Resource management method and device of communication perception integrated system

Abstract

The invention discloses a resource management method and device for a communication perception integrated system, is applied to network equipment, and relates to the technical field of wireless communication. Comprising the following steps: distributing a perception artificial intelligence model training request to a perception node, and receiving a model gradient vector reported by the perception node after training based on local perception data; performing model aggregation based on gradient vectors to obtain a global perception artificial intelligence model; deploying the global sensing artificial intelligence model to a corresponding sensing node; a deployed global perception artificial intelligence model is used for prediction, and a perception resource management decision is generated; initiating a sensing resource allocation or reconfiguration operation according to the decision; and obtaining the perceptual performance data after the operation to verify whether the service quality meets a preset standard. Dynamic intelligent distribution and optimization of the sensing resources are realized through an artificial intelligence technology, the method can adapt to changes of high-dynamic sensing scenes, the utilization efficiency of the sensing resources and the performance of a sensing system are effectively improved, and the service quality of sensing services is ensured.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and more specifically, to a resource management method and apparatus for an integrated communication and sensing system. Background Technology

[0002] Integrated Communication and Sensing (ISAC) is a key technology for 5G-Advanced and 6G networks. By integrating communication and sensing functions on a unified platform, it can meet the application needs of scenarios such as vehicle-to-everything (V2X) and low-altitude economy. However, existing ISAC systems mostly adopt fixed sensing resource configuration and allocation strategies, making it difficult to respond in real time to dynamic changes such as sensing signal fluctuations caused by the high-speed movement of sensing targets and uneven spatiotemporal distribution of traffic. In highly dynamic scenarios, the movement of targets or transceivers can cause drastic fluctuations in sensing signals, significantly degrading sensing performance under fixed resource configuration. At the same time, the uneven distribution of sensing services in the spatiotemporal domain makes fixed resource configuration prone to resource allocation failures during peak traffic periods, while resources remain idle during off-peak periods. The system lacks flexibility, and the overall sensing resource utilization efficiency and sensing accuracy are difficult to guarantee. Summary of the Invention

[0003] The purpose of this invention is to provide a resource management method and apparatus for an integrated communication and sensing system to improve the aforementioned problems. To achieve this objective, the technical solution adopted by this invention is as follows: Firstly, this application provides a resource management method for an integrated communication and sensing system, applied to network devices, comprising: Distribute training requests for perception artificial intelligence (AI) models to one or more perception nodes, and receive model gradient vectors reported by one or more perception nodes after training based on local perception data; The perceptual artificial intelligence model is aggregated based on the gradient vector of the model to obtain a global perceptual artificial intelligence model; Deploy the global perception AI model to the corresponding perception nodes; Receive sensing information, use the deployed global sensing artificial intelligence model to predict the sensing information, and generate sensing resource management decisions; Initiate sensing resource allocation or reconfiguration operations based on the sensing resource management decision; Obtain perception performance data after perception resource allocation or reconfiguration operations to verify that the service quality of perception services meets preset standards.

[0004] Secondly, this application also provides a resource management device for an integrated communication and sensing system, comprising: The training request distribution module is used to distribute training requests for the perception artificial intelligence model to one or more perception nodes, and to receive model gradient vectors reported by multiple perception nodes after training based on local perception data. The model aggregation module is used to aggregate the perceptual artificial intelligence model based on the model gradient vector to obtain a global perceptual artificial intelligence model. The model deployment module is used to deploy the global perception artificial intelligence model to the corresponding perception nodes; The decision generation module is used to receive sensing information, use the deployed global sensing artificial intelligence model to predict the sensing information, and generate sensing resource management decisions. The operation execution module is used to initiate sensing resource allocation or reconfiguration operations based on the sensing resource management decision. The performance verification module is used to obtain perception performance data after perception resource allocation or reconfiguration operations, in order to verify that the service quality of perception services meets preset standards.

[0005] Thirdly, this application also provides a resource management device for an integrated communication and sensing system, comprising: Memory, used to store computer programs; A processor, used to execute the computer program, implements the resource management method of the integrated communication and sensing system.

[0006] Fourthly, this application also provides a readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the resource management method based on the integrated communication and sensing system described above.

[0007] The beneficial effects of this invention are as follows: This invention can predict and dynamically adjust the configuration of sensing resources in real time based on the dynamic changes in the sensing scenario and the sensing volume, thereby keeping the sensing system in an optimized working state. This invention effectively overcomes the performance limitations of a fixed resource allocation model, significantly improving the accuracy of sensing measurements, the utilization efficiency of sensing resources, and the robustness of the entire ISAC system in highly dynamic environments, thus significantly improving the service quality and user experience of sensing services.

[0008] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of the resource management method of the integrated communication and sensing system described in this embodiment of the invention; Figure 2 This is a schematic diagram of perception resource allocation prediction based on perception AI in an embodiment of the present invention; Figure 3 This is a schematic diagram of perception resource reconfiguration prediction based on perception AI in an embodiment of the present invention; Figure 4 This is a schematic diagram of perception measurement result prediction based on perception AI in an embodiment of the present invention; Figure 5 This is a schematic diagram of the predicted sensing event A in an embodiment of the present invention; Figure 6 This is a schematic diagram of the predicted sensing event B in an embodiment of the present invention; Figure 7 This is a schematic diagram of the predicted sensing event C in an embodiment of the present invention; Figure 8 This is a schematic diagram of the predicted sensing event D in an embodiment of the present invention; Figure 9 This is a schematic diagram of the predicted sensing event E in an embodiment of the present invention; Figure 10 This is a schematic diagram of the resource management device structure of the integrated communication and sensing system described in this embodiment of the invention; Figure 11 This is a schematic diagram of the resource management method and device structure of the integrated communication and sensing system described in this embodiment of the invention. (Labels in the diagram:) 800. Resource management method and device for integrated communication and sensing system; 801. Processor; 802. Memory; 803. Multimedia component; 804. I / O interface; 805. Communication component. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0012] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0013] Example 1: like Figure 1 As shown, this embodiment provides a resource management method for an integrated communication and sensing system, which is executed by a network device, namely a sensing function (SF) network element or a radio access network (RAN) node, and the initiator of the sensing artificial intelligence model training request is the artificial intelligence model management function or the SF.

[0014] 1) When the AI ​​model management function initiates a model training request, it includes the following steps: S0. Receive a training request for a perceptual artificial intelligence model from the artificial intelligence model management function; Specifically, the training request for the perceptual AI model includes information such as the perceptual AI model identifier, model feature information, aggregation strategy, and model gradient vector reporting type. The aggregation strategy includes aggregation frequency, weighting method, and aggregation rounds.

[0015] 2) When a network device initiates a model request, the following steps are included: S0a. The perception function SF sends a model training request message to the artificial intelligence model management function. The request message includes perception service characteristics and perception service quality requirements information, specifically: the unique attribute requirements of the perception task itself, such as the type of target to be detected, the required accuracy, the real-time processing, the data throughput, and the scene complexity and QoS information, that is, the specific quantitative requirements of the perception service for service quality, including performance indicators such as real-time performance, accuracy, reliability and resource consumption. S0b. The perception function SF receives model information of the global perception artificial intelligence model returned by the artificial intelligence model management function, which matches the business characteristics and service quality requirements; Specifically, the AI ​​model management function matches the most suitable perception AI model from the perception AI model library; at the same time, it responds to perception AI model requests by sending the matched perception AI model back to SF, including the perception AI model and its number, model feature information, model training requirements, and deployment requirements.

[0016] Based on the above embodiments, this method further includes: S1. The perception function SF distributes a perception artificial intelligence model training request to one or more perception nodes, and receives the model gradient vector reported by the perception nodes after training based on local perception data. The perception nodes include terminals, RAN nodes and cloud computing centers. In this embodiment, SF sends a perception AI model training request to the terminal, RAN node and cloud computing center respectively, requesting to train the perception AI model; Specifically, when SF requests the training of the perception AI model from the terminal and the cloud computing center, the perception AI model training request message includes the perception AI model identifier, model feature information, and model gradient vector reporting type. The model feature information includes, for example, model structure, model training and deployment requirements, etc. The model gradient vector reporting type includes reporting type selection: periodic reporting or event reporting. Specifically, when SF requests the perception AI model training from the RAN node, the perception AI model training request message includes the perception AI model identifier, model feature information, model gradient vector reporting type, perception AI model aggregation strategy and permissions. The model feature information includes model training deployment requirements, model size, i.e. memory usage, computational complexity and latency. In this embodiment, the SF grants the RAN node permission to perform aggregation calculations on the model gradient vectors reported by the terminals under its jurisdiction, along with its weight information. Specifically, the SF grants the RAN node the weights to aggregate the model gradient vectors reported by the terminals, thereby accelerating model aggregation and reducing the network bandwidth, time, and computing resources consumed in transmitting model parameters, gradients, and other data.

[0017] Each sensing node trains the sensing AI model based on local sensing data according to the sensing AI model training request. If the training is successful, the corresponding model gradient vector is generated; if the training fails, the reason for the failure is generated. Each sensing node sends the model gradient vector or the reason for failure to the SF, wherein the model gradient vector sent by the RAN node is the aggregated model gradient vector.

[0018] Based on the above embodiments, this method further includes: S2. Aggregate the perceptual artificial intelligence model based on the model gradient vector to obtain a global perceptual artificial intelligence model; Specifically, SF performs a weighted average aggregation of the perception AI models reported by each perception node. The global perception AI model is a large model composed of multiple independent small models. When calling, the large model can be used alone or only the small models in the large model can be used. Preferably, this embodiment also includes evaluating the quality of the aggregated global perception AI model. If the quality of the global perception AI model meets the requirements, the global perception AI model is reported to the AI ​​model management function.

[0019] Based on the above embodiments, this method further includes: S3. Deploy the global perception artificial intelligence model to the corresponding perception nodes; Specifically, step S3 includes: S31. The perception function SF sends a perception AI model request to the AI ​​model management function to apply for a trained perception AI model that meets the requirements. The request carries perception business characteristics and perception business service quality requirements information. The perception business characteristics include scene complexity. The scene complexity is obtained by weighted scoring and quantification based on one or more environmental multidimensional features, such as target density, dynamic change rate, and average occlusion rate. S32. Receive model information of the globally perceptive artificial intelligence model returned by the artificial intelligence model management function, which matches the business characteristics and service quality requirements; Specifically, the AI ​​model management function selects the most suitable AI model from the AI ​​model library based on the characteristics of the perceived business and the QoS information of the perceived business, and replies to SF with a AI model request response message, carrying the AI ​​model identifier, model structure, model parameters, and model deployment requirements information, such as memory and computing power. S33. Based on the perceived service characteristics and the computing power of the target perceived node, determine the target perceived node from multiple perceived nodes; Specifically, if the real-time requirements of the sensing service are extremely high, such as vehicle collision avoidance, then the RAN node or the sensing node of the terminal that is close to the sensing data source will be used as the target node to reduce transmission latency.

[0020] If the perception business has extremely high accuracy requirements and the model is complex, or requires large-scale data fusion processing, then the cloud computing center will be used as the target node. If the scenario in which the perception business is located has a high degree of complexity, then the node with stronger computing power will be selected as the target node to ensure the inference speed of the complex model.

[0021] S34. Send a model configuration request to the target sensing node, the model configuration request carrying the model information, so that the target sensing node can complete the model loading.

[0022] Specifically, the SF sends a perception AI model configuration request message to the perception node, carrying the perception AI model identifier, model structure, model parameters, and model deployment requirements. The target perception node deploys the perception AI model according to the information carried in the perception AI model configuration request message, and then replies to the SF with a perception AI model configuration request response message. If the perception node does not meet the perception AI model deployment requirements, it includes the reason for the failure in the perception AI model configuration request response message.

[0023] Based on the above embodiments, this method further includes: S4. Receive sensing information, use the deployed global sensing artificial intelligence model to predict the sensing information, and generate sensing resource management decisions; Step S4 includes at least one of the following two application scenarios: First application scenario: Perception resource allocation based on perception AI. Please refer to [link / reference]. Figure 2 : S41a. The network device receives sensing information from a sensing service request or a newly established sensing service reported by a sensing node, including one or more of the following: sensing service type, sensing service quality requirements, sensing channel state information, etc. The sensing service type includes target detection, identification, localization, tracking, imaging, ranging, velocity measurement, angle measurement, and physical environment information sensing. It should be noted that when a sensing terminal switches from a source sensing node to a target sensing node, the switching sensing terminal and its sensing services are considered newly established sensing services by the target sensing node. S42a. Based on the sensing information of the newly established sensing service, the sensing service analysis model, the sensing target channel prediction model, and the sensing resource allocation algorithm model are matched from the global sensing artificial intelligence model. Specifically, based on the business type and service quality requirements, the most suitable perception business analysis model is matched, wherein the perception business analysis model is a global perception AI model or a sub-model contained therein; Specifically, based on environmental state information, a target perception channel prediction model is matched, wherein the target perception channel prediction model is a global perception AI model or a sub-model contained therein; Specifically, based on overall business needs, a matching perception resource allocation algorithm model is used.

[0024] S43a. Use the aforementioned sensing service analysis model to generate sensing service change prediction results for newly established sensing services, including the expected duration of the sensing service, the dynamic trend of service resource demand, and the service load prediction curve based on time series.

[0025] S44a. Use the aforementioned sensing target channel prediction model to generate environmental change prediction results for newly established sensing services, including motion trajectory prediction of sensing targets, time-varying characteristics prediction of sensing channels, and trend prediction of interference conditions. S45a. Input the prediction results of the changes in the sensing services and the prediction results of the changes in the environment into the sensing resource allocation algorithm model to generate the sensing resource decision for the allocation of the newly established sensing services; In this embodiment, the sensing resource allocation algorithm model performs comprehensive optimization calculations according to preset criteria such as priority-driven, quality of service assurance, and utility maximization. The final output is a sensing resource decision customized for the newly established sensing service, including but not limited to the allocated time and frequency resource blocks, sensing reference signal resource configuration, corresponding beamforming parameters, and dynamic resource allocation sequence based on time series.

[0026] Scenario 2: Perception resource reconfiguration prediction based on perception AI. Please refer to [link / reference]. Figure 3 : S41c. Obtain the status information of sensing services and their resources, specifically including: the list of currently active sensing services and their service characteristics, the sensing resources allocated to each sensing service, the historical and real-time performance data or sensing measurement results of each sensing service, or the overall usage status and remaining available resources of the sensing resource pool within the system. S42c. Based on the sensing service and its resource status information, match the sensing service analysis model, the sensing target channel prediction model, and the sensing resource allocation algorithm model from the global sensing artificial intelligence model; S43c. Input the sensing services and their resource status information into the sensing service analysis model to generate sensing service change prediction results. Specifically, the prediction results include the changing trend of the total amount of sensing services in the system, the distribution change prediction of each type of sensing service, the prediction results of system resource demand in the future period, or the dynamic changes in the distribution of service priorities. Preferred, such as Figure 4 As shown: Based on the perception service and its resource status information, M (M≥1) perception artificial intelligence models are matched from the global perception artificial intelligence model; preferably, the factors considered in the matching process include: the adaptability of each model to the current perception scenario, the historical prediction performance of each model, and the complementary characteristics between different models.

[0027] The M perception artificial intelligence models are used to process the perception measurement results in parallel to obtain M initial prediction results. Specifically, each model independently processes the input data based on its own algorithm characteristics and training data, and outputs the corresponding initial prediction result.

[0028] The M initial prediction results are weighted and aggregated according to the weights of each perception AI model to generate the final prediction perception result.

[0029] In this embodiment, the weights of each perceptual AI model are dynamically allocated by the perceptual AI model analysis module based on the matching degree between the model and the current scene. This weighted aggregation mechanism effectively balances the prediction outputs of different models, suppresses the influence of outliers, and thus yields a more stable and reliable final prediction result.

[0030] S44c. Input the sensing service and its resource status information into the sensing target channel prediction model to generate environmental change prediction results. Specifically, the prediction results include: the overall distribution and movement trend of sensing targets within the system coverage area, the spatiotemporal change prediction of system-level channel quality, the system-level prediction of interference distribution and change trends, and the dynamic change prediction of environmental complexity indicators.

[0031] S45c. Input the predicted results of the perceived service changes and the predicted results of the environmental changes into the perceived resource allocation algorithm model to generate a perceived resource reconfiguration scheme and perceived resource reconfiguration information. Specifically, the reconfiguration scheme includes triggering a RAN perceived resource reconfiguration event, triggering a specific service resource reconfiguration event, or triggering a sidechain perceived resource reconfiguration event.

[0032] Based on the above embodiments, this method further includes: S5. The network device initiates a sensing resource allocation or reconfiguration operation based on the sensing resource management decision; Here, this embodiment uses the predicted sensing event A as an example to illustrate the reconfiguration of sensing resources of the RAN node triggered by SF, such as... Figure 5 As shown: Step A.1: The SF collects sensing resource usage and sensing service performance data from one or more RAN nodes under its jurisdiction. It then uses its deployed global sensing AI model (matching sensing service analysis model, sensing target channel prediction model, etc.) to analyze and predict the collected information, obtaining the predicted changes in sensing resource demand. This generates a sensing resource management decision, namely, the SF updates the sensing resource configuration of the RAN nodes. The SF obtains new sensing reference signal resource configuration information for the RAN nodes, including a list of sensing reference signal resource sets. The list of sensing reference signal resource sets includes the number of sensing reference signal resource sets and their information. The sensing reference signal resource set information includes sensing reference signal resource bandwidth, comb size, resource set period, resource repetition factor, number of resource symbols, resource set start time and duration, and the list of sensing reference signal resources. The list of sensing reference signal resources includes the number of sensing reference signal resources, sensing reference signal resource set identifiers, and sensing reference signal resource identifiers. It should be noted that the SF can use the global sensing AI model for prediction in a periodic, non-periodic, or event-triggered manner.

[0033] It should be noted that SF can update the perception resource configuration of one or more RAN nodes. Taking updating the perception resource configuration of RAN node 1 as an example: Step A.2: SF sends a Sensing Reference Signal Resource Configuration Update Message to RAN Node 1, carrying the RAN node identifier, a list of Sensing Reference Signal Resource Sets, etc.

[0034] Step A.3a: After receiving the Sensing Reference Signal Resource Configuration Update message, RAN Node 1 saves and updates its local Sensing Reference Signal Resource Configuration and replies to SF with a Sensing Reference Signal Resource Configuration Update Confirmation Message. Step A.3b: If RAN Node 1 cannot accept the update, it replies to SF with a Sensing Reference Signal Resource Configuration Update Failure Message, carrying the reason for the failure.

[0035] Step A.4: After the sensing reference signal resource configuration of RAN Node 1 is updated, the SF updates the sensing reference signal resource configuration information of the sensing terminal. Specifically, the SF obtains the sensing terminal served by the corresponding RAN Node 1, and then the SF sends a Provide Sensing Assistance Data message to the sensing terminal, carrying the updated sensing reference signal resource configuration information. It should be noted that the Provide Sensing Assistance Data message is a sensing protocol message.

[0036] Step A.5: Optionally, when the sensing reference signal resource configuration information of RAN node 1 is updated, it can send a RAN node sensing reference signal resource configuration update message to the peer RAN node 2, carrying the updated sensing reference signal resource configuration information, including the sensing reference signal resource set list, etc.

[0037] Step A.6a: If RAN Node 2 accepts the update, it replies to RAN Node 1 with a RAN Node Aware Reference Signal Resource Configuration Update Confirmation Message. Step A.6b: If RAN Node 2 cannot accept the update, it replies to RAN Node 1 with a RAN Node Aware Reference Signal Resource Configuration Update Failure Message, carrying the reason for the failure.

[0038] Here, this embodiment uses the predicted sensing event B as an example to illustrate the reconfiguration of sensing resources of the RAN node triggered by the RAN node, such as... Figure 6 As shown: Step B.1: RAN Node 1 uses its locally deployed global perception AI model to predict changes in perception resource demand, generating a perception resource management decision, i.e., updating the RAN perception resource configuration. RAN Node 1 obtains new perception resource configuration information, including a list of perception reference signal resource sets, etc. It should be noted that RAN Node 1 can use the global perception AI model for prediction in a periodic, event-triggered, or hybrid manner. Event-triggered events can include the creation of new perception services, the release of perception services, or changes in the performance of perception service measurement results, etc.

[0039] Step B.2: RAN node 1 sends a RAN sensing reference signal resource configuration update message to SF, carrying the RAN node identifier, sensing reference signal resource set list, etc.

[0040] Step B.3a: After receiving the RAN-aware reference signal resource configuration update message, if the SF accepts the update, it replies to RAN node 1 with a RAN-aware reference signal resource configuration update confirmation message. Step B.3b: If the SF does not accept the update, it replies to RAN node 1 with a RAN-aware reference signal resource configuration update failure message, carrying the reason for the failure.

[0041] Step B.4: After the sensing reference signal resource configuration of RAN Node 1 is updated, the sensing reference signal resource configuration of the sensing terminal is updated. Implementation Method 1: As shown in Step B.4a, the SF obtains the sensing terminal served by the corresponding RAN Node 1, and then the SF sends a sensing assistance data message to the sensing terminal, carrying the updated sensing reference signal resource configuration information. It should be noted that the sensing assistance data is a sensing protocol message. Implementation Method 2: As shown in Steps B.4b.1 and B.4b.2, RAN Node 1 obtains the serving sensing terminal, and then RAN Node 1 sends a radio resource control reconfiguration message to the sensing terminal, carrying the updated sensing reference signal resource configuration information; after receiving the radio resource control reconfiguration message, the sensing terminal saves and locally updates the sensing reference signal resource configuration information.

[0042] Step B.5: Optionally, when the sensing reference signal resource configuration information of RAN Node 1 is updated, it can send a RAN Node Sensing Reference Signal Resource Configuration Update Message to its peer RAN Node 2, carrying the updated sensing resource configuration information. If RAN Node 2 accepts the update, it replies to RAN Node 1 with a RAN Node Sensing Reference Signal Resource Configuration Update Confirmation Message. If RAN Node 2 cannot accept the update, it replies to RAN Node 1 with a RAN Node Sensing Reference Signal Resource Configuration Update Failure Message, carrying the reason for the failure.

[0043] Here, this embodiment uses the predicted sensing event C as an example to illustrate the network-triggered sidechain sensing resource reconfiguration, such as... Figure 7 As shown: Step C.1: Network-side network elements use the perception AI model for prediction. When the prediction result indicates that the network needs to update the sidechain perception resource configuration, a list of sidechain perception reference signal resource sets is obtained, including the number of perception reference signal resource sets and their information. The perception reference signal resource set information includes the perception reference signal resource bandwidth, comb size, resource set period, resource repetition factor, number of resource symbols, resource set start time and duration, and the perception reference signal resource list. The perception reference signal resource list includes the number of perception reference signal resources, the perception reference signal resource set identifier, and the perception reference signal resource identifier.

[0044] Step C.2: Update the sidechain awareness resource configuration of the network update terminal.

[0045] Optionally, Method 1: As shown in step C.2a, the network sends a Sidechain Aware Resource Configuration System Information Block (SIB) to the terminal via broadcast system information. The SIB carries a sidechain awareness resource configuration including a list of sidechain awareness resource sets. The list of sidechain awareness resource sets includes the quantity, awareness reference signal resource identifiers, etc. The terminal receives the sidechain awareness resource configuration SIB, saves it, and updates it locally.

[0046] Optionally, Method Two: As shown in steps C.2b.1 to C.2b.2, if the terminal is not in a connected state, first execute the network-triggered service request process to establish a connection between the terminal and the network; then, the network carries the sidechain-aware resource configuration through a downlink non-access stratum transport (DLNAS TRANSPORT) message. The sidechain-aware resource configuration includes a list of sidechain-aware resource sets. The list of sidechain-aware resource sets includes the quantity, sensing reference signal resource identifier, etc.; the terminal receives the sidechain-aware resource configuration carried by the DLNAS TRANSPORT, saves it, and performs a local update.

[0047] Optionally, Method 3: As shown in steps C.2c.1 to C.2c.2, the network updates the terminal's sidechain-aware resource configuration through the Radio Resource Control (RRC) reconfiguration procedure. The network sends an RRC reconfiguration message to the terminal, carrying sidechain-aware resource configuration information, including a list of sidechain-aware reference signal resource sets, etc. After receiving the RRC reconfiguration message, the terminal saves and uses the sidechain-aware reference signal resource configuration information carried in the message to perform a local update, and replies to the network with an RRC reconfiguration complete message. If the RRC reconfiguration fails, the RRC reconfiguration complete message carries the reason for the failure.

[0048] It should be noted that for Method 1 and Method 2, if the sidechain sensing resource configuration is triggered by the SF network element on the network side, in the network-side processing flow, the SF sends a sensing reference signal configuration request message to the RAN node, carrying sidechain sensing reference signal resource configuration information, including a list of sidechain sensing reference signal resource sets. The list of sensing reference signal resource sets includes the quantity, sensing reference signal resource identifier, etc.; if the processing is successful, the RAN node replies to the SF with a sensing reference signal configuration response message; if the processing fails, the RAN node replies to the SF with a sensing reference signal configuration failure message, carrying the reason for the failure.

[0049] Here, this embodiment uses a predicted sensing event D as an example to illustrate the reconfiguration of sensing resources for network-triggered sensing services, such as... Figure 8 As shown: Step D.1: Prediction using the perception AI model. Based on perception measurement results and / or perception performance information, perception reference signal resources are reallocated from the configured perception reference signal resource set, including perception reference signal identification information, perception resource frequency layer information, and perception aggregation information. Specifically, perception reference signal identification information includes the number of perception reference signal identifiers, the list of perception reference signal resource identifiers, and the perception reference signal resource set identifier; perception resource frequency layer information includes subcarrier spacing, bandwidth, starting physical resource block (PRB), perception reference signal A point, comb size, and cyclic prefix (CP); perception aggregation information includes the number, perception frequency layer index, perception transceiver point (TRP) index, and perception reference signal resource set index. It should be noted that SF network elements or RAN nodes on the network side can trigger the reconfiguration of perception resources for the terminal's perception services.

[0050] Step D.2: When the RAN node triggers the reconfiguration of sensing resources for the sensing service, as shown in step D.2a, the RAN node sends a sensing resource reconfiguration message to the terminal, carrying the reassigned sensing reference signal resource information. Specifically, the sensing resource reconfiguration message sent by the RAN node can be a Medium Access Control Element (MAC CE) or a Radio Resource Control reconfiguration message. When the SF triggers the terminal's sensing resource reconfiguration, as shown in step D.2b, the SF sends a sensing assistance data message to the sensing terminal, carrying the reassigned sensing reference signal resource information.

[0051] Here, this embodiment uses the predicted sensing event E as an example to illustrate the network-triggered sensing power control, such as... Figure 9 As shown: Step E.1: Use the perception AI model to make a prediction, and obtain the prediction result as network-triggered perception power control. It should be noted that the network element that triggers perception power control can be an SF or RAN node.

[0052] It should be noted that there are two implementation methods for SF-triggered sensing power control: Implementation method one executes steps E.2 to E.6; Implementation method two executes steps E.7 to E.8. RAN node-triggered uplink sensing power control executes steps E.8 to E.9.

[0053] Step E.2: SF sends a Sensing Transmission Power Information Request message to the RAN node, requesting the current sensing transmission power of the RAN node.

[0054] Step E.3: If the RAN node successfully processes the Sensing Transmission Power Information Request message, it replies to the SF with a Sensing Transmission Power Information Request Response message, carrying the Sensing Transmission Power Information. If processing fails, it replies to the SF with a Sensing Transmission Power Information Failure message, carrying the reason for the failure.

[0055] Step E.4: SF calculates the new sensing transmit power based on the sensing measurement results, sensing QoS, current sensing transmit power, etc.

[0056] Step E.5: SF sends a sensed power control request message to the RAN node, carrying sensed transmit power information.

[0057] Step E.6: If the RAN node successfully processes the Sensing Power Control Request message, it replies with a Sensing Power Control Request Response message to the SF, and transmits the Sensing Reference Signal using the carried Sensing Transmit Power Information. If processing fails, it replies with a Sensing Power Control Failure message to the SF, carrying the reason for the failure.

[0058] Step E.7: SF sends a sensing measurement result transmission message to the RAN node, carrying the sensing measurement results.

[0059] Step E.8: The RAN node calculates the new sensed transmit power based on the sensed measurement results, sensed QoS, current sensed transmit power, etc.

[0060] Step E.9: Optionally, for uplink sensing power control, the RAN node sends power control information to the terminal, carrying uplink sensing power control information. After receiving the power control message, the terminal uses the carried uplink sensing power control information to transmit an uplink sensing reference signal. The power control can be implemented through MAC CE, downlink control information (DCI), or RRC signaling.

[0061] Based on the above embodiments, this method further includes: S6. Obtain perception performance data after perception resource allocation or reconfiguration operations to verify that the service quality of perception services meets preset standards; Specifically, step S6 includes: S61.SF Acquire sensing performance data, which includes: receiving sensing measurement result reports from sensing terminals, actively monitoring the performance indicators of sensing services, or obtaining sensing service statistics from RAN nodes within their jurisdiction; S62. Verify whether the perception performance data meets the preset service quality standards of the perception service. The verification indicators include: Whether the sensing accuracy meets the required threshold, whether the sensing latency meets business needs, the reliability indicators of the sensing business, and the resource utilization efficiency indicators.

[0062] S63. When perceived performance meets or exceeds service quality requirements, confirm that the current resource allocation scheme is effective and maintain the existing configuration; When a perceived performance degradation or failure to meet service quality requirements occurs, the perceived resource reallocation process is retried, and the verification process is repeated until the service quality requirements are met or all feasible resource combinations have been tried. Specifically, the retried perceived resource reallocation process applies to resource configuration updates initiated by the network side, resource configuration updates initiated by the access network side, and resource reconfiguration processes initiated by the terminal side.

[0063] Example 2: like Figure 10 As shown, this embodiment provides a resource management device for an integrated communication and sensing system, the device comprising: The training distribution and receiving module is used to distribute training requests for perception artificial intelligence models to one or more perception nodes, and to receive model gradient vectors reported by multiple perception nodes after training based on local perception data. The model aggregation module is used to aggregate the perceptual artificial intelligence model based on the model gradient vector to obtain a global perceptual artificial intelligence model. The model deployment module is used to deploy the global perception artificial intelligence model to the corresponding perception nodes; The decision generation module is used to receive sensing information, use the deployed global sensing artificial intelligence model to predict the sensing information, and generate sensing resource management decisions. The operation execution module is used to initiate sensing resource allocation or reconfiguration operations based on the sensing resource management decision. The performance verification module is used to obtain perception performance data after perception resource allocation or reconfiguration operations, in order to verify that the service quality of perception services meets preset standards.

[0064] Based on the above embodiments, a training triggering module is also included, used for: Receive training requests for perceptual artificial intelligence models from the artificial intelligence model management function; or, Send a model training request message to the artificial intelligence model management function, the request message including perceived business characteristics and perceived business service quality requirements information; Receive model information from the AI ​​model management function that matches the business characteristics and service quality requirements of the globally perceived AI model.

[0065] Based on the above embodiments, the model deployment module includes: The model request unit is used to send a perception artificial intelligence model request to the artificial intelligence model management function, wherein the request carries perception business characteristics and perception business service quality requirements information. The model receiving unit is used to receive model information of the globally perceptive artificial intelligence model returned by the artificial intelligence model management function, which matches the business characteristics and service quality requirements; A node determination unit is used to determine a target sensing node from multiple sensing nodes based on the sensing service characteristics and the computing power of the target sensing node. The configuration distribution unit is used to send a model configuration request to the target perception node. The model configuration request carries the model information so that the target perception node can complete the model loading.

[0066] Based on the above embodiments, the decision generation module includes a new business decision unit, used for: Receive sensing information from sensing service requests or newly established sensing services reported by sensing nodes. In response to the sensing information of newly established sensing services, a sensing service analysis model, a sensing target channel prediction model, and a sensing resource allocation algorithm model are matched from the global sensing artificial intelligence model. The perception service analysis model is used to generate perception service change prediction results for newly established perception services. The aforementioned sensing target channel prediction model is used to generate environmental change prediction results for newly established sensing services. The predicted results of changes in the sensing services and the predicted results of changes in the environment are input into the sensing resource allocation algorithm model to generate the sensing resource decision for the allocation of the newly established sensing services.

[0067] Based on the above embodiments, the decision generation module includes a resource optimization unit, used for: Obtain system-wide awareness information and resource status information; Based on the sensing services and their resource status information, the sensing service analysis model, the sensing target channel prediction model, and the sensing resource allocation algorithm model are matched from the global sensing artificial intelligence model. The sensing services and their resource status information are input into the sensing service analysis model to generate sensing service change prediction results. The sensing services and their resource status information are input into the sensing target channel prediction model to generate environmental change prediction results. The predicted results of changes in the sensing services and the predicted results of changes in the environment are input into the sensing resource allocation algorithm model to generate the sensing resource optimization scheme and resource reconfiguration scheme for the sensing services.

[0068] Based on the above embodiments, the operation execution module includes: The terminal configuration unit is used to send a sensing resource allocation or reconfiguration instruction to the sensing node so that the sensing node updates its sensing resources. The resource pool management unit is used to update the configuration information of the sensing resource pool managed by the network device.

[0069] It should be noted that the specific manner in which each module performs its operation in the apparatus described in the above embodiments has been described in detail in the embodiments of the method, and will not be elaborated here.

[0070] Example 3: Corresponding to the above method embodiments, this embodiment also provides a resource management method and apparatus for an integrated communication and sensing system. The resource management method and apparatus for an integrated communication and sensing system described below can be referred to in correspondence with the resource management method for an integrated communication and sensing system described above.

[0071] Figure 11 This is a block diagram illustrating a resource management method and apparatus 800 for an integrated communication and sensing system according to an exemplary embodiment. Figure 11 As shown, the resource management method and apparatus 800 of the integrated communication and sensing system may include: a processor 801 and a memory 802. The resource management method and apparatus 800 of the integrated communication and sensing system may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0072] The processor 801 controls the overall operation of the resource management method and apparatus 800 of the integrated communication and sensing system to complete all or part of the steps in the resource management method of the integrated communication and sensing system. The memory 802 stores various types of data to support the operation of the resource management method and apparatus 800 of the integrated communication and sensing system. This data may include, for example, instructions for any application or method operating on the resource management method and apparatus 800 of the integrated communication and sensing system, as well as application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented using any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the resource management method and apparatus 800 of the integrated communication and sensing system and other devices. Wireless communication includes, for example, Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or one or more combinations thereof. Therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, and an NFC module.

[0073] In an exemplary embodiment, the resource management method and apparatus 800 of the integrated communication and sensing system can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the resource management method of the integrated communication and sensing system described above.

[0074] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When executed by a processor, the program instructions implement the steps of the resource management method of the communication-sensing integrated system described above. For example, the computer-readable storage medium may be the memory 802 including the program instructions described above. The program instructions may be executed by the processor 801 of the resource management method and apparatus 800 of the communication-sensing integrated system to complete the resource management method of the communication-sensing integrated system described above.

[0075] Example 4: Corresponding to the above method embodiments, this embodiment also provides a readable storage medium. The readable storage medium described below can be referred to in conjunction with the resource management method of the integrated communication and sensing system described above.

[0076] A readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the steps of the resource management method of the communication-sensing integrated system described in the above method embodiments are implemented.

[0077] Specifically, the readable storage medium can be a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other readable storage medium capable of storing program code.

[0078] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0079] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A resource management method for an integrated communication and sensing system, characterized in that, Applied to network devices, including: Distribute training requests for perception AI models to one or more perception nodes, and receive model gradient vectors reported by one or more perception nodes after training based on local perception data. The perceptual artificial intelligence model is aggregated based on the gradient vector of the model to obtain a global perceptual artificial intelligence model; Deploy the global perception AI model to the corresponding perception nodes; Receive sensing information, use the deployed global sensing artificial intelligence model to predict the sensing information, and generate sensing resource management decisions; Initiate sensing resource allocation or reconfiguration operations based on the sensing resource management decision; Obtain perception performance data after perception resource allocation or reconfiguration operations to verify that the service quality of perception services meets preset standards.

2. The resource management method for the integrated communication and sensing system according to claim 1, characterized in that, Before distributing a perception AI model training request to one or more perception nodes, the following steps are also included: Receive training requests for perceptual artificial intelligence models from the artificial intelligence model management function; or, Send a model training request message to the artificial intelligence model management function, the request message including perceived business characteristics and perceived business service quality requirements information; Receive model information from the AI ​​model management function that matches the perceived business characteristics and service quality requirements of the global perception AI model.

3. The resource management method for the integrated communication and sensing system according to claim 1, characterized in that, Deploying the global perception AI model to the corresponding perception nodes includes: Send a perception AI model request to the AI ​​model management function, the request carrying perception business characteristics and perception business service quality requirements information; Receive model information of the global perception artificial intelligence model returned by the artificial intelligence model management function, which matches the perception business characteristics and service quality requirements; The target sensing node is determined based on the sensing service characteristics and the computing power of the target sensing node. A model configuration request is sent to the target sensing node, the model configuration request carrying the model information, so that the target sensing node can complete the model loading.

4. The resource management method for the integrated communication and sensing system according to claim 1, characterized in that, Receiving sensing information, using a deployed global sensing AI model to predict the sensing information, and generating sensing resource management decisions, including: Receive sensing information from sensing service requests or newly established sensing services reported by sensing nodes. Based on the sensing information of the newly established sensing service, the sensing service analysis model, the sensing target channel prediction model, and the sensing resource allocation algorithm model are matched from the global sensing artificial intelligence model. The perception service analysis model is used to generate perception service change prediction results for newly established perception services. The aforementioned sensing target channel prediction model is used to generate environmental change prediction results for newly established sensing services. The predicted results of changes in the sensing services and the predicted results of changes in the environment are input into the sensing resource allocation algorithm model to generate the sensing resource decision for the allocation of the newly established sensing services.

5. The resource management method for the integrated communication and sensing system according to claim 1, characterized in that, Receiving sensing information, using a deployed global sensing AI model to predict the sensing information, and generating sensing resource management decisions, including: Acquire information about the status of sensing services and their resources; Based on the sensing services and their resource status information, the sensing service analysis model, the sensing target channel prediction model, and the sensing resource allocation algorithm model are matched from the global sensing artificial intelligence model. The sensing services and their resource status information are input into the sensing service analysis model to generate sensing service change prediction results. The sensing services and their resource status information are input into the sensing target channel prediction model to generate environmental change prediction results. The predicted results of changes in the sensing services and the predicted results of changes in the environment are input into the sensing resource allocation algorithm model to generate the sensing resource optimization scheme and resource reconfiguration scheme for the sensing services.

6. The resource management method for the integrated communication and sensing system according to claim 5, characterized in that, The sensing services and their resource status information are input into the sensing service analysis model to generate sensing service change prediction results, including: Based on the perception service and its resource status information, M perception artificial intelligence models are matched from the global perception artificial intelligence model, where M≥1; The perception measurement results are processed in parallel using the M perception artificial intelligence models to obtain M initial prediction results; The M initial prediction results are weighted and aggregated to generate the final prediction result of perceived business changes.

7. The resource management method for the integrated communication and sensing system according to claim 1, characterized in that, Based on the aforementioned sensing resource management decision, a sensing resource allocation or reconfiguration operation is initiated. When the network device is a sensing function (SF) and the sensing node is a RAN node, the operation includes: Based on the aforementioned sensing resource management decision, SF obtains the updated RAN node sensing reference signal resource configuration information; Send a Sensing Reference Signal Resource Configuration Update Message to the RAN Node, the Sensing Reference Signal Resource Configuration Update Message carrying the RAN Node Identifier and a list of Sensing Reference Signal Resource Sets; After receiving the Sensing Reference Signal Resource Configuration Update message, if the RAN node accepts the update, it saves and updates its local Sensing Reference Signal Resource Configuration and replies to the SF with a Sensing Reference Signal Resource Configuration Update Confirmation Message; if it does not accept the update, the RAN node replies to the SF with a Sensing Reference Signal Resource Configuration Update Failure Message and the reason for the failure. When the RAN node's sensing reference signal resource configuration is updated, the SF updates the sensing reference signal resource configuration information of the sensing terminal.

8. The resource management method of the integrated communication and sensing system according to claim 7, characterized in that, The sensing reference signal resource set list includes the number of sensing reference signal resource sets and sensing reference signal resource set information; wherein, the sensing reference signal resource set information includes one or more of the following: sensing reference signal resource bandwidth, comb size, resource set period, resource repetition factor, resource set start time and duration, number of resource symbols, and sensing reference signal resource list; the sensing reference signal resource list includes one or more of the following: number of sensing reference signal resources, sensing reference signal resource set identifier, and sensing reference signal resource identifier.

9. The resource management method of the integrated communication and sensing system according to claim 1, characterized in that, Based on the aforementioned sensing resource management decision, a sensing resource allocation or reconfiguration operation is initiated. When the network device is a sensing function (SF) and the sensing node is a RAN node, the operation includes: The RAN node sends a RAN sensing reference signal resource configuration update message to the SF. The reference signal resource configuration update message includes the RAN node identifier and a list of sensing reference signal resource sets. After receiving the RAN-aware reference signal resource configuration update message, if the SF accepts the update, it replies to RAN node 1 with a RAN-aware reference signal resource configuration update confirmation message; if the SF does not accept the update, it replies to the RAN node with a RAN-aware reference signal resource configuration update failure message and the reason for the failure. When the sensing reference signal resource configuration of the RAN node is updated, the sensing reference signal resource configuration of the sensing terminal is also updated.

10. A resource management device for an integrated communication and sensing system, characterized in that, include: The training request distribution module is used to distribute training requests for the perception artificial intelligence model to one or more perception nodes, and to receive model gradient vectors reported by multiple perception nodes after training based on local perception data. The model aggregation module is used to aggregate the perceptual artificial intelligence model based on the model gradient vector to obtain a global perceptual artificial intelligence model. The model deployment module is used to deploy the global perception artificial intelligence model to the corresponding perception nodes; The decision generation module is used to receive sensing information, use the deployed global sensing artificial intelligence model to predict the sensing information, and generate sensing resource management decisions. The operation execution module is used to initiate sensing resource allocation or reconfiguration operations based on the sensing resource management decision. The performance verification module is used to obtain perception performance data after perception resource allocation or reconfiguration operations, in order to verify that the service quality of perception services meets preset standards.

Images