Configuration updating method and device

Abstract

The invention discloses a configuration updating method and a related device. The method comprises the following steps: acquiring monitoring data corresponding to a first model; whether configuration information of a first channel is updated or not is determined according to changes of the monitoring data, the first channel is used for transmitting all or part of first data, and the first data comprises at least one of the monitoring data, training data of a first model and reasoning data of the first model. By monitoring the model parameters, whether the change of the parameters corresponding to the model needs to cause updating of the channel configuration information for transmitting the model parameters or not is judged, so that the suitability of the configuration related parameters of the channel and the corresponding parameters of the model is ensured, and the effectiveness of model life cycle management is improved.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a configuration update method and apparatus. Background Technology

[0002] Over the past few decades, wireless communication systems have evolved and been researched and developed from first-generation analog communication to 5G New Radio (NR) and future wireless communication technologies. Throughout this complex evolution, high throughput and massive connectivity have remained core challenges for wireless communication networks. To address these challenges, 5G communication has set technological goals such as enhanced mobile broadband (eMBB), ultra-reliable low-latency communications (URLLC), and massive machine-type communications (mMTC). Future wireless communication systems will inevitably evolve towards higher throughput, lower latency, higher reliability, greater connection capacity, and higher spectrum utilization.

[0003] When a terminal device reports data to the network side through a physical channel, the usual practice is for the terminal to map a logical channel with a fixed configuration corresponding to the quality of service (QoS) of the data to the physical channel, and then transmit the reported data. However, due to the increase in data throughput and the need for lower latency data transmission, when the QoS of the reported data changes, the original logical channel is difficult to adapt to the changed QoS, resulting in a decrease in data transmission speed. Summary of the Invention

[0004] This application provides a configuration update method that can improve channel adaptability and improve communication performance.

[0005] In a first aspect, embodiments of this application provide a configuration update method, which can be applied to the terminal side, such as a terminal or a communication module in the terminal, or a circuit or chip in the terminal responsible for communication functions (such as a modem chip, also known as a baseband chip, or a system-on-chip (SoC) chip or system-in-package (SIP) chip containing a modem core). Taking the application of this method to a terminal as an example, the method includes: acquiring monitoring data corresponding to a first model; determining whether to update the configuration information of a first channel based on changes in the monitoring data, wherein the first channel is used to transmit all or part of the first data, and the first data includes at least one of the monitoring data, the training data of the first model, and the inference data of the first model.

[0006] By using the above method, by monitoring the model parameters, it can be determined whether changes in the parameters corresponding to the model require updating the channel configuration information for transmitting model parameters, so as to ensure the compatibility between the channel configuration parameters and the corresponding model parameters and improve the effectiveness of model lifecycle management.

[0007] In one possible design, the monitoring data includes performance data, distribution data, and time data; determining whether to update the configuration information of the first channel based on changes in the monitoring data includes: updating the configuration information of the first channel in response to the performance data meeting a first performance threshold within a first observation period; and / or updating the configuration information of the first channel in response to an anomaly in the distribution data within a second observation period; and / or updating the configuration information of the first channel in response to the time data meeting a first time threshold.

[0008] By using the above method, we can identify various scenarios in which channel parameters need to be updated, thereby improving the reliability of channel configuration information updates.

[0009] In one possible design, the observation is a periodic observation; the performance data meeting a first performance threshold within a first observation period includes: the performance data meeting the first performance threshold within a first number of observation periods; the distribution data exhibiting anomalies within a second observation period includes: the distribution of the distribution data exhibiting anomalies within a second number of observation periods.

[0010] Using the above method can reduce the time spent monitoring model-related parameters and lower power consumption.

[0011] In one possible design, the configuration information includes bit rate and priority.

[0012] By using the above method, updating the channel's bit rate and priority can speed up the transmission of terminal data and improve transmission efficiency.

[0013] In one possible design, the performance data satisfies a first performance threshold, including: the performance data is less than or equal to the first performance threshold during the observation period.

[0014] In one possible design, the first performance threshold is the difference between the base sub-threshold and the dynamic sub-threshold.

[0015] By adopting the above method, we can avoid frequent updates of channel configuration information caused by the model parameters switching back and forth around the basic sub-threshold, thus reducing power consumption.

[0016] In one possible design, the performance data includes model accuracy or communication metrics.

[0017] In one possible design, the distribution data includes the distribution of input data and output data; the distribution data is abnormal, including: the samples of the input data and output data do not conform to the distribution of historical input data and output data; and / or, the samples of the input data and output data do not conform to the distribution of training input data and output data; and / or, the drift of the input data and output data during the observation time is greater than or equal to a first drift threshold.

[0018] In one possible design, the method further includes: restoring the configuration information to its previous state in response to the performance data not meeting the first performance threshold during a third observation time; or, restoring the configuration information to its previous state in response to the performance data meeting the second performance threshold during the third observation time; and / or, restoring the configuration information to its previous state in response to the distribution data being normal during a fourth observation time; wherein the third observation time is after the first observation time, and the fourth observation time is after the second observation time.

[0019] By using the above method, when the relevant parameters of the model recover from anomalies to normal, the channel configuration information can also be restored to the state before the update, thus avoiding the accelerated transmission of model-related parameters from affecting the transmission of other data and improving communication efficiency.

[0020] In one possible design, the method further includes updating the configuration information according to the priority of the monitoring data.

[0021] By using the above method, abnormal events in the model can be classified, and high-priority abnormal events can be processed first, thereby improving the management effectiveness of configuration information updates.

[0022] In one possible design, the time data includes: the difference between the end time of training data collection and the end time of training data transmission for the first model, and / or, the difference between the end time of aggregation and the end time of model data transmission for the first model, and / or, the difference between the end time of inference data collection and the end time of inference data transmission for the first model.

[0023] In one possible design, the first time threshold is the transmission delay; the time data satisfies the first time threshold by: the time data being less than the transmission delay.

[0024] In one possible design, the method further includes: receiving configuration information of the transmission delay, wherein the configuration information indicates the magnitude of the transmission delay.

[0025] Secondly, at least one embodiment of this application also provides a configuration update apparatus, comprising: a transceiver unit for acquiring monitoring data corresponding to a first model; and a processing unit for determining whether to update the configuration information of a first channel based on changes in the monitoring data, wherein the first channel is used to transmit all or part of first data, and the first data includes at least one of the monitoring data, training data of the first model, and inference data of the first model.

[0026] In one possible design, the monitoring data includes performance data, distribution data, and time data; the processing unit is further configured to: update the configuration information of the first channel in response to the performance data meeting a first performance threshold within a first observation period; and / or, update the configuration information of the first channel in response to an anomaly occurring in the distribution data within a second observation period; and / or, update the configuration information of the first channel in response to the time data meeting a first time threshold.

[0027] In one possible design, the observation is a periodic observation; the performance data meeting a first performance threshold within a first observation period includes: the performance data meeting the first performance threshold within a first number of observation periods; the distribution data exhibiting anomalies within a second observation period includes: the distribution of the distribution data exhibiting anomalies within a second number of observation periods.

[0028] In one possible design, the configuration information includes bit rate and priority.

[0029] In one possible design, the performance data satisfies a first performance threshold, including: the performance data is less than or equal to the first performance threshold during the observation period.

[0030] In one possible design, the first performance threshold is the difference between the base sub-threshold and the dynamic sub-threshold.

[0031] In one possible design, the performance data includes model accuracy or communication metrics.

[0032] In one possible design, the distribution data includes the distribution of input data and output data; the distribution data is abnormal, including: the samples of the input data and output data do not conform to the distribution of historical input data and output data; and / or, the samples of the input data and output data do not conform to the distribution of training input data and output data; and / or, the drift of the input data and output data during the observation time is greater than or equal to a first drift threshold.

[0033] In one possible design, the processing unit is further configured to: restore the configuration information to its previous state in response to the performance data not meeting the first performance threshold during a third observation time; or, restore the configuration information to its previous state in response to the performance data meeting the second performance threshold during the third observation time; and / or, restore the configuration information to its previous state in response to the distribution data being normal during a fourth observation time; wherein the third observation time is after the first observation time, and the fourth observation time is after the second observation time.

[0034] In one possible design, the processing unit is further configured to update the configuration information according to the priority of the monitoring data.

[0035] In one possible design, the time data includes: the difference between the end time of training data collection and the end time of training data transmission for the first model, and / or, the difference between the end time of aggregation and the end time of model data transmission for the first model, and / or, the difference between the end time of inference data collection and the end time of inference data transmission for the first model.

[0036] In one possible design, the first time threshold is the transmission delay; the time data satisfies the first time threshold by: the time data being less than the transmission delay.

[0037] In one possible design, the transceiver unit is further configured to: receive configuration information corresponding to the transmission delay, wherein the configuration information indicates the magnitude of the transmission delay.

[0038] Thirdly, this application provides a communication device capable of executing the method described in the first aspect. The communication device possesses the functions described in the first aspect. For example, the communication device includes modules, units, or means corresponding to the operations involved in the first aspect. These modules, units, or means can be implemented through software, hardware, or a combination of software and hardware. The operations performed by this communication device and its beneficial effects can be found in the method described in the first aspect and its beneficial effects.

[0039] Fourthly, the application provides a communication device comprising a memory and one or more processors. The memory stores part or all of the computer program or instructions necessary for implementing the functions described in the first aspect. The one or more processors are executable to carry out the computer program or instructions, which, when executed, cause the communication device to implement the methods in any possible design or implementation of the first aspect.

[0040] In one possible design, the communication device may further include an interface circuit, through which the processor communicates with other devices or components.

[0041] In one possible design, the communication device may also include the memory.

[0042] The aforementioned communication device may be a terminal, or a communication module in a terminal, or a chip in a terminal that is responsible for communication functions, such as a modem chip (also known as a baseband chip), or a chip-on-a-chip system or system-in-package chip containing a modem module.

[0043] Fifthly, this application provides a computer-readable storage medium storing instructions or programs that, when executed on a communication device, cause the communication device to execute the methods of the first aspect or any possible implementation thereof.

[0044] Sixthly, this application provides a computer program product, the computer program product including a computer program or instructions, and instructions for the method in any possible implementation of the first aspect when the computer program or instructions are run on a computer. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the architecture of the communication system used in the embodiments of this application;

[0046] Figure 2 This is a schematic diagram of a configuration update method provided in an embodiment of this application;

[0047] Figure 3A This is a schematic diagram of performance data monitoring provided in an embodiment of this application;

[0048] Figure 3B This is another performance data monitoring diagram provided in the embodiments of this application;

[0049] Figure 3C This is another performance data monitoring diagram provided in the embodiments of this application;

[0050] Figure 3D This is another performance data monitoring diagram provided in the embodiments of this application;

[0051] Figure 4A This is a schematic diagram of terminal data reporting time provided in an embodiment of this application;

[0052] Figure 4B This is another schematic diagram of terminal data reporting time provided in the embodiments of this application;

[0053] Figure 4CThis is another schematic diagram of terminal data reporting time provided in the embodiments of this application;

[0054] Figure 5 This is a schematic diagram of a communication system structure provided in an embodiment of this application;

[0055] Figure 6 This is a schematic diagram of a communication device structure provided in an embodiment of this application; and

[0056] Figure 7 This is a schematic diagram of another communication device structure provided in an embodiment of this application. Detailed Implementation

[0057] This application provides a communication method, apparatus, storage medium, and computer program product for improvement. The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0058] The technical solution of this application can be applied to terrestrial networks (TN), non-terrestrial networks (NTN), or scenarios where NTN and TN are integrated. NTN systems can be, for example, satellite communication systems, high altitude platform station (HAPS) communication systems, global navigation satellite systems (GNSS), etc. TN systems can be, for example, 4th generation (4G) communication systems (e.g., long term evolution (LTE) systems), worldwide interoperability for microwave access (WiMAX) communication systems, 5th generation (5G) communication systems (e.g., new radio (NR) systems), 6th generation mobile communication (6G) systems, and future mobile communication systems, etc.

[0059] To better understand the embodiments of this application, the network architecture of the embodiments of this application will be described below. Please refer to... Figure 1 , Figure 1 This is a schematic diagram of the architecture of the communication system used in the embodiments of this application. It should be noted that... Figure 1 This is a schematic diagram of one possible, non-limiting system. For example... Figure 1As shown, the communication system 10 includes a radio access network (RAN) 100 and a core network (CN) 200. Optionally, the communication system 10 may also include an Internet 300. The RAN 100 includes at least one RAN node (e.g., Figure 1 110a and 110b (collectively referred to as 110) and at least one terminal (such as Figure 1 RAN 100, denoted as RAN 120a-120j, is collectively referred to as RAN 120. RAN 100 may also include other RAN nodes, such as wireless relay equipment and / or wireless backhaul equipment. Figure 1 (Not shown in the image). Terminal 120 is connected to RAN node 110 wirelessly. RAN node 110 is connected to core network 200 wirelessly or via wired connection. The core network elements in core network 200 and RAN node 110 in RAN 100 can be different physical devices, or they can be the same physical device integrating core network logical functions and radio access network logical functions, or they can be a single physical device integrating some core network element functions and some RAN node 110 functions. Terminals can be interconnected with each other, and RAN nodes 110 can be interconnected with each other via wired or wireless connection. Figure 1 This is just an illustration; the communication system may also include other network devices, such as wireless repeaters and wireless backhaul devices. Figure 1 It is not shown in the middle.

[0060] RAN 100 can be a cellular system related to the 3rd Generation Partnership Project (3GPP), such as 4G, 5G mobile communication systems, or future-oriented evolution systems (such as 6G mobile communication systems). RAN 100 can also be an open access network (O-RAN or ORAN), a cloud radio access network (CRAN), or a wireless fidelity (WiFi) system. RAN 100 can also be a communication system that integrates two or more of the above systems.

[0061] RAN node 110, sometimes also referred to as radio access network equipment, access network device, RAN entity, or access node, constitutes part of the communication system and is used to help terminals achieve wireless access. Multiple RAN nodes 110 in communication system 10 can be of the same type or different types. In some scenarios, the roles of RAN node 110 and terminal 120 are relative, for example... Figure 1Network element 120i can be a helicopter or a drone, and it can be configured as a mobile base station. For terminals 120j that access RAN 100 through network element 120i, network element 120i is a base station; however, for base station 110a, network element 120i is a terminal. RAN node 110 and terminal 120 are sometimes referred to as communication devices, for example... Figure 1 Network elements 110a and 110b can be understood as communication devices with base station functions, while network elements 120a-120j can be understood as communication devices with terminal functions.

[0062] In one possible scenario, RAN node 110 can be a base station, an evolved NodeB (eNodeB), an access point (AP), a transmission reception point (TRP), a next-generation NodeB (gNB), a next-generation base station in a 6th-generation (6G) mobile communication system, a base station in a future mobile communication system, or an access node in a WiFi system, etc. Figure 1 110a), micro base stations or indoor stations (such as Figure 1 The RAN node 110 can be a relay node or donor node, or a radio controller in a CRAN scenario. Optionally, the RAN node 110 can also be a server, wearable device, vehicle, or in-vehicle equipment. For example, the radio access network equipment in vehicle-to-everything (V2X) technology can be a roadside unit (RSU). All or part of the functions of the RAN node 110 in this application can also be implemented through software functions running on hardware, or through virtualization functions instantiated on a platform (e.g., a cloud platform). The RAN node 110 in this application can also be a logical node, logical module, or software capable of implementing all or part of the functions of the RAN node 110.

[0063] In another possible scenario, multiple RAN nodes 110 collaborate to assist the terminal in achieving wireless access, with each RAN node 110 implementing a portion of the base station's functions. For example, a RAN node 110 can be a central unit (CU), a distributed unit (DU), a CU-control plane (CP), a CU-user plane (UP), or a radio unit (RU), etc. CUs and DUs can be configured separately or included in the same network element, such as a baseband unit (BBU). RUs can be included in radio frequency equipment or radio frequency units, such as remote radio units (RRUs), active antenna units (AAUs), or remote radio heads (RRHs).

[0064] In different systems, CU (or CU-CP and CU-UP), DU, or RU may have different names, but those skilled in the art will understand their meaning. For example, in an ORAN system, CU can also be called O-CU (open CU), DU can also be called O-DU, CU-CP can also be called O-CU-CP, CU-UP can also be called O-CU-UP, and RU can also be called O-RU. For ease of description, this application uses CU, CU-CP, CU-UP, DU, and RU as examples. Any of the units among CU (or CU-CP, CU-UP), DU, and RU in this application can be implemented through software modules, hardware modules, or a combination of software and hardware modules.

[0065] A terminal can also be called a terminal device, user equipment (UE), mobile station, mobile terminal, etc. Terminals can be widely used in various scenarios, such as device-to-device (D2D), vehicle-to-everything (V2X) communication, machine-type communication (MTC), Internet of Things (IoT), virtual reality, augmented reality, industrial control, autonomous driving, telemedicine, smart grids, smart furniture, smart offices, smart wearables, smart transportation, smart cities, etc. Terminals can be mobile phones, tablets, computers with wireless transceiver capabilities, wearable devices, vehicles, drones, helicopters, airplanes, ships, robots, robotic arms, smart home devices, etc. The embodiments of this application do not limit the device form of the terminal.

[0066] For ease of description, the following description uses a base station as an example of RAN node 110. Base stations and terminals can be fixed or mobile. Base stations and terminals can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on water; and they can be deployed on aircraft, balloons, and satellites. The embodiments of this application do not limit the application scenarios of the base stations and terminals.

[0067] The roles of base stations and terminals can be relative, for example, Figure 1 The helicopter or drone 120i can be configured as a mobile base station. For terminals 120j accessing the wireless access network 100 via 120i, terminal 120i is a base station; however, for base station 110a, 120i is a terminal, meaning that 110a and 120i communicate via a wireless air interface protocol. Of course, 110a and 120i can also communicate via a base station-to-base station interface protocol; in this case, 120i is also a base station relative to 110a. Therefore, both base stations and terminals can be collectively referred to as communication devices. Figure 1 The 110a and 110b in the text can be referred to as communication devices with base station functions. Figure 1 The 120a-120j in the text can be referred to as communication devices with terminal functions.

[0068] Communication between base stations and terminals, between base stations, and between terminals can be conducted using licensed spectrum, unlicensed spectrum, or both simultaneously. Communication can be conducted using spectrum below 6 GHz, spectrum above 6 GHz, or both simultaneously. The embodiments of this application do not limit the spectrum resources used for wireless communication.

[0069] In the embodiments of this application, the functions of the base station can be executed by modules (such as chips) within the base station, or by a control subsystem that includes base station functions. This control subsystem, including base station functions, can be a control center in the aforementioned application scenarios such as smart grids, industrial control, intelligent transportation, and smart cities. Similarly, the functions of the terminal can be executed by modules (such as chips or modems) within the terminal, or by a device that includes terminal functions.

[0070] In this application, the base station sends downlink signals or downlink information to the terminal, with the downlink information carried on the downlink channel; the terminal sends uplink signals or uplink information to the base station, with the uplink information carried on the uplink channel. To communicate with the base station, the terminal needs to establish a radio connection on a cell controlled by the base station. The cell with which the terminal has established a radio connection is called the terminal's serving cell. When the terminal communicates with this serving cell, it is also susceptible to interference from signals from neighboring cells.

[0071] To facilitate understanding of the relevant content of the embodiments of this application, some terms and processes involved in the embodiments of this application will be explained below. This part is only for the purpose of understanding and should not be regarded as a disclosure or specific limitation of the technical solution of this application.

[0072] 1. The Media Access Control (MAC) layer, supported by the Logical Link Control (LLC) layer, provides access control functions to the Physical Layer (PHY) (such as addressing modes, access coordination, frame check sequence generation and inspection, and delimitation of logical link control packet data units). The MAC layer provides data transmission services on logical channels, the type of which is defined based on the type of data transmitted on those channels. When data is transmitted on multiple logical channels and the total data volume exceeds the current transport time interval (TTI), the MAC layer's multiplexing function is triggered.

[0073] The MAC layer's multiplexing function combines data packets from multiple logical channels into a single data packet, which is then transmitted through a physical layer channel (such as the Physical Downlink Shared Channel, PDSCH). The order in which data packets from multiple logical channels are multiplexed into a single data packet is determined by the logical channel priorities. Specifically, higher-priority logical channels are grouped first, meaning the data packets are grouped in descending order of priority. Higher-priority logical channels receive more transmission opportunities, resulting in higher transmission rates or lower latency. Currently, the priority of logical channels is primarily determined by the Quality of Service (QoS) parameter of the data carried by the corresponding logical channel.

[0074] Logical channels can be divided into downlink logical channels and uplink logical channels based on the direction of data transmission. In downlink logical channels, access network devices determine the indication priority of a logical channel based on its type and the QoS parameters of the data it carries. In uplink logical channels, once a terminal receives a scheduling opportunity (i.e., obtains uplink data transmission resources), it can determine the indication priority of each uplink logical channel based on its own circumstances.

[0075] Logical channels are typically divided into control channels and traffic channels. Control channels are mainly used to transmit signaling or synchronization data; traffic channels are used to transmit encoded voice or user data. Control channels have a higher priority than traffic channels. The priority among traffic channels depends on the QoS parameters of the data carried by the corresponding logical channel.

[0076] 2. The physical layer provides the transmission medium and interconnection devices for data communication between devices, ensuring a reliable environment for data transmission. In the transmitting device, the physical layer processes the data streams (transport blocks / transport block sets) from the MAC layer and higher layers through multiplexing and channel coding, mapping from the transport channel to the physical channel, and spreading and modulation of the physical channel to form the data stream for the wireless interface, which is then transmitted over the wireless interface. In the receiving device, the process is reversed.

[0077] 3. Service quality

[0078] QoS is a metric for the overall performance of services experienced by users within a network. To quantitatively measure QoS, aspects of the service are considered, including packet loss, bit rate, throughput, transmission delay, and availability. QoS encompasses requirements for all aspects of connectivity, such as service response time, loss, signal-to-noise ratio, crosstalk, echo, outages, frequency response, and / or loudness level. In 5G NR, each QoS flow packet (e.g., data packet) is classified and labeled using a QoS Flow Identifier (QFI). For example, the first QoS flow is associated with video packets, and the second QoS flow is associated with video streaming packets.

[0079] Within a 5G network, a 5G QoS Identifier (5QI) mechanism can be used, where individual QoS flow packets are classified into different QoS categories. In this way, QoS values ​​can be configured based on the category, so that each QoS class has its own assigned QoS characteristics (e.g., packet delay and packet loss), and different packets can have different QoS.

[0080] 4. Logical Channel Prioritization Process

[0081] For the uplink, the UE creates a MAC protocol data unit (PDU) to transmit using the allocated resources. This is to ensure that the UE meets the QoS of each configured radio bearer for PDU-based service flows. Based on the uplink transmission resource grant message signaled on the physical downlink control channel (PDCCH), the UE can determine the amount of data for each logical channel to be included in the new MAC PDU, and, if necessary, allocate space for the MAC control element (CE).

[0082] In some cases, when a new transmission is performed, a logical channel prioritization process is applied, where data from the highest priority logical channel is included in the MAC PDU first, followed by data from the next highest priority logical channel, and so on, until the MAC PDU size allocated by the network entity is completely filled or there is no more data to send.

[0083] In some cases, radio resource control (RRC) can control the scheduling of uplink data by signaling for each logical channel: priority, prioritized bit rate (PBR), and bucket size duration (BSD).

[0084] The UE can maintain a variable Bj for each logical channel j. Bj can be initialized to zero when the associated logical channel is established and incremented by PBR*TTI every transmission time interval (TTI), where PBR is the priority bit rate of logical channel j. However, the value of Bj cannot exceed the bucket size, and if the value of Bj is greater than the bucket size of logical channel j, it is set to the bucket size. The bucket size of a logical channel is equal to PBR×BSD, where PBR and BSD are configured by the upper layer.

[0085] 5. Data reporting for artificial intelligence (AI) models

[0086] AI model data includes monitoring data, training data, and inference data. Model training includes base station-side training or terminal-side training.

[0087] Typically, during training at the base station, the terminal needs to report the monitoring data according to a fixed QoS. If the base station collects training data from multiple terminals, training will begin at a unified time after the collection is completed. The collection of training data is a relatively long process, and each terminal needs to report multiple times and continuously. The training data will also be reported according to a fixed QoS.

[0088] For example, in federated learning, multiple terminals independently train using local data and then report the weights or gradients of their local models to the base station. The base station aggregates the model weights or gradients from multiple terminals and then sends the aggregated model weights or gradients to each terminal. This process is repeated until the model converges.

[0089] For distributed inference AI models, there are no sidelinks between multiple terminals. The AI ​​model needs to perform inference across multiple terminals. For example, the first part of the AI ​​model is deployed on the first terminal. After the first terminal completes inference, it sends the first intermediate result to the base station. The base station sends the first intermediate result to the second terminal. The second terminal performs inference based on the first intermediate result and sends the second intermediate result to the base station. The base station sends the second intermediate result to the base station. The base station sends the second intermediate result to the third terminal. The third terminal performs inference based on the second intermediate result and obtains the final model output. Considering the energy consumption of inference and the fact that the terminal may perform other tasks, the terminal performs inference at fixed times (hereinafter referred to as the inference time). Therefore, it is necessary to report inference data before fixed time nodes (for example, the first terminal needs to report the inference result before the second terminal's inference time). In addition, the inference processing capabilities of different terminals are different, so the inference time of different terminals is also different.

[0090] 6. Activation Model

[0091] On the terminal side, models can be divided into active and inactive models. Active models are the models currently in actual use, while inactive models are models that have not yet been used. For example, when the model on the terminal side is a channel state information (CSI) prediction model, the CSI prediction results of the active model can be used in the actual communication system to determine schemes such as precoding, modulation and coding scheme (MCS), while the CSI prediction results of the inactive model are not used in the actual communication system. The terminal obtains the output results of the inactive model to monitor the inactive model so that the terminal can preliminarily determine the performance of the model before the inactive model is activated.

[0092] In a communication system, each type of uplink data corresponds to a fixed logical channel parameter configuration. Each logical channel parameter configuration can be adapted to a QoS stream. The parameters of the logical channel can be configured by the base station. However, the QoS stream corresponding to the uplink data may change (for example, when training an AI model, the parameters corresponding to the model may become abnormal, requiring accelerated reporting). If the previous logical channel parameter configuration is still used, it cannot match the changed QoS stream, and the data may not be transmitted within the scheduled time. Therefore, the parameters of the logical channel need to be updated. Existing communication systems cannot determine whether the logical channel needs to be updated based on changes in QoS.

[0093] In order to determine whether a logical channel needs to be updated when data QoS changes, at least one embodiment of this application provides a configuration update method.

[0094] Figure 2 This illustration shows a schematic flowchart of a configuration update method provided in at least one embodiment of this application, applied to a terminal device, such as a terminal or a communication module in a terminal, or a circuit or chip in a terminal responsible for communication functions (such as a modem chip, also known as a baseband chip, or a system-on-chip (SoC) chip or system-in-package (SIP) chip containing a modem core). Taking the application of this method to a terminal as an example, as shown in Figure 3, the method includes steps S201-S202:

[0095] Step S201: Obtain the monitoring data corresponding to the first model.

[0096] For example, the first model can be an artificial intelligence (AI) model, a neural network model, a big data model, etc., and the monitoring data can include the performance data, distribution data, and time data corresponding to the first model.

[0097] Step S202: Determine whether to update the configuration information of the first channel based on the changes in the monitoring data, wherein the first channel is used to transmit all or part of the first data, and the first data includes at least one of monitoring data, training data of the first model, and inference data of the first model.

[0098] For example, the first channel can be a logical channel as described above.

[0099] Exemplarily, the configuration information may include a bit rate and a priority. The bit rate may be, for example, the PBR as described above, or other parameters representing the data-carrying capacity of the channel, which is not limited in this application. The priority may be used to represent the priority of mapping a logical channel to a physical layer channel or the priority of transmitting the data carried by the logical channel.

[0100] Exemplarily, the monitoring data may further include performance data, distribution data, and time data corresponding to the first model. The distribution data may include, for example, the distribution of input data and output data. When the terminal monitors the first model and detects one or more of the following situations, it updates the configuration information of the first channel: the performance data meets the first performance threshold within the first observation time; the distribution data is abnormal within the second observation time; the time data meets the first time threshold.

[0101] The process of triggering the update of the configuration information of the first channel in the above three scenarios will be described in detail below.

[0102] Scenario 1: The performance data meets the first performance threshold within the first observation time.

[0103] In Scenario 1, the performance data may include, for example, communication metrics or model accuracy (Accuracy). The communication metrics may be, for example, throughput or signal to interference plus noise ratio (SINR) or reference signal received power (RSRP). The communication metrics may be obtained by the terminal measuring the system performance or link quality when the first model is activated. When the above communication metrics change significantly within a period of time, the configuration information of the first channel is updated.

[0104] In a possible implementation, when the performance data is less than or equal to the first performance threshold within the observation time, or when the performance data is greater than or equal to the first performance threshold within the observation time, the configuration information of the first channel is updated. Below, an example of updating the configuration information of the first channel when the performance data is less than or equal to the first performance threshold within the observation time is used for illustration.

[0105] The first performance threshold is the difference between the base sub-threshold and the dynamic sub-threshold, that is, Mea s <Thre1 - offset, where Mea s represents the measured value of the performance data, Thre1 is the base sub-threshold, and offset is the dynamic sub-threshold.

[0106] It should be noted that the basic sub-threshold represents the threshold corresponding to the measured value. Different performance data correspond to different measurement thresholds. The offset is used to control the dynamic range of the threshold (the offset can also be regarded as the dwell value of the threshold). It is used to ensure that the first channel configuration information is updated only after the performance data drops to a certain level, so as to avoid frequent triggering of configuration information updates when the performance data switches back and forth around the threshold, thereby reducing the frequency of configuration information updates and reducing power consumption.

[0107] Furthermore, the basic sub-threshold and the dynamic sub-threshold can also represent two different levels of the first channel configuration information update stage. For example, when the performance data meets the basic sub-threshold, the first channel configuration information is updated for the first time. When the performance data meets the difference between the basic sub-threshold and the dynamic sub-threshold, the first channel configuration information is updated for the second time. The change in configuration information corresponding to the first update can, for example, be smaller than the change in configuration information corresponding to the second update.

[0108] Figure 3A and Figure 3B This is a schematic diagram of performance data monitoring shown in at least one embodiment of this application. The following is in conjunction with... Figure 3A and Figure 3B The process of determining the update of the first channel configuration information in Scenario 1 is explained in detail.

[0109] like Figure 3A As shown, when the detected performance data is lower than Thre1-offset within the observation time (Td in the figure) (i.e., the performance data is abnormal), the update of the first channel configuration information is triggered. The dynamic sub-threshold offset and the observation time Td can ensure that the system performance is lower than a certain level and persists for a period of time before the update of the first channel configuration information is triggered, ensuring the reliability of the triggering condition. Alternatively, the update of the first channel configuration information can also be triggered if the performance data is lower than Thre1-offset within a first time range. Here, the first time range is within the observation time, and the proportion of the first time range within the observation time is greater than or equal to a specific threshold (e.g., 80%). The first time range can be a continuous time range or a discontinuous time range.

[0110] like Figure 3B As shown, the observation time can satisfy periodicity, that is, the observation as described above is periodic observation. In one possible implementation, the performance data satisfies the first performance threshold within a first number of observation periods (e.g., within two observation window time ranges); this application does not impose any restrictions here. Figure 3BIn this application, the observation time for performance data is divided into multiple windows of equal duration. Some windows are fixed observation windows (the first and second observation windows in the figure). A certain number (one in the figure) of candidate observation windows are spaced between sequentially adjacent fixed observation windows. Candidate observation windows may or may not be observed. For example, if an observation was performed in a window preceding a candidate observation window, and the result shows that the performance data meets a first performance threshold, then that candidate observation window is observed; otherwise, it is not observed. Furthermore, the time ranges of the fixed observation windows and candidate observation windows can be set to different values, different time windows can overlap in time, the time lengths of observation windows of the same type can differ, and the number of candidate observation windows spaced between sequentially adjacent fixed observation windows can also differ. This application does not impose any limitations on these aspects. By setting periodic observation windows, this application can reduce observation time and power consumption.

[0111] In one possible implementation, in this scenario, updating the information reported by the terminal can be triggered simultaneously with determining that the configuration information for the first channel needs to be updated.

[0112] For example, before triggering the update of the reported information content, the terminal reports communication indicators. After triggering the update of the reported information content, the terminal simultaneously reports the communication indicators and the inference data of the first model, such as the input data and output data of the first model.

[0113] For another example, before triggering the update of the reported information content, the terminal reports the model accuracy. After triggering the update of the reported information content, the terminal simultaneously reports the model accuracy, the input data and output data of the first model, and the model's true result (ground truth).

[0114] The input data, output data, and / or actual results of the first model can be, for example, data that causes anomalies in the current performance data. This allows the network device receiving the data (e.g., the base station) to update the data in a timely manner based on the data that causes the anomalies. For example, this can be done by aligning the differences between the training data and the inference data, updating the training data, or performing corresponding control operations (e.g., switching, updating, or rolling back the first model), thus ensuring the effective management of the first model's lifecycle.

[0115] In one possible example, when the accuracy of the first model meets the first performance threshold, the first channel configuration information can also be updated based on the accuracy of the second model. The first model can be, for example, the active model on the terminal side, and the second model can be, for example, the inactive model on the terminal side.

[0116] Figure 3C This is another performance data monitoring diagram shown in at least one embodiment of the present application, such as... Figure 3C As shown, when the accuracy of the first model decreases (the solid line in the figure), while the accuracy of the second model increases (the dashed line in the figure), or the accuracy of the first model satisfies Accuracy ≤ Thre1 - offset within the observation time, and the accuracy of the second model satisfies Accuracy > Thre1 + offset within the observation time, or the accuracy of the first model satisfies Accuracy < Thre1 - offset within the observation time, and the accuracy of the second model satisfies Accuracy ≥ Thre1 + offset within the observation time, that is, the accuracy of the active model is poor and the accuracy of the non - active model is good, then the update of the logical channel parameters is triggered, and the updated logical channel parameters can be set according to the non - active model, for example.

[0117] Scenario 2: The distribution of the distributed data is abnormal within the second observation time, for example, the distribution of the input data and the output data is abnormal.

[0118] In Scenario 2, when the terminal detects one or more of the following situations in the first model, it can be considered that the distribution of the input data and the output data is abnormal: The samples of the input data and the output data do not conform to the distribution of the historical input data and the output data; The samples of the input data and the output data do not conform to the distribution of the training input data and the output data; The drift of the input data and the output data within the observation time is greater than or equal to the first drift threshold. For example, the K - S (Kolmogorov - Smirnov) test can be used to determine whether the current input data or output data conforms to the distribution of the historical input data and output data collected by the terminal, or whether the current input data or output data conforms to the distribution of the training input data and output data; Another example is that the drift of the input data or output data within the observation time is greater than a specific threshold (such as the numerical threshold of the population stability index (PSI)).

[0119] The observation time in this scenario can be the same as that in Scenario 1, for example. In the scenario of periodic observation, if the distribution of the input data and the output data is abnormal within the second number of observation cycles (such as within two observation window times), then the terminal triggers the update of the first channel configuration information. The rest of the information related to the observation time can refer to Scenario 1, which will not be elaborated herein.

[0120] In a possible implementation, when triggering the update of the first channel configuration information in this scenario, the update of the content of the terminal reporting information can also be triggered. The specific update of the reported data content can refer to Scenario 1, which will not be elaborated herein.

[0121] It should be noted that the events that trigger the update of the first channel configuration information in Scenario 1 and Scenario 2 can occur simultaneously. For example, if the terminal detects anomalies in the distribution of input data of communication indicators and models at the same time, the first channel configuration information can be updated according to the priority of the monitoring data.

[0122] For example, when the terminal detects an abnormality in the communication metrics, the priority value of the event is 1, triggering a first update to the configuration information of the first channel. When the terminal detects an abnormality in the distribution of the model's input and output data, the priority value of the event is 2, triggering a second update to the configuration information of the first channel. The lower the priority value of the event, the higher the priority. Thus, when the terminal detects an abnormality in both the communication metrics and the distribution of the model's input data, it triggers a first update to the configuration information of the first channel.

[0123] In one possible implementation, in scenarios one and two of this application, after updating the first channel configuration information, the first channel configuration information can also be restored to make it the same as before the update. For example, in response to the performance data not meeting the first performance threshold during the third observation time, the configuration information is restored to the state before the update; or, in response to the performance data meeting the second performance threshold during the third observation time, the configuration information is restored to the state before the update; and / or, in response to the normal distribution of input and output data during the fourth observation time, the configuration information is restored to the state before the update; wherein the third observation time is after the first observation time, and the fourth observation time is after the second observation time.

[0124] For example, Figure 3D This is another performance data monitoring diagram shown in at least one embodiment of the present application, such as... Figure 3D As shown, after the terminal triggers the update of the first channel configuration information based on the abnormality of the performance data, when the terminal detects that the performance data is greater than Thre1+offset within the observation time Td, it triggers the restoration of the first channel configuration information. For example, the relevant information of the observation time and threshold can be referred to Scenario 1, which will not be repeated here.

[0125] For example, after the terminal triggers the update of the first channel configuration information based on the abnormality of the performance data, when the terminal detects that the performance data is greater than Thre1-offset within the observation time Td, it triggers the restoration of the first channel configuration information.

[0126] For example, after the terminal triggers the update of the first channel configuration information based on the anomaly of the input data and output data, when the terminal detects that the input data and output data are normally distributed within the observation time Td, it triggers the restoration of the first channel configuration information.

[0127] Scenario 3: The time data meets the first time threshold.

[0128] In Scenario 3, for example, when a terminal reports training data nears the reporting termination point, it needs to report the training data as quickly as possible. If the reporting is done according to the original configuration information of the first channel, the training data may not be reported in time or may not be reused in the current transmission resources. Another example is when the base station aggregates the weights or gradients reported by multiple terminals. Terminals with weaker training capabilities require more local training time, so they may only report near the model aggregation time point. Alternatively, if a terminal has a large amount of data during a local training session, it may report the local training data later, and the reported data volume may be large. Therefore, the configuration information of the first channel needs to be updated to accelerate the reporting and complete the reporting before the model aggregation time point. Yet another example is when each terminal involved in the model needs to report inference data before the inference time of its corresponding next terminal. Due to the different inference capabilities of different terminals, terminals with weaker inference capabilities require longer inference time. The current configuration parameters of the first channel cannot guarantee that the current terminal will report inference data before the inference time of the next terminal. Therefore, the configuration information of the first channel needs to be updated.

[0129] For example, time data could be the difference between the end of the training data collection for the first model and the end of the training data transmission; or the difference between the end of the aggregation for the first model and the end of the model data transmission; or the difference between the end of the inference data collection for the first model and the end of the inference data transmission.

[0130] The first time threshold could be, for example, the transmission delay of data reported by the terminal. When the time threshold is less than the transmission delay, an update to the first channel configuration information is triggered. The terminal can receive configuration information indicating the transmission delay through downlink control information (DCI), RRC, or broadcast messages.

[0131] Figure 4A , Figure 4B as well as Figure 4C This is a schematic diagram of terminal data reporting time shown in at least one embodiment of this application. The following is in conjunction with... Figure 4A , Figure 4B as well as Figure 4C This application explains whether the time data in Scenario 3 meets the first time threshold.

[0132] For example, such as Figure 4AAs shown, Td1 is the end time of the first model training data collection, Tc1 is the terminal training data transmission time (i.e., the current time), and Ed1 is the expected delay of the training data transmission (i.e., the first time threshold). Td1 can be configured by the base station through DCI or RRC. Since Td1 can be the end time of the overhead chain data collection for multiple terminals, the base station can also indicate it through broadcast. Ed1 can be configured through RRC to indicate the average delay between two transmission resources.

[0133] When Td1-Tc1<Ed1, meaning the remaining time for the terminal to report training data is less than the expected delay for training data transmission, the terminal cannot complete the reporting of training data before the end of the first model training data collection time, or the data cannot be reused in the current transmission resources, thus triggering an update of the first channel configuration information to speed up the transmission of training data.

[0134] For example, such as Figure 4B As shown, the base station needs to aggregate the weights or gradients (i.e. model data) reported by the first terminal UE1 and the second terminal UE2. Before the end of the first model data transmission, both UE1 and UE2 promptly reuse the model data with other data and report it (not shown in the figure). Td2 is the end time of the model data transmission corresponding to the aggregation of the first model, Tc2 is the model data transmission time corresponding to UE2 (i.e., the current time), and Ed2 is the expected delay of model data transmission (i.e., the first time threshold). The configuration of Td2 and Ed2 can refer to Td1 and Ed1 mentioned above, and will not be repeated here.

[0135] Before the end of the second model data transmission, UE1 completed the reporting of model data before Td2. However, when Td2-Tc2 < Ed2, that is, the remaining time for UE2 to report model data is less than the expected delay of model data transmission, UE2 cannot complete the reporting of model data before the end of the model data transmission for the first model aggregation. The first channel parameters corresponding to the model weights or gradients of UE2 cannot guarantee that the model data of UE2 will be sent before Td2. Furthermore, the next transmission resource of UE2 may be located after UE2, which triggers the update of the first channel configuration information to speed up the transmission of model data.

[0136] For example, such as Figure 4CAs shown, the second terminal UE2 performs inference at a fixed time. The first terminal UE1 needs to send the intermediate results of the inference to the base station before the inference time of UE2. Td3 is the end time of the inference data collection of the first model, Ed3 is the expected delay of UE1's inference data transmission (i.e., the first time threshold), Tc3 is the time of UE1's inference data transmission (i.e., the current time), and En is the expected delay of the base station sending the inference data of UE1 to UE2. Td3 can be configured in the same way as Td1, for example, by directly sending Ed3, or by determining it through the inference time of UE2 and En.

[0137] When Td3-Tc < Ed3, meaning the remaining time for UE1 to report inference data is less than the expected delay for inference data transmission, the first channel parameters corresponding to the inference result of UE1 cannot guarantee that the transmission of the inference result will be completed before Td3. In other words, UE1 cannot complete the uploading of inference data before the end of the inference data collection of the first model, and the next transmission resource of UE1 may be located after Td3. Therefore, the update of the first channel configuration information is triggered to speed up the transmission of inference data.

[0138] In this embodiment, by monitoring the model parameters, it is determined whether changes in the parameters corresponding to the model require updating the channel configuration information for transmitting the model parameters, so as to ensure the compatibility between the channel parameters and the corresponding model parameters and improve the effectiveness of model lifecycle management.

[0139] Figure 5 This application provides a schematic diagram of a communication system structure for at least one embodiment, as shown below. Figure 5 As shown, the communication system 50 includes a terminal 51 and a network device 52. The terminal 51 is configured to perform the terminal function in any of the configuration update methods described above. The network device 52 is configured to perform the network device function in any of the configuration update methods described above. For example, the network device 52 may be a base station.

[0140] Figure 6 This is a schematic diagram of a communication device provided in this application. This communication device can be used to implement any possible function in the above method embodiments, and therefore can also achieve the beneficial effects of the above method embodiments.

[0141] like Figure 6 As shown, the communication device 600 includes a processing unit 610.

[0142] In one possible implementation, the communication device 600 may also include a transceiver unit 620.

[0143] In one possible implementation, the communication device 600 may also include a storage unit 630.

[0144] In one possible implementation, the communication device 600 may further include a transceiver unit 620 and a storage unit 630.

[0145] In this embodiment of the application, the communication device 600 may be Figure 5 The terminal 51 shown can also be a module (such as a chip) applied to the terminal 51, or the communication device 600 can be... Figure 5 The network device 52 shown can also be a module (such as a chip) applied to the network device 52.

[0146] When the communication device 600 is used to implement the functions of a terminal as described above, the transceiver unit 620 is used to acquire monitoring data corresponding to the first model. The processing unit 610 is used to determine whether to update the configuration information of the first channel based on changes in the monitoring data, wherein the first channel is used to transmit all or part of the first data, and the first data includes at least one of monitoring data, training data of the first model, and inference data of the first model. The storage unit 630 is used to store any data, computer instructions, and / or computer programs that may be involved in the various embodiments of this application. A more detailed description of the above processing unit 610 and transceiver unit 620 can be found in [reference]. Figure 2 The relevant descriptions in the method embodiments shown.

[0147] In one possible implementation, the monitoring data includes performance data, distribution data, and time data; the processing unit is further configured to: update the configuration information of the first channel in response to the performance data meeting a first performance threshold within a first observation period; and / or, update the configuration information of the first channel in response to an anomaly in the distribution data within a second observation period; and / or, update the configuration information of the first channel in response to the time data meeting a first time threshold.

[0148] In one possible implementation, the observation is periodic; the performance data meets a first performance threshold within a first observation period, including: the performance data meets the first performance threshold within a first number of observation periods; the distribution data shows anomalies within a second observation period, including: the distribution of the distribution data shows anomalies within a second number of observation periods.

[0149] In one possible implementation, the configuration information includes bit rate and priority.

[0150] In one possible implementation, the performance data satisfies a first performance threshold, including: the performance data is less than or equal to the first performance threshold during the observation period.

[0151] In one possible implementation, the first performance threshold is the difference between the basic sub-threshold and the dynamic sub-threshold.

[0152] In one possible implementation, performance data includes model accuracy or communication metrics.

[0153] In one possible implementation, the distribution data includes the distribution of input data and output data; anomalies in the distribution data include: the samples of input data and output data do not conform to the distribution of historical input data and output data; and / or, the samples of input data and output data do not conform to the distribution of training input data and output data; and / or, the drift of input data and output data during the observation time is greater than or equal to a first drift threshold.

[0154] In one possible implementation, the processing unit is further configured to: restore the configuration information to its previous state in response to the performance data not meeting the first performance threshold during the third observation time; or, restore the configuration information to its previous state in response to the performance data meeting the second performance threshold during the third observation time; and / or, restore the configuration information to its previous state in response to the distribution data being normal during the fourth observation time; wherein the third observation time is after the first observation time, and the fourth observation time is after the second observation time.

[0155] In one possible implementation, the processing unit is also used to update configuration information according to the priority of the monitoring data.

[0156] In one possible implementation, the time data includes: the difference between the end time of training data collection and the end time of training data transmission for the first model, and / or, the difference between the end time of aggregation and the end time of model data transmission for the first model, and / or, the difference between the end time of inference data collection and the end time of inference data transmission for the first model.

[0157] In one possible implementation, the first time threshold is the transmission delay; the time data satisfies the first time threshold, including: the time data is less than the transmission delay.

[0158] In one possible implementation, the transceiver unit is further configured to: receive configuration information corresponding to the transmission delay, wherein the configuration information indicates the magnitude of the transmission delay.

[0159] Optionally, the transceiver unit 620 may be a transceiver, which may include an antenna and radio frequency circuitry, etc.

[0160] The processing unit 610 may be a processor (or processing circuitry), such as a baseband processor, which may include one or more CPUs.

[0161] Figure 7 This is a schematic diagram of a communication device provided in this application. This communication device can be used to implement any possible function in the above method embodiments, and therefore can also achieve the beneficial effects of the above method embodiments.

[0162] like Figure 7 As shown, the communication device 700 includes at least one processor 710. In one possible implementation, the communication device 700 may further include interface circuitry 720.

[0163] In one possible implementation, the communication device 700 may also include a memory 730.

[0164] In one possible implementation, the communication device 700 may also include a memory 730 and an interface circuit 720.

[0165] In some embodiments, the processor 710 and the memory 730 are coupled to each other; and / or, the processor 710 and the interface circuit 720 are coupled to each other. It is understood that the interface circuit 720 can be a transceiver or an input / output interface. The memory 730 can be used to store computer instructions executed by the processor 710, or to store input data required by the processor 710 to execute computer instructions, or to store data generated by the processor 710 after executing computer instructions.

[0166] Figure 6 and Figure 7 The communication device shown is merely an example, and in practical applications, the communication device can have more than... Figure 6 and Figure 7 The more or fewer components shown can be combined into two or more components, or they can have different component configurations, and Figure 6 In this context, the processing unit can also be called a processing module or processor; the transceiver unit can also be called a transceiver module or transceiver; and the storage unit can also be called a storage module or memory.

[0167] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.

[0168] The method steps in this application embodiment can be implemented in hardware or in software instructions executable by a processor. The software instructions can consist of corresponding software modules, which can be stored in random access memory, flash memory, read-only memory, programmable read-only memory, erasable programmable read-only memory, electrically erasable programmable read-only memory, registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. The storage medium can also be a component of the processor. The processor and storage medium can reside in an ASIC. Alternatively, the ASIC can reside in a network device or terminal. The processor and storage medium can also exist as discrete components in a network device or terminal.

[0169] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this application are performed entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video optical disc; or it can be a semiconductor medium, such as a solid-state drive. The computer-readable storage medium may be a volatile or non-volatile storage medium, or may include both types of storage media.

[0170] In the various embodiments of this application, unless otherwise specified or logically conflicting, the terms and / or descriptions between different embodiments are consistent and can be referenced by each other. The technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0171] It is understood that the various numbers used in the embodiments of this application are merely for ease of description and are not intended to limit the scope of the embodiments of this application. The order of the sequence numbers of the above-mentioned processes does not necessarily imply a specific order of execution; the order of execution of the processes should be determined by their functions and inherent logic.

[0172] It is understood that descriptive terms used in the embodiments of this application, such as: in response to, when, if, under the circumstances, can be used interchangeably.

Claims

1. A configuration update method, characterized in that, include: Obtain the monitoring data corresponding to the first model; Whether to update the configuration information of the first channel is determined based on the changes in the monitoring data, wherein the first channel is used to transmit all or part of the first data, and the first data includes at least one of the monitoring data, the training data of the first model, and the inference data of the first model.

2. The method according to claim 1, characterized in that, The monitoring data includes performance data, distribution data, and time data; The step of determining whether to update the configuration information of the first channel based on changes in the monitoring data includes: updating the configuration information of the first channel in response to the performance data meeting a first performance threshold within a first observation period; and / or, In response to an anomaly in the distribution data during the second observation period, the configuration information of the first channel is updated; and / or, In response to the time data meeting a first time threshold, the configuration information of the first channel is updated.

3. The method according to claim 2, characterized in that, The observations were periodic. The condition that the performance data meets the first performance threshold within the first observation time includes: the performance data meets the first performance threshold within a first number of observation periods; The abnormality of the distribution data during the second observation period includes: the abnormality of the distribution data within the second number of observation periods.

4. The method according to claim 1, characterized in that, The configuration information includes bit rate and priority.

5. The method according to claim 2, characterized in that, The performance data satisfies the first performance threshold, including: The performance data is less than or equal to the first performance threshold during the observation period.

6. The method according to claim 5, characterized in that, The first performance threshold is the difference between the basic sub-threshold and the dynamic sub-threshold.

7. The method according to claim 2, characterized in that, The performance data includes model accuracy or communication metrics.

8. The method according to claim 2, characterized in that, The distributed data includes the distribution of input data and output data; The distribution data showed anomalies, including: The samples of the input and output data do not conform to the distribution of historical input and output data; and / or, The samples of the input and output data do not conform to the distribution of the training input and output data; and / or, The drift of the input data and output data during the observation time is greater than or equal to a first drift threshold.

9. The method according to claim 2, characterized in that, The method further includes: In response to the performance data not meeting the first performance threshold during the third observation period, the configuration information is restored to its state before the update; or, In response to the performance data meeting the second performance threshold within the third observation period, the configuration information is restored to its state before the update; and / or, In response to the distribution data being normal during the fourth observation period, the configuration information is restored to its state before the update. The third observation time is after the first observation time, and the fourth observation time is after the second observation time.

10. The method according to claim 2, characterized in that, The method further includes: The configuration information is updated according to the priority of the monitoring data.

11. The method according to claim 2, characterized in that, The time data includes: The difference between the end time of training data collection and the end time of training data transmission for the first model, and / or, The difference between the aggregation end time and the model data transmission time of the first model, and / or, The difference between the end time of inference data collection and the end time of inference data transmission for the first model.

12. The method according to claim 2, characterized in that, The first time threshold is the transmission delay; The time data satisfies the first time threshold, including: the time data is less than the transmission delay.

13. The method according to claim 12, characterized in that, The method further includes: Receive the configuration information of the transmission delay, wherein the configuration information indicates the magnitude of the transmission delay.

14. A configuration update device, characterized in that, include: The transceiver unit is used to acquire monitoring data corresponding to the first model; The processing unit is configured to determine whether to update the configuration information of the first channel based on changes in the monitoring data, wherein the first channel is used to transmit all or part of the first data, and the first data includes at least one of the monitoring data, the training data of the first model, and the inference data of the first model.

15. The apparatus according to claim 14, characterized in that, The monitoring data includes performance data, distribution data, and time data; The processing unit is further configured to: update the configuration information of the first channel in response to the performance data meeting a first performance threshold within a first observation time; and / or, In response to an anomaly in the distribution data during the second observation period, the configuration information of the first channel is updated; and / or, In response to the time data meeting a first time threshold, the configuration information of the first channel is updated.

16. The apparatus according to claim 15, characterized in that, The observations were periodic. The condition that the performance data meets the first performance threshold within the first observation time includes: the performance data meets the first performance threshold within a first number of observation periods; The abnormality of the distribution data during the second observation period includes: the abnormality of the distribution data within the second number of observation periods.

17. The apparatus according to claim 14, characterized in that, The configuration information includes bit rate and priority.

18. The apparatus according to claim 15, characterized in that, The performance data satisfies the first performance threshold, including: The performance data is less than or equal to the first performance threshold during the observation period.

19. The apparatus according to claim 18, characterized in that, The first performance threshold is the difference between the basic sub-threshold and the dynamic sub-threshold.

20. The apparatus according to claim 15, characterized in that, The performance data includes model accuracy or communication metrics.

21. The apparatus according to claim 15, characterized in that, The distributed data includes the distribution of input data and output data; The distribution data showed anomalies, including: The samples of the input and output data do not conform to the distribution of historical input and output data; and / or, The samples of the input and output data do not conform to the distribution of the training input and output data; and / or, The drift of the input data and output data during the observation time is greater than or equal to a first drift threshold.

22. The apparatus according to claim 15, characterized in that, The processing unit is also used for: In response to the performance data not meeting the first performance threshold during the third observation period, the configuration information is restored to its state before the update; or, In response to the performance data meeting the second performance threshold within the third observation period, the configuration information is restored to its state before the update; and / or, In response to the distribution data being normal during the fourth observation period, the configuration information is restored to its state before the update. The third observation time is after the first observation time, and the fourth observation time is after the second observation time.

23. The apparatus according to claim 15, characterized in that, The processing unit is also used for: The configuration information is updated according to the priority of the monitoring data.

24. The apparatus according to claim 15, characterized in that, The time data includes: The difference between the end time of training data collection and the end time of training data transmission for the first model, and / or, The difference between the aggregation end time and the model data transmission time of the first model, and / or, The difference between the end time of inference data collection and the end time of inference data transmission for the first model.

25. The apparatus according to claim 15, characterized in that, The first time threshold is the transmission delay; The time data satisfies the first time threshold, including: the time data is less than the transmission delay.

26. The apparatus according to claim 25, characterized in that, The transceiver unit is also used for: Receive configuration information corresponding to the transmission delay, wherein the configuration information indicates the magnitude of the transmission delay.

27. A communication device, comprising: At least one processor is configured to invoke the computer instructions in memory to cause the configuration update device to perform the method as described in any one of claims 1 to 13.

28. A computer-readable storage medium storing instructions or programs that, when executed on a cell selection device, implement the method as described in any one of claims 1-13.

29. A computer program product comprising a computer program or instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1-13.

Images