wireless access network 120. A core network 130 can be part of the communication system, and can be dependent or independent of the radio access technology used in the communication system 100. In addition, the communication system 100 includes a public switched telephone network (PSTN) 140, the Internet 150, and other networks 160.

[0108] Figure 2 Figure 1 is a schematic diagram of a communication system 100 according to an embodiment of the present application.

[0109] Figure 2 An example communication system 100 is shown. Generally, the communication system 100 is capable of transmitting data and other content among a plurality of wireless or wireline elements. The communication system 100 can have the purpose of providing voice, data, video, and / or text content, among other content, through broadcast, multicast, and unicast, among other techniques. The communication system 100 can operate by sharing resources (e.g., carrier spectrum bandwidth) among its constituent elements. The communication system 100 can include a terrestrial communication system and / or a non-terrestrial communication system. The communication system 100 can provide a wide range of communication services and applications (e.g., earth monitoring, remote sensing, passive sensing and positioning, navigation and tracking, automated delivery and mobility, among others). The communication system 100 can provide a high level of availability and robustness through the joint operation of a terrestrial communication system and a non-terrestrial communication system. For example, the integration of a non-terrestrial communication system (or components thereof) into a terrestrial communication system can enable a heterogeneous network comprising multiple tiers. The heterogeneous network can achieve better overall performance compared to a legacy communication network through efficient multi-link joint operation, more flexible function sharing, and faster physical layer link switching between the terrestrial network and the non-terrestrial network.

[0110] The ground communication system and the non-terrestrial communication system can be considered as subsystems of a communication system. In the illustrated example, the communication system 100 includes electronic devices (EDs) 110a-110d (generally referred to as EDs 110), radio access networks (RANs) 120a-120b, a non-terrestrial communication network 120c, a core network 130, a public switched telephone network (PSTN) 140, the Internet 150, and other networks 160. The RANs 120a-120b include respective base stations (BSs) 170a-170b, which can be generally referred to as terrestrial transmit and receive points (T-TRPs) 170a-170b. The non-terrestrial communication network 120c includes an access node 120c, which can be generally referred to as a non-terrestrial transmit and receive point (NT-TRP) 172.

[0111] Alternatively or additionally, any of the EDs 110 can be configured to connect with, access, or communicate with any other T-TRP 170a-170b, NT-TRP 172, Internet 150, core network 130, PSTN 140, other networks 160, or any combination of the above. In some examples, the ED 110a can communicate uplink and / or downlink transmissions with the T-TRP 170a via an interface 190a. In some examples, the ED 110a, ED 110b, and ED 110d can also communicate directly with each other via one or more sidelink air interfaces 190b. In some examples, the ED 110d can communicate uplink and / or downlink transmissions with the NT-TRP 172 via an interface 190c.

[0112] The air interfaces 190a and 190b can use similar communication techniques, e.g., any applicable wireless access technique. For example, the communication system 100 can implement one or more channel access methods in the air interfaces 190a and 190b, e.g., code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), or single-carrier FDMA (SC-FDMA). The air interfaces 190a and 190b can utilize other higher-dimensional signal spaces that can involve combinations of orthogonal and non-orthogonal dimensions. The air interface 190c can enable communication between the ED 110d and one or more NT-TRPs 172 through a wireless link or simply a link. In some examples, the link is a dedicated connection for unicast transmission, a connection for broadcast transmission, or a connection between a group of EDs and one or more NT-TRPs for groupcast transmission.

[0113] The RANs 120a and 120b are in communication with the core network 130 to provide the EDs 110a, 110b, and 110c with access to various services, such as voice, data, and other services. The RANs 120a and 120b and / or the core network 130 can be in direct or indirect communication with one or more other RANs (not shown) that can or can not be of the same type as the RANs 120a, 120b, or both, and that can or can not be in direct communication with the core network 130. The core network 130 can also serve as a gateway for the RANs 120a and 120b, or the EDs 110a, 110b, and 110c, or both, to other networks (for example, the PSTN 140, the Internet 150, and the other networks 160) by providing an interface with the other networks. Further, some or all of the EDs 110a, 110b, and 110c can include functionality for communicating with different wireless networks over different wireless links using different wireless technologies and / or protocols. The EDs 110a, 110b, and 110c can communicate with service providers or switches (not shown) and the Internet 150 through wired communication channels instead of or in addition to wireless communication. The PSTN 140 can include a circuit- switched telephone network for providing plain old telephone service (POTS). The Internet 150 can include a network of computers and subnetworks (intranets) or both, and can incorporate protocols such as Internet Protocol (IP), transmission control protocol (TCP), and user datagram protocol (UDP). The EDs 110a, 110b, and 110c can be multi-mode devices capable of operating according to multiple wireless access technologies and include multiple transceivers needed to support these technologies.

[0114] Figure 3 is a schematic diagram of an ED 110 and base stations 170a, 170b, and / or 170c, which are embodiments of the present application.

[0115] Figure 3Another example of an ED 110 and base stations 170a, 170b, and / or 170c is shown. The ED 110 is used to connect people, objects, machines, etc. The ED 110 can be widely used in various scenarios, such as cellular communication, device-to-device (D2D), vehicle to everything (V2X), peer-to-peer (P2P), machine-to-machine (M2M), machine-type communication (MTC), internet of things (IoT), virtual reality (VR), augmented reality (AR), industrial control, autonomous driving, telemedicine, smart grid, smart home, smart office, smart wearable, smart transportation, smart city, drone, robot, remote sensing, passive sensing, positioning, navigation and tracking, automatic distribution, mobility, etc.

[0116] Each ED 110 represents any suitable end user device for wireless operation, which can include (or can be referred to as) a user equipment (UE / user device), a wireless transmit / receive unit (WTRU), a mobile station, a fixed or mobile subscriber unit, a cellular phone, a station (STA), a machine type communication (MTC) device, a personal digital assistant (PDA), a smartphone, a notebook, a computer, a tablet, a wireless sensor, a consumer electronics, a smart book, a vehicle, a car, a truck, a bus, a train, or an IoT device, an industrial device or apparatus in the above devices (e.g., a communication module, a modem, or a chip), etc. Future generation ED 110 can be referred to using other terms. The base stations 170a and 170b are T-TRPs, which will be referred to as T-TRPs 170 hereinafter. Likewise, the NT-TRPs are shown in Figure 3

[0117] ​ED 110 includes a transmitter 201 and a receiver 203 coupled to one or more antennas 204. Only one antenna 204 is shown in the figure. One, part, or all of the antennas may also be panels. The transmitter 201 and receiver 203 may, for example, be integrated as a transceiver. The transceiver is used to modulate data or other content for transmission through at least one antenna 204 or a network interface controller (NIC). The transceiver is also used to demodulate data or other content received through at least one antenna 204. Each transceiver includes any suitable structure for generating signals for wireless or wired transmission and / or for processing signals received wirelessly or wiredly. Each antenna 204 includes any suitable structure for transmitting and / or receiving wireless or wired signals.

[0118] ED 110 includes at least one memory 208. Memory 208 stores instructions and data used, generated, or acquired by ED 110. For example, memory 208 may store software instructions or modules for implementing some or all of the functions and / or embodiments described herein, and executed by one or more processing units 210. Each memory 208 includes any suitable one or more volatile and / or non-volatile storage and retrieval devices. Any suitable type of memory can be used, such as random access memory (RAM), read-only memory (ROM), hard disk, optical disk, subscriber identity module (SIM) card, memory stick, secure digital (SD) memory card, and processor cache, etc.

[0119] ED 110 may also include one or more input / output devices (not shown) or interfaces (e.g., connected to...). Figure 1 (Wired interface of Internet 150 in the network). One or more input / output devices support interaction with users or other devices in the network. Each input / output device includes any suitable structure for providing or receiving information from the user, such as a speaker, microphone, keypad, keyboard, display, or touchscreen, including network interface communication.

[0120] The ED 110 also includes a processor 210 for performing various operations, including operations related to preparing transmissions for uplink transmissions to the NT-TRPs 172 and / or the T-TRPs 170, operations related to processing downlink transmissions received from the NT-TRPs 172 and / or the T-TRPs 170, and operations related to processing sidelink transmissions to and from another ED 110. Processing operations related to preparing transmissions for uplink transmissions can include operations such as encoding, modulation, transmit beamforming, and generating symbols for transmission. Processing operations related to processing downlink transmissions can include operations such as receive beamforming, demodulation, and decoding received symbols. According to embodiments, the receiver 203 can receive downlink transmissions (possibly using receive beamforming), and the processor 210 can extract signaling (e.g., by detecting and / or decoding the signaling) from the downlink transmissions. An example of the signaling can be reference signals transmitted by the NT-TRPs 172 and / or the T-TRPs 170. In some embodiments, the processor 276 implements transmit beamforming and / or receive beamforming based on indications of beam directions (e.g., beam angle information (BAI)) received from the T-TRPs 170. In some embodiments, the processor 210 can perform operations related to network access (e.g., initial access) and / or downlink synchronization, such as operations related to detecting synchronization sequences, decoding and acquiring system information, etc. In some embodiments, the processor 210 can perform channel estimation, e.g., using reference signals received from the NT-TRPs 172 and / or the T-TRPs 170.

[0121] Although not shown, the processor 210 can form part of the transmitter 201 and / or the receiver 203. Although not shown, the memory 208 can be part of the processor 210.

[0122] The processor 210, and the processing components of the transmitter 201 and the receiver 203, respectively, can be implemented by the same or different one or more processors for executing instructions that are stored in a memory (e.g., the memory 208). Alternatively, part or all of the processor 210, and the processing components of the transmitter 201 and the receiver 203, can be implemented using a specially-programmed field-programmable gate array (FPGA), a graphical processing unit (GPU), or an application-specific integrated circuit (ASIC), etc.

[0123] T-TRP 170 can be referred to using other names in some implementations, e.g., base station, base transceiver station (BTS), radio base station, network node, network equipment, network-side device, transmission / reception node, Node B, evolved Node B (eNodeB or eNB), home eNodeB, next generation Node B (gNB), transmission point (TP), site controller, access point (AP) or wireless router, relay, remote radio head, ground node, ground network device, or ground base station, base band unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), positioning node, etc. T-TRP 170 can be a macro BS, a pico BS, a relay node, a donor node, etc., or a combination thereof. T-TRP 170 can refer to the above devices, or to an apparatus (e.g., a communication module, modem, or chip) in the above devices.

[0124] In some embodiments, various parts of T-TRP 170 can be distributed. For example, some modules of T-TRP 170 can be located at a location remote from a device that houses an antenna of T-TRP 170, and can be coupled to the device that houses the antenna through a communication link (not shown), sometimes referred to as front-haul, e.g., common public radio interface (CPRI). Thus, in some embodiments, the term “T-TRP 170” can also refer to modules that perform ED 110 position determination, resource allocation (scheduling), message generation and encoding / decoding, etc. processing operations on the network side, which are not necessarily part of the device that houses the antenna of T-TRP 170. These modules can also be coupled to other T-TRPs. In some embodiments, T-TRP 170 can actually be multiple T-TRPs that work together, e.g., through coordinated multipoint transmission, to serve ED 110.

[0125] The T-TRP 170 includes at least one transmitter 252 and at least one receiver 254 coupled to one or more antennas 256. Only one antenna 256 is shown in the figure. One, some or all of the antennas can also be panels. The transmitter 252 and receiver 254 can be integrated as a transceiver. The T-TRP 170 also includes a processor 260 for performing various operations, including operations related to preparing transmissions for downlink transmissions to the ED 110, processing uplink transmissions received from the ED 110, preparing transmissions for backhaul transmissions to the NT-TRP 172, and processing transmissions received from the NT-TRP 172 over the backhaul. The processing operations related to preparing transmissions for downlink or backhaul transmissions can include operations such as encoding, modulation, precoding (e.g., MIMO precoding), transmit beamforming, and generating symbols for transmission. The processing operations related to processing received transmissions in the uplink or over the backhaul can include operations such as receive beamforming, demodulation, and decoding received symbols. The processor 260 can also perform operations related to network access (e.g., initial access) and / or downlink synchronization, such as generating contents of a synchronization signal block (SSB), generating system information, etc. In some embodiments, the processor 260 also generates an indication of a beam direction, e.g., a BAI, which can be scheduled for transmission by the scheduler 253. The processor 260 performs other network-side processing operations described herein, e.g., determining a location of the ED 110, determining a location where to deploy the NT-TRP 172, etc. In some embodiments, the processor 260 can generate signaling, e.g., for configuring one or more parameters of the ED 110 and / or one or more parameters of the NT-TRP 172. Any signaling generated by the processor 260 is transmitted by the transmitter 252. Note that the “signaling” used here can also be referred to as control signaling. Dynamic signaling can be transmitted in a control channel, e.g., a physical downlink control channel (PDCCH), and static or semi-static higher layer signaling can be included in packets transmitted in a data channel, e.g., a physical downlink shared channel (PDSCH).

[0126] The scheduler 253 can be coupled to the processor 260. The scheduler 253 can be included within or operate separately from the T-TRP 170, which can schedule uplink, downlink, and / or backhaul transmissions, including issuing scheduling grants and / or configuring grant-free (“configured grant”) resources. The T-TRP 170 also includes memory 258 that stores information and data. The memory 258 stores instructions and data used, generated, or collected by the T-TRP 170. For example, the memory 258 can store software

[0127] Although not shown, the processor 260 can form part of the transmitter 252 and / or the receiver 254. Further, the processor 260 can implement the scheduler 253, although not shown. Although not shown, the memory 258 can be part of the processor 260.

[0128] The processor 260, the scheduler 253, and the processing components of the transmitter 252 and the receiver 254 can each be implemented by one or more processors executing instructions as stored in a memory (e.g., the memory 258). Alternatively, some or all of the processor 260, the scheduler 253, and the processing components of the transmitter 252 and the receiver 254 can be implemented using special-purpose circuitry, such as FPGA, GPU, or ASIC.

[0129] Although NT-TRP 172 is shown as a drone merely as an example, NT-TRP 172 can be implemented by any suitable non-ground form. Also, in some implementations, NT-TRP 172 can have other names, such as non-ground node, non-ground network device, or non-ground base station. NT-TRP 172 includes transmitter 272 and receiver 274 coupled to one or more antennas 280. Only one antenna 280 is shown in the figure. One, some or all of the antennas can also be panels. Transmitter 272 and receiver 274 can be integrated as a transceiver. NT-TRP 172 also includes processor 276 for performing various operations, including operations related to preparing transmissions for downlink transmissions to ED 110, processing uplink transmissions received from ED 110, preparing transmissions for backhaul transmissions to T-TRP 170, and processing transmissions received from T-TRP 170 over the backhaul. Processing operations related to preparing transmissions for downlink or backhaul transmissions can include operations such as encoding, modulation, precoding (e.g., MIMO precoding), transmit beamforming, and generating symbols for transmission. Processing operations related to processing received transmissions in uplink or over the backhaul can include operations such as receive beamforming, demodulation, and decoding received symbols. In some embodiments, processor 276 implements transmit beamforming and / or receive beamforming based on beam direction information (e.g., BAI) received from T-TRP 170. In some embodiments, processor 276 can generate signaling, e.g., to configure one or more parameters of ED 110. In some embodiments, NT-TRP 172 implements physical layer processing, but not higher layer functions such as functions of medium access control (MAC) or radio link control (RLC) layers. Since this is merely one example, NT-TRP 172 can generally implement higher layer functions in addition to physical layer processing.

[0130] NT-TRP 172 also includes memory 278 that stores information and data. Although not shown, processor 276 can form part of transmitter 272 and / or receiver 274. Although not shown, memory 278 can be part of processor 276.

[0131] The processing components of the processor 276, and transmitter 272 and receiver 274, respectively, can be implemented by the same or different one or more processors that execute instructions stored in a memory (e.g., memory 278). Alternatively, some or all of the processing components of the processor 276, and transmitter 272 and receiver 274, can be implemented using specially designed hardware (e.g., an ASIC, a FPGA, a GPU, or the like), or a combination of hardware and software. In some embodiments, the NT-TRP 172 can actually be multiple NT-TRPs that work together, e.g., through coordinated multipoint transmission, to serve the ED 110.

[0132] The T-TRP 170, NT-TRP 172, and / or ED 110 can include other components, but for the sake of clarity, these components have not been illustrated.

[0133] Figure 4 is a schematic diagram of an apparatus that is an embodiment of a unit or module of the present application.

[0134] According to Figure 4 , one or more steps in the embodiments methods provided can be performed by corresponding units or modules. Figure 4 Units or modules in an apparatus, e.g., in the ED 110, in the T-TRP 170, or in the NT-TRP 172, are shown. For example, a signal can be transmitted by a transmitting unit or module. For example, a signal can be transmitted by a transmitting unit or module. A signal can be received by a receiving unit or module. A signal can be processed by a processing unit or module. Other steps can be performed by an artificial intelligence (AI) or machine learning (ML) module. The corresponding units or modules can be implemented using hardware, one or more components or devices executing software, or a combination thereof. For example, one or more of these units or modules can be an integrated circuit, such as a programmed FPGA, GPU, or ASIC. It will be appreciated that if these modules are implemented by a processor using software, the processor can retrieve the modules from memory in whole or in part, individually or collectively, as necessary for processing, in one or multiple instances, and the modules themselves can include instructions for further deployment and instantiation.

[0135] Other details regarding the ED 110, T-TRP 170, and NT-TRP 172 are known to those of skill in the art. Accordingly, these details have been omitted here.

[0136] Figure 5 is a schematic diagram of an AI-based communication device.

[0137] A wireless system includes multiple connected devices. Device 500 is a base station (BS) or a user equipment (UE). Device 500 can have three systems: a perception system 510, a communication system 520, and / or an AI system 530. The perception system 510 perceives and collects signals and data, the communication system 520 transmits and receives signals and data, and the AI system 530 trains and infers AI implementations. An exemplary AI implementation is based on deep learning with two cycles, a training cycle and an inference cycle. In some possible application scenarios, the training cycle can also be referred to as a learning cycle, and the inference cycle can also be referred to as a reasoning cycle.

[0138] Deep learning consists of two cycles: training (or learning) and inference (or reasoning). In the training cycle, the coefficients of neurons are learned from training data to achieve a specific training purpose or goal. In the inference or reasoning cycle, input data samples are fed into a trained neural network, which outputs predictions.

[0139] In the training cycle, the AI system 530 of the device 500 can train one or more DNNs, where the perception system 510 of the device 500 can generate signals and / or data. The communication system 520 of the device 500 can receive signals or data from another device or other devices. During and / or after the AI system 530 completes the training, the communication of the device can transmit the training results to another device or other devices.

[0140] In the inference cycle, the AI system 530 of the device 500 can perform an inference or a series of inferences on one or more DNNs to complete one or more tasks, where the perception system 510 of the device 500 can generate signals and / or data, and the communication system 520 of the device 500 can receive signals or data from another device or other devices. After the AI system 530 of the device 500 completes the inference, the communication system 520 of the device 500 can transmit the inference results to another device or other devices.

[0141] The AI implementation can switch between the two cycles or stay in both cycles at the same time. For example, the AI system 530 of the device 500 can train a second DNN, but still perform inferences on a first DNN.

[0142] In a training period, the AI system 530 of the device 500 can work in a single-user mode. In this mode, the AI system 530 trains one or more DNNs with data provided by the perception system 510 of the device 500. Examples of the data include local perception data and local channel data. The local perception data includes RGB data, light detection and ranging (LiDAR) data, temperature data, barometric pressure data, outage data, etc. The local channel data includes channel state information (CSI), received signal strength indicator (RSSI), latency data, etc.

[0143] Alternatively, the AI system 530 of the device 500 can work in a cooperative mode. In this mode, the AI system 530 trains one or more DNNs with data received by the communication system 520 of the device 500. Examples of the data include perception data, channel data, neuron data, and latent output data. The perception data includes RGB data, LiDAR data, temperature data, barometric pressure data, outage data, etc. The channel data includes CSI, RSSI, latency data, etc. The neuron data includes a plurality of neurons or a plurality of gradients. The latent output data includes a plurality of latent outputs.

[0144] Figure 6 is a schematic diagram of a device 500 receiving reference data samples from a device 600. The AI system 530 of the device 500 in the cooperative mode can use the following data: accumulate the perception data received by the communication system 520 of the device 500 into a training data set; accumulate the channel data received by the communication system 520 of the device 500 into a training data set; set the local neurons by the neurons received by the communication system 520 of the device 500, which is a typical federated learning scheme; input the latent output received by the communication system 520 of the device 500 to its DNN.

[0145] Alternatively, the AI system 530 of the device 500 in the collaborative mode can use the data received by the communication system 520 of the device 500 together with its local data, for example: mix the local perception data provided by the perception system 510 of the device 500 with the perception data received by the communication system 520 of the device 500 into one training data set; mix the local channel data provided by the perception system 510 of the device 500 with the channel data received by the communication system 520 of the device 500 into one training data set; average the local neurons owned by the AI system 530 of the device 500 with the neurons received by the communication system 520 of the device 500, which is a typical federated learning scheme; average the local latent outputs owned by the AI system 530 of the device 500 and input them into its DNN.

[0146] Figure 7 is a schematic diagram of a reference data sample consisting of multiple groups, which is an embodiment of the present application. In the training period, the communication system 520 of the device 500 can receive some reference data samples in the single-user mode or the collaborative mode. Some devices send the reference data samples in the broadcast, multicast or unicast channel. Other devices send one or more indicators about which one or more layers the reference data samples are related to, where, for example, there are three groups of reference data samples: the first group of reference data samples is indicated to be related to the input layer of the DNN, the second group of reference data samples is indicated to be related to one latent layer output of the DNN, and the third group of reference data samples is indicated to be related to the layer output of the DNN.

[0147] The AI system 530 of the device 500 can measure the distance between its local data samples and the reference data samples group by group. The AI system 530 of the device 500 can sample its local layer input, local latent layer output and / or layer output randomly, non-randomly, uniformly or non-uniformly. Then, the AI system 530 of the device 500 measures the distance between the local samples and the reference samples received by the communication system 520 of the device 500. If the average distance of all groups is always below one or more predefined thresholds, the AI system 530 of the device 500 can tell that the current training process is working as expected, otherwise the AI system 530 can tell that it is abnormal.

[0148] In the case of a device without an AI system but with a perception and communication system, the perception system of the device can still measure the distance between its local data samples and the reference data samples related to the layer input of the DNN. If the average distance on the layer input is below a predefined threshold, the perception system of the device can consider that the perception device is capturing “good” data, otherwise the perception device is capturing bad data. The communication system of the device can only send good data to other devices without sending bad data to other devices, or the communication system of the device can mark the perception data with the distance before sending it to other devices.

[0149] The UE can report information about its data to the BS, and then the BS determines whether the data is significantly different from the training data. If the difference is too large, the BS can switch the working mode from the AI mode to the non-AI mode, or switch to another AI model. However, it may be considered as an invasion of user privacy for the UE to directly report the original data. It is inefficient or violates privacy policies to send the original data over the air. Therefore, how to safely and efficiently send data state information is a technical problem to be solved.

[0150] In order to protect the original data and save bandwidth, a set of reference data samples are encoded or compressed into a lower-dimensional space than their original space. The encoder or compressor can be linear or nonlinear. A linear encoder can be implemented using some standard bases such as Fourier bases, DCT, wavelets, etc., or a linear encoder can be implemented using some custom bases. These bases can be composed of unitary matrices (orthogonal). A nonlinear encoder can be implemented with some DNN. Figure 8 is a schematic diagram of a DNN-based approximation of an embodiment of the present application.

[0151] Unlike traditional compression schemes that are built for reliable reconstruction, when compressing data into a lower-dimensional space, the encoder deliberately avoids reliable reconstruction, but preserves as much topological distance as possible. That is, the relative distance of two data samples in their original signal space can be well preserved after encoding into a low-dimensional space.

[0152] Figure 9 is a flowchart of a communication method of an embodiment of the present application.

[0153] 710, sending the first coefficient.

[0154] The first coefficient is determined according to the first data and the reference bases, and the dimension of the first coefficient is less than the dimension of the first data.

[0155] The first data includes monitoring data or measured data of a user equipment or a network equipment. In addition, the first data is monitoring data or measured data related to an AI model. The network equipment in this embodiment can be a BS. If the first data is data sent by the UE to the BS through the uplink, the data is monitoring data or measured data of the UE. If the first data is data sent by the BS to the UE through the downlink, the data is monitoring data or measured data of the BS.

[0156] One or more reference bases are predefined or configured. The reference base is one of the plurality of predefined or configured reference bases. For example, the BS can configure the reference bases for the UE. The reference bases can be orthogonal bases, and any two columns of the reference bases are completely orthogonal to each other. A typical orthogonal base is a DFT base.

[0157] 720, communicate according to the first coefficient.

[0158] Figure 10 This is a flowchart illustrating a communication method according to an embodiment of this application. To protect the original data and save bandwidth, a set of reference data samples is encoded or compressed into a space with a lower dimension than its original space. The encoder or compressor can be linear or nonlinear. A linear encoder can be implemented using standard bases such as Fourier bases, discrete cosine transform (DCT), or wavelets. Alternatively, a linear encoder can be implemented using custom bases, which can consist of unitary (orthogonal) matrices. A nonlinear encoder can be implemented using DNNs.

[0159] In the embodiments given below, the UE projects a high-dimensional signal onto a low-dimensional signal (coefficients) through a transformation (orthogonal basis U). Reporting coefficients instead of raw data is effective and beneficial for privacy protection.

[0160] 810, Configure or predefine one or more reference bases.

[0161] The coefficient of reference basis indicator (CRBI) is used to indicate coefficients relative to a reference basis (e.g., an orthogonal basis). Let {u1, u2, ..., u...} r Let} be the orthogonal set of vectors in the subspace Rn. This set forms the basis U of the subspace Rn. The elements represented by the basis U in the subspace Rn can be written as a finite weighted linear combination of the basis elements. The coefficients of this weighted linear combination are called the components or coordinates of the vectors relative to the basis U. ).

[0162] Figure 11 This is a schematic diagram illustrating an embodiment of this application of projecting a high-dimensional signal onto a low-dimensional signal. For example, ,in, It is an n×1 primitive space, and U is an n×r orthogonal basis. It is an r×1 spectral subspace. n is an integer greater than 1, r < 1. <n。 This refers to the data that the UE needs to report, such as perceived data, measured data, AI / machine learning (ML) data, channel data, environmental data, etc. U is both a reference basis and an orthogonal basis; any two columns of U are completely orthogonal to each other. Embodiments of this application can use columns as a basis, which can be easily applied to a basis matrix with rows as the basis. Simply put, U... HOne typical orthogonal basis is the discrete fourier transform (DFT) basis. is a CRBI, used as a reference coefficient.

[0163] is expressed as an n x 1 reference sample, and U is an n x r matrix. Each column of U can be expressed as a weighted linear combination of each column of U: where, is an r x 1 spectral coefficient or weight. In the case of r << n, is the equivalent low-dimensional space signal (vector) of . Matrix U is a unitary matrix that satisfies and . Then, matrix U H is an encoder or compressor that compresses the high-dimensional (n x 1) reference sample into a low-dimensional (r x 1) reference sample

[0164] In one possible implementation scenario, multiple reference bases (U A , U B , U C …) are configured or predefined. The BS configures which reference base to use, for example, U X . The UE reports the CRBI based on U X . According to equation , the UE knows U and , and thus can calculate the coefficient .

[0165] In one possible implementation scenario, one reference matrix U is configured or predefined, and one or more pruning bases are indicated or predefined as reference bases. The reference matrix Y is a matrix with M rows and N columns. The pruning bases of the reference bases are the K columns of Y, for example, the first K columns of Y, where K is configured, K ≤ N. Optionally, it can be specified which K columns of Y are selected as pruning bases.

[0166] 820, the UE determines the coefficient of its reference base.

[0167] The reference base (U) is configured or predefined. The BS can configure one or more reference signals, and the UE can obtain the original data by measuring the reference signal. Optionally, the reference signal can also not be configured, and the UE can obtain the original data by sensing the reference signal. The UE determines its CRBI by . U is a unitary matrix that satisfies the conjugate transpose of the matrix equal to the inverse of the matrix, i.e. , I is the identity matrix.

[0168] The UE can obtain one or more reporting data from a single slot. Based on the observation time interval (or not limited), the UE shall obtain the reported CRBI value in the uplink slot. Illustratively, the UE reports the CRBI value in the uplink slot n. The UE can obtain the corresponding one or more CRBI values by measuring the data within the configured time window n-5 to n-1. The UE can choose to report multiple CRBI values, or choose to report the average value / maximum value / minimum value of multiple CRBI values.

[0169] 830, the UE reports the CRBI or the index of the CRBI.

[0170] Illustratively, the UE obtains P reporting data from the time window n-5 to n-1, which can be obtained by The UE can obtain P CRBI values corresponding to the P reporting data. The UE can choose to report the average value, maximum value, or minimum value of the P CRBI values. The reporting data includes the monitoring data or measured data of the UE.

[0171] The UE can directly report the CRBI, or report the index corresponding to the CRBI. The BS can configure the physical uplink control channel (PUCCH) or the physical uplink shared channel (PUSCH) for the UE to report the CRBI. The CRBI reporting supports periodic, aperiodic, and semi-static.

[0172] In some possible application scenarios, the UE reports the index corresponding to the CRBI. In this scenario, one or more CRBI tables are predefined or configured. A reference base can be associated with one CRBI table, or associated with multiple CRBI tables. When the reference base is associated with multiple CRBI tables, the BS indicates which CRBI table to use.

[0173] The UE reports the CRBI index of the CRBI table. As shown in Table 1, 4 bits are used to indicate the CRBI index. Although the CRBI values in Table 1 are all represented by the same {c0, c1, …, c r}, each CRBI index corresponds to a different CRBI value. In some possible implementations, the value of r in {c0, c1, …, c r} is different in different rows of the CRBI table, for example, some are {c0, c1, …, c5} and some are {c0, c1, …, c6}.

[0174] Table 1

[0175] In some possible implementations, one CRBI index can correspond to one CRBI range, and Table 1 should not be understood as limiting the present application.

[0176] In the communication method provided in the present embodiment, the UE can report its data information to the BS with minimum air interface overhead, and the BS can then determine whether the data is significantly different from the training data, thereby improving the efficiency of data reporting and protecting the privacy of the data.

[0177] Figure 12 is a flowchart of a communication method according to an embodiment of the present application. In the present embodiment, differential CRBI index reporting can be used.

[0178] 910, determine a reference CRBI index.

[0179] The BS can indicate the reference CRBI index, or the reference CRBI index can be configured or predefined.

[0180] 920, report an offset level to the BS.

[0181] The UE reports the offset level to the BS. According to the offset level and the reference CRBI index, the BS knows the current data CRBI index. Exemplarily, the differential CRBI can be obtained by equation (1).

[0182] Offset level = current data CRBI index - reference CRBI index (1) In the communication method provided in the present embodiment, the UE can report its data information to the BS with minimum air interface overhead, and the BS can then determine whether the data is significantly different from the training data, thereby improving the efficiency of data reporting and protecting the privacy of the data.

[0183] In addition, the communication method provided in the present application can also be applied to downlink (DL) transmission of the BS to the UE for indicating the CRBI or CRBI index, for indicating the data information on the BS side. The specific implementation can refer to the description of Figures 9 to 12 and will not be described herein again.

[0184] Figure 13 is a schematic diagram of a matrix U determined according to an embodiment of the present application.

[0185] Each column of the matrix U can be a standard basis such as a Fourier basis, a DCT basis, a wavelet basis, etc. Alternatively, r columns of the matrix U can be constructed on the distribution of the reference sample x group. An example process for calculating the matrix U on the distribution of , ... may be as follows: Accumulate a sufficient number (M) of n x 1 samples , ... , M « n; juxtapose them into an n x M matrix The order of the data samples is not important; apply a descending singular value decomposition (SVD) to : where U is an n x r unitary matrix representing the commonality between all M reference samples .

[0186] Because a set of reference data samples corresponds to a layer of output, each set of reference data samples has its own matrix U. The first set has a compressed version of matrix U1 , the second set has a compressed version of matrix U2 .

[0187] The communication system of the device receives the first matrix U1 and the first set of reference samples (compressed) , and the second matrix U2 and the second set of reference samples (compressed) .

[0188] Figure 14 is a schematic diagram of a first sampling matrix P1 of an embodiment of the present application.

[0189] The first matrix U1 is n1 x r1 and the second matrix U2 is n2 x r2. If n1 and / or n2 are very large numbers, then a first sampling matrix P1 can be applied to the first matrix U1 and a second sampling matrix P2 can be applied to the second matrix U2. The first sampling matrix P1 is m1 x n1 (m1 « n1) with only one "1" per row indicating the position of the to be sampled. The second sampling matrix P2 is m2 x n2 (m2 « n2) with only one "1" per row indicating the position of the to be sampled. The first sampling matrix P1 can "compress" the first matrix U1 (n1 x r1) to m1 x r1 θ1, i.e. θ1 = P1 U1. Because θ1 is much smaller than U1 (because m1 « n1), θ1 can be a better replacement for U1. The second sampling matrix P2 can "compress" the second matrix U2 (n2 x r2) to m2 x r2 θ2, i.e. θ2 = P2 U2. Because θ2 is much smaller than U2 (because m2 « n2), θ2 can be a better replacement for U2.

[0190] Figure 15 is a schematic diagram of a sampling matrix compressing a matrix U of an embodiment of the present application.

[0191] In one possible implementation, the device's communication system receives a first compact matrix θ1, a first sampling matrix P1, and a first set of reference samples (compressed). The device's communication system receives a second compact matrix θ2, a second sampling matrix P2, and a second set of reference samples (compressed). .

[0192] Alternatively, the device's communication system receives a first compact matrix. The left inverse, the first sampling matrix P1, and the first set of reference samples (compressed). The device's communication system receives the second compression matrix. The inverse of the second sampling matrix P2 and the second set of reference samples (compressed). .

[0193] Figure 16 This is a schematic diagram of a scoring distance in a low-frequency space according to an embodiment of this application.

[0194] The device's communication system can receive measurements of the first two samples. and The first scoring function for the distance between The device's communication system can receive measurements of two samples from the second group. and The second scoring function for the distance between The first rating function d1 and the second rating function d2 can be the same or different. The first rating function... Second scoring function This could be a dot product, inner product, Euclidean distance, etc. Alternatively, it could be the first scoring function. Second scoring function It can be based on DNN.

[0195] Alternatively, the device's communication system can receive measurements of the two distributions in the first group. and The first scoring function for the distance between The device's communication system can receive measurements from two distributions in the second group. and The second scoring function for the distance between The first rating function d1 and the second rating function d2 can be the same or different. The first rating function... Second scoring function may be mutual information, a Hilbert-Schmidt independence criterion (HSIC) measure, a KL scattering, a graph edit distance, a Wasserstein distance, a Jensen-Shannon divergence (JSD) distance, or the like. Alternatively, the first scoring function and the second scoring function may be DNN-based.

[0196] Figure 17 is a flowchart of an embodiment of a communication method that is an embodiment of the present application.

[0197] 1710, when the first difference value is less than or equal to the first threshold value, sending the first data to a network device.

[0198] When the first difference value is less than or equal to the first threshold value, the UE sends the first data to the network device. The first difference value is the difference between the first data and the first anchor point. The first anchor point includes one or more reference data. The first threshold value is a predefined or configured threshold value corresponding to the first anchor point.

[0199] The first data includes monitoring data or measured data of the user equipment. In addition, the first data is monitoring data or measured data related to an AI model. The network device in this embodiment can be a base station (BS).

[0200] The first anchor point can be one of N anchor points configured by the BS for the UE, N≥1. The configuration signal can be a radio resource control (RRC), a medium access control-control element (MAC-CE), or a downlink control information (DCI), and can be broadcast, groupcast, or unicast.

[0201] The BS configures a threshold value associated with the anchor point for reporting. In some possible application scenarios, the BS separately configures the threshold value of each anchor point, or predefines the threshold value of the anchor point. In some possible application scenarios, the protocol predefines the threshold value of multiple anchor points. For example, the protocol predefines a threshold value for all anchor points. The threshold value can be reconfigured over time.

[0202] 1720, communicating according to the first data.

[0203] Figure 18This is a flowchart illustrating an embodiment of a communication method according to this application. Embodiments of this application identify data quality on the UE side through one or more anchor points, wherein data quality includes data accuracy and data diversity.

[0204] 1810, BS configures one or more anchor points.

[0205] For the UE, the BS configures one or more anchor points (the number of anchor points is N). The configuration signal can be radio resource control (RRC), medium access control-control element (MAC-CE), or downlink control information (DCI), and can be broadcast, multicast, or unicast.

[0206] Anchor points are a set of reference data, such as reference coefficients ( Reference data () The set can be a vector, such as a one-dimensional array, where the size of the vector is r, and r is predefined or configured. The size of the set is K. (j=1, 2, … K), where K is predefined or configured. Each of the N anchor points includes K reference coefficients.

[0207] 1820, UE calculates its data (e.g., coefficients ( The difference between the anchor point and the anchor point.

[0208] Reference data (e.g., coefficients) The coefficients are the reference coefficients of the reference basis (orthogonal basis U). During the information exchange between the UE and the BS, the UE projects the high-dimensional signal onto the low-dimensional signal (coefficients) through transformation (orthogonal basis U). In ), the transformation equation is... ,in, It refers to the reported data, where U is the reference base. These are reference coefficients. One column U is one of the bases, meaning any two columns U are completely orthogonal to each other. Reported data includes monitoring or measured data from user devices. Additionally, reported data includes monitoring or measured data related to the AI ​​model.

[0209] The reference base (U) is configured or predefined. The UE uses... Determine its reference basis indicator (CRBI) coefficient. H It is the conjugate transpose of U. I is the identity matrix.

[0210] UE computes the difference between its data (e.g., coefficients (a ) ) and the reference data c j (j = 1, 2, … K) in the anchor point, which can be computed by any one of Equations (2), (3), and (4). is the difference between the reported data and the reference data c in the anchor point. is the reported data of the user equipment. is the jthreference data in the anchor point. denotes inner product. denotes norm, which is a method to measure the size of a vector, matrix, tensor, or function. f denotes other custom functions. 1≤j≤K and 1≤i≤r.

[0211] (2) (3) (4) It should be understood that Equations (2) to (4) are only examples, and the UE computes the difference between its data (e.g., coefficients (a ) ) and the reference data c ( j = 1, 2, … K) in the anchor point can also be computed by dot product, Euclidean distance, or DNN-based algorithm, etc., and the above examples should not be understood as a limitation of the present application.

[0212] The anchor point is a set of reference data, and the UE computes the difference between its data and the anchor point according to a method that can be indicated by the BS or predefined, such as Equations (5) or (6). is the difference between the reported data and the anchor point. can be the minimum value of the difference between the reported data and the K reference data in the anchor point, or can be the average value of the difference between the reported data and the K reference data in the anchor point.

[0213] (5) (6) Alternatively, the difference between the data of the UE and the anchor point can also be obtained by mutual information, Hilbert-Schmidt independence criterion (HSIC) metric, Kullback-Leibler (Kullback-Leibler, KL) scattering, graph edit distance, Wasserstein distance, Jensen-Shannon divergence (JSD) distance, DNN-based algorithm, etc.

[0214] 1830, the UE reports the data whose difference is less than or equal to the threshold value.

[0215] The BS configures a threshold value associated with the anchor point for reporting. In some possible application scenarios, the BS separately configures a threshold value for each anchor point, or predefines a threshold value for the anchor point. In some possible application scenarios, the protocol predefines threshold values for multiple anchor points. For example, the protocol predefines a threshold value for all anchor points. The threshold value can be reconfigured over time.

[0216] Figure 19 is a schematic diagram of UE data reporting when one anchor point is configured. When there is one anchor point, the UE calculates the difference between its data and the anchor point according to the method that can be indicated by the BS or predefined, such as equation (5) or (6). When is less than or equal to the threshold value, the UE will report its data to the BS, where the data format can be raw data or the coefficients of the reference basis When is greater than the threshold value, the UE will not report its data to the BS.

[0217] Figure 20 is a schematic diagram of UE data reporting when multiple anchor points are configured. When there are multiple anchor points (N anchor points, N > 1), the UE calculates the distance between its data and anchor point n (n = 1 to N) . For all n, if > threshold_n, the UE will not report its data. threshold_n is the threshold value corresponding to anchor point n. If at least one anchor point satisfies ≤ threshold_n, the UE reports its data to the BS.

[0218] Through anchor point configuration, the UE can determine the quality of the data and report high-quality data to the BS, so as to quickly and accurately train the neural network model.

[0219] Figure 21is a flowchart of an embodiment of a communication method of an embodiment of the present application. Diversity is used to describe the correlation between training samples. High-quality data is essential for training accurate AI models with good generalization performance, while diversified data is necessary to avoid bias and overfitting. Therefore, the UE needs to identify diversified data for reporting to reduce air interface overhead.

[0220] 2110, the UE determines the latest anchor point for its data.

[0221] For the UE, one or more anchor points (the number of anchor points is N) are configured, where an anchor point is a set of reference data. The UE calculates the difference between its data and anchor point n (n = 1 to N) Then find the index k of the latest anchor point. For example, find the index k of the latest anchor point by equation (7).

[0222] (7) The data with the latest anchor point index k is referred to as the data associated with anchor point k in the embodiments of the present application.

[0223] 2120, the BS indicates the anchor point priority.

[0224] The BS configures a priority index for each anchor point. The anchor point index value indicates the priority of the anchor point. For example, the smaller the index value, the higher the priority. Higher priority of an anchor point means that the UE data associated with the anchor point should have higher priority, for example, be reported first.

[0225] Optionally, a priority value can also be configured for each anchor point individually, without using an index value. The above options should not be understood as a limitation of the present application.

[0226] For data with higher priority, the priority is higher in medium access control (MAC) or physical (PHY) layer multiplexing. Multiple data can be multiplexed in a transport block (TB) in the MAC layer or the PHY layer, and the highest priority is the first to be included in the TB.

[0227] Figure 22 is a schematic diagram of UE data reporting according to the priority index of an embodiment of the present application. For example, the BS observes that there are enough data samples associated with anchor point-1 (for example, data collected in an outdoor environment), but much fewer data samples associated with anchor point-2 (for example, data collected in an indoor environment). The BS configures a higher priority for anchor point-2 to collect more corresponding data to train or fine-tune its model.

[0228] 2130, the UE first reports data belonging to the anchor point with the highest priority.

[0229] For UE reporting data, the UE first reports data belonging to the highest priority anchor point.

[0230] The UE can also report its associated anchor point index to the BS according to the above rules. Then, if the BS finds that data associated with anchor point k is needed, the BS instructs the UE to report data associated with anchor point k and also indicates the resource for feedback.

[0231] Optionally, only data with a difference from the latest anchor point (index k) less than a threshold value can be considered as data associated with anchor point k.

[0232] The method provided in the application realizes diversified data reporting and improves the generalization performance of AI / ML.

[0233] Figure 23 is a schematic block diagram of a communication apparatus 2300 according to an embodiment of the application. The communication apparatus 2300 comprises: a sending module 2310 configured to send first data to a network device when a first difference value is less than or equal to a first threshold value, the first difference value being a difference value between the first data and a first anchor point, the first anchor point comprising one or more reference data, the first threshold value being a predefined or configured threshold value corresponding to the first anchor point; and a processing module 2320 configured to perform communication according to the first data.

[0234] In a possible implementation, the first difference value is a minimum value of K second difference values or an average value of the K second difference values, the first anchor point comprises K reference data, a jth second difference value in the K second difference values being a difference value between the first data and a jth reference data in the K reference data, K≥1 and 1≤j≤K.

[0235] In a possible implementation, the sending module is further configured to send an index of a second anchor point to the network device, the second anchor point being one of N anchor points, wherein a third difference value corresponding to the second anchor point is a minimum third difference value in N third difference values, an nth third difference value in the N third difference values being a difference value between the first data and an nth anchor point, N≥1 and 1≤n≤N.

[0236] In a possible implementation, any anchor point in the N anchor points corresponds to a priority, and the sending module is further configured to send the first data in a priority order of the second anchor point in the N anchor points.

[0237] In a possible implementation, the index value of the anchor point is a priority corresponding to the anchor point.

[0238] In a possible implementation, the N anchor points are configured by a radio resource control (RRC), a medium access control-control element (MAC-CE), or a downlink control information (DCI) signal from the network device.

[0239] In a possible implementation, the first data is transmitted by a transport block at a medium access control (MAC) layer or a physical (PHY) layer of the user equipment.

[0240] In a possible implementation, the first data includes monitoring data or measured data of the user equipment.

[0241] In a possible implementation, the first data includes any one or more of the following: perception data, measured data, channel data, neuron data of an artificial intelligence (AI) model, and latent output data of the AI model.

[0242] In a possible implementation, the first data is a coefficient of a predefined or configured orthogonal basis , the first anchor point includes K reference coefficients c k of the orthogonal basis, k = 1, 2, …, K, and 1 ≤ k ≤ K.

[0243] As shown in Figure 24 , the communication apparatus 2400 can include a processor 2410 and a transceiver 2420. Optionally, the communication apparatus 2400 can further include a memory 2430. The memory 2430 can be used to store indication information, and can also be used to store codes, instructions, and the like to be executed by the processor 2410.

[0244] The memory 2430 can include random access memory, flash memory, read-only memory, programmable read-only memory, non-volatile memory, registers, and the like. The processor 2410 can be a central processing unit (CPU).

[0245] For other functions and operations of the communication apparatus 2400, refer to the processes of the method embodiments of Figures 5 to 22 , and to avoid repetition, they will not be described here.

[0246] Embodiments of the present application also provide a computer storage medium, which can store program instructions to perform the steps in the above method.

[0247] Optionally, the storage medium can be specifically the memory 2430.

[0248] The embodiment of the present application further provides a computer program product. The computer program product comprises computer program codes. When the computer program codes run on a computer, the computer is caused to execute the steps in the method.

[0249] Optionally, all or part of the computer program codes can be stored in the first storage medium. The first storage medium can be packaged with the processor or packaged separately from the processor.

[0250] The embodiment of the present application further provides a system chip. The system chip comprises an input / output interface, at least one processor, at least one memory and a bus. The at least one memory is used for storing instructions, and the at least one processor is used for calling the instructions of the at least one memory to execute the operations in the method in the above embodiment.

[0251] A person of ordinary skill in the art can understand that all or part of the processes in the embodiment method can be implemented by a computer program instructing relevant hardware. The program can be stored in a computer readable storage medium. When the program runs, the processes of the method in the embodiment are executed. The above storage medium can include a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0252] In the multiple embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the described device embodiments are merely exemplary. For example, the unit division is merely a logical function division, and an actual implementation can be other division manners. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be implemented by some interfaces. The indirect couplings or communication connections between the devices or units can be implemented in electronic, mechanical or other forms.

[0253] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, and can be located in one position, or distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiment solutions.

[0254] In addition, the functional units in the embodiments of the present application can be integrated into one processing unit, or each unit can exist physically as separated, or two or more units are integrated into one unit.

[0255] The above merely provides exemplary embodiments of the present application. Various modifications and variations can be made to the present application by those skilled in the art without departing from the scope of the present application.

Claims

1. A communication method, characterized in that, include: When the first difference is less than or equal to the first threshold, the first data is sent to the network device. The first difference is the difference between the first data and the first anchor point. The first anchor point includes one or more reference data. The first threshold is a predefined or configured threshold corresponding to the first anchor point. Communication is performed based on the first data.

2. The method according to claim 1, characterized in that, The first difference is the minimum of K second differences or the average of K second differences. The first anchor point includes K reference data. The j-th second difference among the K second differences is the difference between the first data and the j-th reference data among the K reference data, where K ≥ 1 and 1 ≤ j ≤ K.

3. The method according to claim 1 or 2, characterized in that, Also includes: Send the index of the second anchor point to the network device. The second anchor point is one of N anchor points. The third difference corresponding to the second anchor point is the smallest third difference among the N third differences. The nth third difference among the N third differences is the difference between the first data and the nth anchor point, where N≥1 and 1≤n≤N.

4. The method according to claim 3, characterized in that, Each of the N anchor points corresponds to a priority level. Sending the first data to the network device includes: The first data is sent according to the priority order of the second anchor point among the N anchor points.

5. The method according to any one of claims 1 to 4, characterized in that, The index value of the anchor point corresponds to the priority of that anchor point.

6. The method according to any one of claims 1 to 5, characterized in that, The N anchor points are configured by radio resource control, media access control-control elements, or downlink control information signals from the network device.

7. The method according to any one of claims 1 to 6, characterized in that, The first data is sent via a transport block at the media access control layer or physical layer of the user equipment.

8. The method according to any one of claims 1 to 7, characterized in that, The first data includes monitoring data or measured data from user equipment.

9. The method according to any one of claims 1 to 8, characterized in that, The first data includes any one or more of the following: perception data, measured data, channel data, neuron data of the artificial intelligence model, and potential output data of the AI ​​model.

10. The method according to any one of claims 1 to 9, characterized in that, The first data is the coefficients of a predefined or configured orthogonal basis. The first anchor point includes K reference coefficients c of the orthogonal basis. k , k=1, 2...K, 1≤k≤K.

11. A communication device, characterized in that, include: The sending module is configured to send first data to the network device when the first difference is less than or equal to the first threshold, wherein the first difference is the difference between the first data and the first anchor point, the first anchor point includes one or more reference data, and the first threshold is a predefined or configured threshold corresponding to the first anchor point. The processing module is used to communicate based on the first data.

12. The communication device according to claim 11, characterized in that, The first difference is the minimum of K second differences or the average of K second differences. The first anchor point includes K reference data. The j-th second difference among the K second differences is the difference between the first data and the j-th reference data among the K reference data, where K ≥ 1 and 1 ≤ j ≤ K.

13. The communication device according to claim 11 or 12, characterized in that, The sending module is further configured to send the index of the second anchor point to the network device. The second anchor point is one of N anchor points. The third difference corresponding to the second anchor point is the smallest third difference among the N third differences. The nth third difference among the N third differences is the difference between the first data and the nth anchor point, where N≥1 and 1≤n≤N.

14. The communication device according to claim 13, characterized in that, Each of the N anchor points corresponds to a priority, and the sending module is further configured to send the first data in the order of priority of the second anchor point among the N anchor points.

15. The communication device according to any one of claims 11 to 14, characterized in that, The index value of the anchor point corresponds to the priority of that anchor point.

16. The communication device according to any one of claims 11 to 15, characterized in that, The N anchor points are configured by radio resource control, media access control-control elements, or downlink control information signals from the network device.

17. The communication device according to any one of claims 11 to 16, characterized in that, The first data is sent via a transport block at the media access control layer or physical layer of the user equipment.

18. The communication device according to any one of claims 11 to 17, characterized in that, The first data includes monitoring data or measured data from user equipment.

19. The communication device according to any one of claims 11 to 18, characterized in that, The first data includes any one or more of the following: perception data, measured data, channel data, neuron data of the artificial intelligence model, and potential output data of the AI ​​model.

20. The communication device according to any one of claims 11 to 19, characterized in that, The first data is the coefficients of a predefined or configured orthogonal basis. The first anchor point includes K reference coefficients c of the orthogonal basis. k , k=1, 2...K, 1≤k≤K.

21. A communication device, characterized in that, The device includes a processor and a memory, the processor being connected to the memory; wherein the memory is used to store instructions, and the processor is used to execute the instructions; when the processor executes the instructions stored in the memory, the processor performs the method according to any one of claims 1 to 10.

22. A communication system, characterized in that, Includes network equipment and communication devices according to any one of claims 10 to 20.

23. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a processor, cause the processor to perform the method according to any one of claims 1 to 10.

24. A computer program product, characterized in that, It includes computer program code that, when run on a computer, causes the computer to perform the method according to any one of claims 1 to 10.

Images