Target sensing method, communication node, storage medium and program product

Abstract

Provided in the present application are a target sensing method, a communication node, a storage medium and a program product. The method is applied to a receiving node in an integrated sensing and communication system, and comprises: receiving a signal reflected by a target; and on the basis of the signal reflected by the target, a sensing model configured by a receiving node itself, and received auxiliary information, determining a target sensing result, wherein a sensing model function of the receiving node corresponding to the auxiliary information is different from a sensing model function of the sensing model configured by the receiving node itself. The stable sensing of a target and an environment in an area that requires sensing is realized, thereby reducing the requirements of AI sensing for prior information about the environment, and improving the sensing precision.

Description

Target perception method, communication node, storage medium and program product Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a target perception method, a communication node, a storage medium, and a program product. Background Art

[0002] During propagation, wireless signals are often affected by surrounding objects and the environment, causing changes in signal amplitude, phase, and other characteristics. By analyzing the received signals, not only can the communication information carried by the signals be obtained, but also information reflecting the characteristics of the target objects or environment can be extracted. Integrated Communication and Sensing (ISAC) refers to the ability to perceive targets or environmental objects in the area while achieving communication functions in the same system through spectrum sharing, hardware sharing, and signal sharing. Current ISAC systems are mainly used in high-precision positioning and tracking, synchronous imaging and map reconstruction, gesture and expression recognition, and other aspects.

[0003] Artificial Intelligence (AI) perception uses artificial intelligence algorithms or models, such as neural networks, to achieve high-precision estimation of parameters such as target distance, angle, and speed, or high-resolution reconstruction of the perceived environment. Compared to traditional environmental perception methods, AI perception offers the advantages of high estimation accuracy and rapid inference speed. Applying AI perception to ISAC systems can significantly improve the speed and performance of ISAC systems in environmental perception.

[0004] However, AI perception requires high levels of prior knowledge and environmental stability. When the actual channel environment differs from the channel environment in the training data, the accuracy of AI perception will be severely reduced. Furthermore, the mapping relationship between input and output in AI perception is more complex. Simultaneously estimating parameters such as the number, location, and velocity of perceived targets using a single model may be difficult to implement or require significant training overhead, making it difficult to meet current communication needs. Summary of the Invention

[0005] The present application provides a target perception method, communication node, storage medium and program product to solve the problem of reduced AI perception accuracy due to the lack of prior knowledge of some AI models when AI perception is applied to the ISAC system. It achieves stable perception of targets and environments within the required perception area, reduces the requirements of AI perception for prior environmental information, and improves perception accuracy.

[0006] To achieve the above objectives, an embodiment of the present application provides a target perception method, which is applied to a receiving node in a synaesthesia integration system, comprising:

[0007] Receive target reflected signal;

[0008] Determine the target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the received auxiliary information;

[0009] The perception model function of the receiving node corresponding to the auxiliary information is different from the perception model function of the perception model configured by the receiving node itself.

[0010] To achieve the above-mentioned purpose, an embodiment of the present application provides a communication node, including: a memory, a processor, a program stored on the memory and runnable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, a target perception method such as any one of the embodiments of the present application is realized.

[0011] To achieve the above-mentioned purpose, an embodiment of the present application provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the target perception method of any one of the embodiments of the present application.

[0012] To achieve the above-mentioned objectives, an embodiment of the present application provides a computer program product, including a computer program, which implements the target perception method of any one of the embodiments of the present application when executed by a processor.

[0013] The target perception method, communication node, storage medium and program product provided in the embodiments of the present application receive a target reflection signal; determine the target perception result according to the target reflection signal, the perception model configured by the receiving node itself and the auxiliary information received; wherein the perception model function of the receiving node corresponding to the auxiliary information is different from the perception model function of the perception model configured by the receiving node itself. By adopting the above technical solution, perception models with different perception model functions are set in different receiving nodes in the synaesthesia integration system, so that after each receiving node in the synaesthesia integration system receives the target reflection signal reflected by the required perception target, it can determine part of the perception information of the required perception target based on the perception model configured by itself, and then determine the final required target perception result in combination with other types of perception information received from other receiving nodes for the required perception target. Since the perception models used to realize different functions are respectively set on different nodes in the synaesthesia integration system, the number of perception model parameters configured on each receiving node is small, which reduces the training overhead of the perception model corresponding to each receiving node. At the same time, since different receiving nodes in the integrated synaesthesia system perceive the required perception targets differently from different perception function directions, the prior knowledge required for the perception models of different perception model functions is different. There is no need to provide complete prior knowledge for each perception model. Even if the prior knowledge of some perception models is missing, the remaining perception models can still give relatively accurate partial perception results, so that the final combined target perception results can still maintain a high degree of accuracy, realizing stable perception of targets and environments in the required perception area, reducing the requirements of AI perception for environmental prior information, and improving perception accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] FIG1 is a flow chart of a target sensing method according to an embodiment of the present application;

[0015] FIG2 is a flow chart of another target perception method provided in an embodiment of the present application;

[0016] FIG3 is an example diagram of a data transmission process of multi-model collaborative target perception provided by an embodiment of the present application;

[0017] FIG4 is a diagram illustrating an example of a data transmission process for multi-model collaborative target perception according to an embodiment of the present application;

[0018] FIG5 is another example diagram of a data transmission process for multi-model collaborative target perception provided by an embodiment of the present application;

[0019] FIG6 is another example diagram of a data transmission process for multi-model collaborative target perception provided by an embodiment of the present application;

[0020] FIG. 7 is a diagram illustrating an example of a process for reporting information of a perception signal receiving node deployment model provided in an embodiment of the present application.

[0021] FIG8 is an example diagram of a data transmission process of multi-model collaborative supervision provided in an embodiment of the present application;

[0022] FIG9 is a diagram illustrating an example of a data transmission process for multi-model collaborative supervision provided in an embodiment of the present application;

[0023] FIG10 is a flowchart illustrating a process of shutting down or updating a perception model according to an embodiment of the present application;

[0024] FIG11 is a flowchart illustrating another process of shutting down or updating a perception model according to an embodiment of the present application;

[0025] FIG12 is a flowchart illustrating an example of a perception model update process provided by an embodiment of the present application;

[0026] FIG13 is a flowchart illustrating another process of updating a perception model according to an embodiment of the present application;

[0027] FIG14 is a schematic diagram of the structure of a target sensing device provided in an embodiment of the present application;

[0028] FIG15 is a schematic diagram of the structure of a communication node provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] To make the purpose, technical solutions and advantages of this application more clear, the embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of this application can be combined with each other in any way.

[0030] The steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions. Also, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be performed in an order different from that shown here.

[0031] During propagation, wireless signals are often affected by surrounding objects and the environment. The ISAC system, through spectrum sharing, hardware sharing, and signal sharing, achieves both communication and perception of the target environment or objects within the area. The ISAC system includes at least a transmitting node, a receiving node, and a network-side node. Both the transmitting and receiving nodes can be base stations or user equipment (UE), and the network-side node is a network-side unit with a sensing function (SF). Traditional sensing methods can be divided into six modes: inter-base station transmission and reception, base station transmission and UE reception, base station self-transmission and self-reception, inter-UE transmission and reception, and UE transmission and base station collection and UE self-transmission and self-reception. Regardless of the sensing mode, the sensing process includes the transmitting node transmitting a sensing reference signal (Sensing RS) to the target within the sensing area; sensing the target or background environment reflecting the sensing reference signal; the receiving node receiving the sensing reference signal reflected by the target to generate a measurement result; the receiving node itself using a sensing algorithm to estimate the sensing result, or the receiving node reporting the measurement result to the SF, which then uses the sensing algorithm to estimate the sensing result of the target.

[0032] When applying AI perception to the ISAC system, although different AI perception models with different perception model functions can be configured on different receiving nodes in the ISAC system, its overall perception process is basically the same as the traditional perception process. The difference is that the measurement results in the traditional perception method are generally the arrival time (TOA), arrival angle (AOA) and Doppler shift (Doppler Shift) of the perception target; while for the ISAC system under AI perception, the measurement results of the receiving node are generally the channel impulse response (CIR) matrix, which is formed by the superposition of the environment, clutter and multiple perception target reflection signals in the perception area. Compared with AI positioning, the AI ​​positioning model inputs the CIR matrix generated by the UE measurement and directly outputs the UE estimated position. In AI perception, the model outputs information such as the estimated number of targets, the location and speed of each target, and other information. The mapping relationship between its input and output is more complex. It would be difficult to implement an AI perception model at each receiving node in the ISAC system to simultaneously estimate the number, location, and speed of perceived targets. This would require a large number of training data samples and a longer training time, resulting in significant training overhead. Furthermore, because AI perception models require high levels of prior knowledge and environmental stability, when the actual channel environment differs from the channel environment in the training data, the accuracy of AI perception models that directly output all perception results will be severely reduced, making it difficult to meet the current needs for perceiving targets and environmental information within communication environments.

[0033] To solve the above problems, the present application provides a target perception method, which can reduce the demand of the AI ​​perception model in each receiving node for prior environmental information through the interaction between the output results of the AI ​​perception model in multiple receiving nodes located in the perception area of ​​the ISAC system, while improving the accuracy of the final perception results.

[0034] In an exemplary embodiment, Figure 1 is a flow chart of a target perception method provided in an embodiment of the present application. The method can be applicable to situations where targets and environmental information are perceived within a required perception area. The method can be executed by a target perception device, which can be executed by software and / or hardware and integrated on a communication node. Exemplarily, the communication node can be a receiving node in an ISAC system.

[0035] As shown in FIG1 , the target perception method provided in the embodiment of the present application specifically includes the following steps:

[0036] S110: Receive a target reflected signal.

[0037] In this embodiment, the target reflected signal may be specifically understood as a signal received by the receiving node after the sensing reference signal transmitted by the transmitting node is reflected by the sensing target or background environment within the sensing area.

[0038] Specifically, when the ISAC system needs to perceive a target within the perception area, the transmitting node in the ISAC system will first transmit a perception reference signal into the perception area. The perception reference signal will be reflected when it encounters an environmental target or perception target within the perception area. At the same time, the reflected signal will be superimposed with the clutter in the environment to form a target reflection signal, which will eventually be received by the receiving node in the ISAC system.

[0039] It is understandable that the sensing reference signal may be reflected at different angles when reflected by an environmental target or a sensing target within the sensing area, forming different target reflection signals to be received by a receiving node in the corresponding reflection direction.

[0040] S120: Determine a target perception result according to the target reflection signal, a perception model configured by the receiving node itself, and the received auxiliary information.

[0041] The perception model function of the receiving node corresponding to the auxiliary information is different from the perception model function of the perception model configured by the receiving node itself.

[0042] In this embodiment, the perception model can be specifically understood as an AI model with different perception functions, installed in each receiving node of the ISAC system. It is understood that the lifecycle management (LCM) of the perception model in AI perception is similar to that of existing models in AI positioning and AI beam management, and can also include several steps: data collection, model training, model inference, and model supervision and update. The perception model in the embodiment of the present application can be trained based on perception measurement results collected by the base station, UE, or SF, combined with a training sample set generated by labels corresponding to the perception function. Auxiliary information can be specifically understood as information sent to the receiving node by receiving nodes other than the receiving node in the ISAC system, containing results output by the perception models configured in other receiving nodes that differ from the perception model function of the receiving node. Target perception results can be specifically understood as information obtained after sensing the desired perception target within the perception area, describing the state of the perceived target from different perception function directions. Optionally, different perception function directions can include sensing the target from distance, angle, and speed. The perception model function can be specifically understood as the direction in which different perception models can perceive the perception target, such as determining the distance between the perception target and the receiving node, determining the number of perception targets within the perception area, or determining the position and speed of the perception target. Accordingly, the output results output by the receiving nodes with different perception model functions are combined to form the target perception result. For example, the target perception result may include information such as the number of perception targets perceived within the perception area, the position and speed of each perception target, etc., which is not limited in the embodiments of the present application.

[0043] Specifically, the CIR matrix obtained from the target reflection signal measurement is input into the receiving node's configured perception model. This allows the receiving node to perceive the targets within the perception area using the corresponding perception model function of the configured perception model, obtaining a partial perception result corresponding to the configured perception model function of the receiving node. Simultaneously, the receiving node also receives auxiliary information from other receiving nodes in the ISAC system, obtaining perception results output by perception models with different perception model functions than the receiving node's. The receiving node then combines the partial perception result determined by the receiving node with the other perception results contained in the received auxiliary information to obtain a target perception result that encompasses all the perception targets within the perception area.

[0044] The target perception method provided in the embodiment of the present application receives the target reflection signal; determines the target perception result according to the target reflection signal, the perception model configured by the receiving node itself and the received auxiliary information; wherein the perception model function of the receiving node corresponding to the auxiliary information is different from the perception model function of the perception model configured by the receiving node itself. By adopting the above technical solution, perception models with different perception model functions are set in different receiving nodes in the synaesthesia integration system, so that after each receiving node in the synaesthesia integration system receives the target reflection signal reflected by the required perception target, it can determine part of the perception information of the required perception target based on the perception model configured by itself, and then determine the final required target perception result in combination with other types of perception information received from other receiving nodes for the required perception target. Since the perception models used to realize different functions are respectively set on different nodes in the synaesthesia integration system, the amount of perception model parameters configured on each receiving node is small, which reduces the training overhead of the perception model corresponding to each receiving node. At the same time, since different receiving nodes in the integrated synaesthesia system perceive the required perception targets differently from different perception function directions, the prior knowledge required for the perception models of different perception model functions is different. There is no need to provide complete prior knowledge for each perception model. Even if the prior knowledge of some perception models is missing, the remaining perception models can still give relatively accurate partial perception results, so that the final combined target perception results can still maintain a high degree of accuracy, realizing stable perception of targets and environments in the required perception area, reducing the requirements of AI perception for environmental prior information, and improving perception accuracy.

[0045] In one embodiment, before receiving the target reflected signal, the method further includes:

[0046] Reporting at least one of a perception model function and a perception model number of the perception model configured by the receiving node itself to the network side node;

[0047] Receives the perception configuration information sent by the network side node.

[0048] In this embodiment, the perception model number can be specifically understood as a number used to indicate the identity information of the perception model configured in the receiving node. It can be understood that since the perception models configured in each receiving node in the ISAC system are different, in order to distinguish and uniformly manage different receiving nodes and the perception models configured in the receiving nodes, a perception model number corresponding to each perception model and unique in the ISAC system can be set for each perception model. The perception configuration information can be specifically understood as information sent by the network-side node to the transmitting node and the receiving node in the ISAC system, which contains the information required for perception within the perception area to configure the transmitting node and the receiving node. Optionally, the perception configuration information may include information such as the range required for perception by the ISAC system, the target type for the target required to be perceived within the perception range, and the upper limit of the number of perception targets. Exemplarily, the perception configuration information may include the lower limit and upper limit of the required perception target distance, the lower limit and upper limit of the required perception target angle, the lower limit and upper limit of the required perception target speed, the lower limit and upper limit of the required number of perception targets, and the target type required to be perceived, etc., wherein the target type may include pedestrian targets, bird targets, vehicle targets, drone targets, etc., which can be set according to actual needs. The embodiments of the present application do not limit this.

[0049] Specifically, when constructing an ISAC system, multiple communication nodes within the desired perception area can be used as receiving nodes or transmitting nodes in the ISAC system. Furthermore, since AI perception models with different perception model functions can be configured in different receiving nodes, each receiving node can determine the perception model function and perception model number of its own configured perception model after completing configuration of the AI ​​perception model. To facilitate subsequent target perception configuration for each node in the ISAC system by the network-side node, each receiving node in the ISAC system can report at least one of the perception model function and perception model number of the configured perception model to the network-side node after completing configuration of its own perception model. This allows the network node to generate perception configuration information based on the received perception model function and / or perception model number, as well as the predetermined perception requirements for the perception area, and send the perception configuration information to each transmitting node and receiving node in the ISAC system. This allows each transmitting node and each receiving node to complete the corresponding configuration after receiving the perception configuration information, and perform target perception within the perception area based on the configuration.

[0050] In one embodiment, the sensing configuration information includes at least multi-model cooperative mode signaling. Accordingly, the target sensing result is determined based on the target reflection signal, the sensing model configured by the receiving node itself, and the received auxiliary information, including:

[0051] Obtain multi-model collaboration mode signaling from the perception configuration information;

[0052] Determine the target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the multi-model cooperation mode signaling;

[0053] Among them, the multi-model collaboration mode signaling includes the model collaboration order of each perception model participating in the collaboration, and at least one of the perception model function and the perception model number.

[0054] In this embodiment, the multi-model collaboration mode signaling can be specifically understood as the signaling sent by SF to the receiving node in the ISAC system, so that the receiving node can clearly understand the model collaboration mode between the perception models with multiple different perception model functions in the ISAC system. The model collaboration sequence can be specifically understood as the processing order of the received reflection signals by the multiple perception models participating in AI perception when determining the perception results. For example, assuming that the perception models participating in target perception are Model1, Model2 and Model3, and the required perception models are executed in the order of Model2->Model1->Model3, a model collaboration sequence of the example content FirstModel2, SecondModel1, ThirdModel3 can be generated, and the model collaboration sequence is sent to the receiving node in the ISAC system as part of the multi-model collaboration mode signaling.

[0055] Specifically, after receiving the perception configuration information, the receiving node can obtain the multi-model collaboration mode signaling contained therein. The multi-model collaboration mode signaling should include at least one of the perception model function and perception model number of each perception model currently participating in the collaboration, so that the receiving node can clearly identify the perception model that will participate in perceiving the perception area. At the same time, the multi-model collaboration mode signaling must also include the model collaboration order of each perception model participating in the collaboration. After the receiving node has clearly identified the identity information of each perception model participating in the collaboration based on the multi-model collaboration mode signaling, it can determine the processing order of each perception model for the received reflection signal based on the obtained model collaboration order. At the same time, since the receiving node has clearly defined the perception model function and perception model number of its own configured perception model, the receiving node can determine the position of its own corresponding perception model in the model collaboration, and then the receiving node can complete the reception of auxiliary information given by other receiving nodes in the ISAC system according to the multi-model collaboration mode signaling, process the target reflection signal according to its own configured perception model, and send the processing results of its own configured perception model to other receiving nodes according to the multi-model collaboration mode signaling, and finally combine the output results of its own configured perception model with the received auxiliary signal to determine the target perception result of the perception area.

[0056] In one embodiment, when the perception models participating in the collaboration include a perception model whose perception model function is to estimate the number of targets, determining a target perception result based on a target reflection signal, a perception model configured by the receiving node itself, and auxiliary information received based on multi-model collaboration mode signaling includes:

[0057] Determine the target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the model cooperation sequence;

[0058] The auxiliary information includes first auxiliary information and second auxiliary information.

[0059] In this embodiment, the perception model whose perception model function is to estimate the number of targets can be specifically understood as an AI perception model used to estimate the number of perceived targets within the perception area. The first auxiliary information can be specifically understood as information generated by the perception model whose perception model function is to estimate the number of targets after processing the target reflection signal received by its corresponding receiving node, and used to assist other receiving nodes in the ISAC system in determining the target perception result. The second auxiliary information can be specifically understood as information generated by the perception model whose perception model function is not to estimate the number of targets after processing the target reflection signal received by its corresponding receiving node, and used to assist other receiving nodes in the ISAC system in determining the target perception result.

[0060] Specifically, when the perception models participating in the collaboration include a perception model whose perception model function is target number estimation, since the output of the perception model whose perception model function is target number estimation is the number of perceived targets within the perception area, this number of perceived targets can be used to adjust the perception models of other receiving nodes in the ISAC system so that the number of output results of the adjusted perception models is consistent with the number of perceived targets. Therefore, the perception model whose perception model function is target number estimation can be selected as the first perception model in the model collaboration order. Furthermore, each receiving node in the ISAC system can perform auxiliary information reception and target perception processing based on the target reflection signal it receives, its configured perception model, and the position of its configured perception model in the model collaboration order. Ultimately, the output result of its configured perception model is combined with the received auxiliary signal to determine the target perception result in the perception area.

[0061] In one example, when the perception model function of the perception model configured by the receiving node itself is target number estimation, the perception model corresponding to the receiving node is placed first in the model cooperation order; accordingly, determining the target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the model cooperation order includes:

[0062] determining a channel impulse response matrix based on the target reflected signal;

[0063] The channel impulse response matrix is ​​input into the perception model configured by the receiving node itself to determine the target number estimation result;

[0064] Sending the target number estimation result to other receiving nodes in the synaesthesia integration system, so that the target number estimation result is used as the first auxiliary information received by each other receiving node;

[0065] receiving second auxiliary information sent by each other receiving node;

[0066] The target number estimation result and each second auxiliary information are combined to determine the target perception result.

[0067] Specifically, when the perception model function of the perception model configured by the receiving node itself is target number estimation, the receiving node can be considered to be the first receiving node in the model collaboration order, that is, the receiving node itself must first complete the estimation of the number of perceived targets within the perception area. At this time, a corresponding CIR matrix can be generated based on the received target reflection signal, and this CIR matrix can be input into the perception model configured by the receiving node itself to obtain the target number estimation result output by the perception model. The receiving node then sends the target number estimation result as first auxiliary information to other receiving nodes in the ISAC system. After each other receiving node completes the adjustment of its own perception model based on the first auxiliary information, it inputs the CIR matrix corresponding to the received target reflection signal into its own perception model according to the model collaboration order, and feeds back the generated result as second auxiliary information according to the model collaboration order to the receiving node whose perception model function is target number estimation. In this case, the receiving node whose perception model function is target number estimation has completed the acquisition of the perception results of each perception model function in the ISAC system. By combining the target number estimation results obtained by its own processing with each second auxiliary information, it can obtain a complete perception result for each perception target in the perception area at different angles, that is, the target perception result.

[0068] It can be understood that the first auxiliary information includes but is not limited to the target number estimation result output by the perceptual model whose perceptual model function is target number estimation; the second auxiliary information includes but is not limited to the result output by the perceptual model whose perceptual model function is not target number estimation.

[0069] In one example, when the perception model function of the perception model configured in the receiving node itself is not target number estimation, the perception model whose perception model function is target number estimation is placed first in the model cooperation order; accordingly, determining the target perception result based on the target reflection signal, the perception model configured in the receiving node itself, and the auxiliary information received according to the model cooperation order includes:

[0070] determining a channel impulse response matrix based on the target reflected signal;

[0071] receiving first auxiliary information sent by a receiving node whose perception model function is target number estimation; wherein the first auxiliary information is a target number estimation result;

[0072] adjusting a perception model configured by the receiving node itself according to the first auxiliary information, and inputting the channel impulse response matrix into the adjusted perception model to determine a partial perception result;

[0073] Sending the partial perception results to other receiving nodes in the synaesthesia integration system so that the partial perception results are used as second auxiliary information received by each of the other receiving nodes;

[0074] receiving second auxiliary information sent by each receiving node other than the receiving node for which the perception model function is target number estimation;

[0075] The first auxiliary information, the partial perception result and each second auxiliary information are combined to determine the target perception result.

[0076] In this embodiment, a partial perception result can be understood as a perception result of a corresponding perception model function for a target within a perception area, obtained by a receiving node whose perception model function is not target number estimation, after processing target reflection signals received by the receiving node. Exemplarily, the partial perception result may include at least one of the position, velocity, distance, angle, and Doppler of the perceived target.

[0077] Specifically, if the perception model function of the receiving node's own perception model is not target number counting, the receiving node is considered not the first receiving node in the model collaboration order, and the perception model whose perception model function is target number estimation is the first receiving node in the model collaboration order. In this case, the receiving node must first receive the first auxiliary information provided by the receiving node whose perception model function is target number estimation. Based on the first auxiliary information, the receiving node adjusts the number of output neurons of its own perception model so that the number of partial perception results for different perception targets ultimately output by its own perception model is the same as the target number estimation result contained in the first auxiliary information. Furthermore, when its own perception model reaches the corresponding position in the model collaboration order, the receiving node inputs the CIR matrix generated based on the received target reflection signal into its own perception model to obtain the partial perception result output by the perception model. The receiving node then transmits the partial perception result as second auxiliary information to other receiving nodes in the ISAC system. Furthermore, the receiving node may receive second auxiliary information from each of the other receiving nodes located before and after its own in the model collaboration order. After the second auxiliary information of the receiving nodes corresponding to all perception models in the model collaboration sequence has been received, it can be considered that the receiving node itself has completed the acquisition of the perception results of each perception model function in the ISAC system. Then, the receiving node can combine the partial perception results obtained by its own processing and the received first auxiliary information and each second auxiliary information to obtain the complete perception results for each perception target in the perception area in different perception function directions, that is, to obtain the target perception results.

[0078] In one embodiment, when the perception models participating in the collaboration do not include a perception model whose perception model function is to estimate the number of targets, determining a target perception result based on a target reflection signal, a perception model configured by the receiving node itself, and auxiliary information received based on multi-model collaboration mode signaling includes:

[0079] determining a channel impulse response matrix based on the target reflected signal;

[0080] Input the channel impulse response matrix into the perception model configured by the receiving node itself to determine the partial perception results;

[0081] Receive auxiliary information sent by other receiving nodes in the synaesthesia integration system according to the model cooperation order in the multi-model cooperation mode signaling, and send part of the perception results to each other receiving node, so as to use the part of the perception results as auxiliary information received by each other receiving node;

[0082] The partial perception results are combined with various auxiliary information to determine the target perception results.

[0083] Specifically, if the participating perception models do not include a perception model whose perception model function is to estimate the number of targets, the auxiliary information generated and transmitted by each receiving node in the ISAC system can be considered to be the second auxiliary information in the above-described embodiment, and the model collaboration order in the multi-model collaboration mode signaling is solely based on the functionality of the different perception models. In this case, each receiving node in the ISAC system will generate a corresponding CIR matrix based on the received target reflection signal and input the CIR matrix into its configured perception model to obtain a partial perception result. Based on the model collaboration order in the multi-model collaboration mode signaling, when its own order is reached, it will send its own partial perception result as auxiliary information to other receiving nodes in the ISAC system, while simultaneously receiving auxiliary information sent by other receiving nodes based on the model collaboration order. Once all auxiliary information corresponding to all perception models in the model collaboration order has been received, all receiving nodes in the ISAC system can be considered to have completed the acquisition of perception results for each perception target within the perception area in the different perception function directions. The receiving nodes can then combine their own partial perception results with the received auxiliary information to obtain the target perception result.

[0084] In one embodiment, the auxiliary information is information transparently transmitted from other receiving nodes in the ISAC system to the receiving node; or, the auxiliary information is information forwarded from other receiving nodes in the ISAC system to the receiving node via network-side node management.

[0085] Specifically, each receiving node in the ISAC system can transparently transmit its own generated auxiliary information directly to the receiving node at the other end. Alternatively, the information can be first sent to a network-side node, which then manages and distributes the auxiliary information to the desired receiving node, completing the transmission of auxiliary information between receiving nodes in the ISAC system.

[0086] The target perception method provided herein places AI perception models with different perception model functions on different receiving nodes in an ISAC system. When perceiving a target within a perception area, the perception models on each receiving node complete partial perception within the perception area, and the resulting partial perception results are sent as auxiliary information to other receiving nodes in the ISAC system. This allows each receiving node to ultimately obtain perception results for all perception function directions within the perception area. Because the perception models on each receiving node only need to complete the perception of their corresponding perception model functions, there is no need to provide complete prior knowledge for the perception models of each receiving node in the ISAC system. Even if prior knowledge for some perception models is missing, the remaining perception models can still provide relatively accurate partial perception results, ensuring that the resulting combined target perception results maintain high accuracy. Furthermore, a different model collaboration order is set based on whether the participating perception models include a perception model with a target number estimation function. This ensures that the perception model with a target number estimation function is prioritized during target perception. The resulting target number estimation result is then used to adjust the perception models of other receiving nodes in the ISAC system before completing perception, resulting in more accurate perception results and improved perception accuracy.

[0087] In an exemplary embodiment, Figure 2 is a flow chart of another target perception method provided in an embodiment of the present application. The embodiment of the present application is further optimized based on the above-mentioned optional technical solutions. When different receiving nodes determine the target perception results based on the received target reflection signal, their own configured perception model and the received auxiliary information, in order to ensure the accuracy of the output results of their own configured perception models, the effectiveness of the perception model configured by the current receiving node itself is supervised through the configuration of the receiving node itself or other nodes in the ISAC system, so that when the accuracy of the perception model configured by the current receiving node itself is insufficient, it can be processed in time according to the supervision results, thereby ensuring the effectiveness of each perception model in the ISAC system during its working process, and ultimately improving the accuracy of the determined target perception results.

[0088] As shown in FIG2 , the target perception method provided in the embodiment of the present application specifically includes the following steps:

[0089] S210. Report at least one of the perception model function and the perception model number of the perception model configured by the receiving node itself to the network side node.

[0090] S220. Receive perception configuration information sent by the network side node.

[0091] S230: Receive a target reflected signal.

[0092] S240: Determine a target perception result according to the target reflection signal, a perception model configured by the receiving node itself, and the received auxiliary information.

[0093] The perception model function of the receiving node corresponding to the auxiliary information is different from the perception model function of the perception model configured by the receiving node itself.

[0094] S250. Obtain model supervision mode signaling from the perception configuration information.

[0095] In this embodiment, the model supervision mode signaling can be specifically understood as signaling sent by the SF to the receiving node in the ISAC system to indicate the mode in which the perception model configured in the receiving node accepts supervision.

[0096] Specifically, in addition to the multi-model collaboration mode signaling, the perception configuration information may also include model supervision mode signaling to indicate the way in which the perception model in the corresponding receiving node needs to be supervised. Therefore, when the receiving node obtains the perception configuration information, it can directly obtain the model supervision mode signaling from the perception configuration information to clarify the model supervision mode during the working process of the perception model configured for itself.

[0097] S260. Determine the model supervision mode of the receiving node according to the model supervision mode signaling.

[0098] Among them, the model supervision method includes at least one of the following: multi-model collaborative supervision and radar cross-section supervision.

[0099] In this embodiment, multi-model collaborative supervision can be specifically understood as a model supervision method that uses the perception models of other nodes in the ISAC system to assist in supervising the accuracy of the output results of the perception model configured for the current receiving node. Radar Cross-Section (RCS) supervision can be specifically understood as a model supervision method that uses the RCS value range information pre-configured in the receiving node through perception configuration information to supervise the RCS value output by the perception model configured on the receiving node itself.

[0100] Specifically, for different receiving nodes in the ISAC system, SF can configure corresponding model supervision modes for them when generating model supervision mode signaling. When the model supervision mode is determined to be multi-model collaborative supervision, the model information used to supervise the perception model configured by the receiving node will be carried in the perception configuration information; when the model supervision mode is determined to be RCS supervision, the RCS value range information will be carried in the perception configuration information, so that the receiving node can supervise its own configured perception model based on the received RCS value range information and the RCS value output by its own configured perception model.

[0101] S270. Perform model supervision on the perception model configured by the receiving node itself through a model supervision method to determine the model supervision result.

[0102] Specifically, after the receiving node clarifies its own model supervision method, it can supervise the model output of the perception model configured by the receiving node itself according to the model supervision method to obtain the final model supervision result.

[0103] It is understandable that S240 and S250-S270 can be executed simultaneously, and the embodiment of the present application takes simultaneous execution as an example.

[0104] In one embodiment, when the model supervision mode is multi-model collaborative supervision, model supervision is performed on the perception model configured by the receiving node itself through the model supervision mode to determine the model supervision result, including:

[0105] Determine, based on the perception configuration information, a supervision model corresponding to the perception model configured by the receiving node itself;

[0106] Send the channel impulse response matrix determined according to the target reflection signal to the supervisory node corresponding to the supervisory model, and accumulate the channel impulse response matrix;

[0107] When the cumulative time unit quantity corresponding to each channel impulse response matrix reaches a preset cumulative threshold, each channel impulse response matrix is ​​input into the perception model configured by the receiving node itself to determine the output result to be supervised;

[0108] Receive supervisory auxiliary information fed back by the supervisory node;

[0109] Compare the output result to be supervised with the auxiliary information of supervision, and determine the model supervision result based on the comparison result;

[0110] The supervisory model is a perception model that has functional overlap or functional conversion with the perception model configured by the receiving node itself.

[0111] In this embodiment, the supervisory model can be specifically understood as a perception model that is set on other nodes in the ISAC system except the current receiving node, and is used to supervise the perception model configured by the current receiving node itself. It can be understood that the supervisory model is a perception model that is input by the model of the desired supervised perception model, and then inputs it into itself for processing similar to or the same as the supervised perception model, and then compares the output results of the two to determine the output accuracy of the supervised perception model. Therefore, it is necessary to ensure that the output result type of the supervisory model is consistent with that of the supervised perception model, or at least to ensure that the output result of the supervisory model can be converted into the form of the output result of the supervised perception model, so as to realize the supervision of the supervised perception model by the supervisory model. Therefore, the supervisory model should be a perception model that has functional overlap or functional conversion with the perception model configured by the receiving node itself. Exemplarily, the perceptual model function of the supervised perceptual model may be one or all of the number, position or speed of the perceptual targets, and the perceptual model function of the supervisory model may also be one or all of the number, position or speed of the perceptual targets; it should be clarified that when the perceptual model function of the supervised perceptual model is the speed of the perceptual targets, and the perceptual model function of the supervisory model is the position of the perceptual targets, the perceptual model functions of the two do not overlap, but since the speed of the perceptual targets can be determined based on the position of the perceptual targets, the supervision of the supervised perceptual model can also be achieved through the supervisory model. At this time, it can be considered that the relationship between the supervised perceptual model and the supervisory model is one that can be converted into functions.

[0112] In this embodiment, since in AI perception, the receiving node often needs to accumulate the CIR matrix obtained by processing the target reflection signal received within multiple time units, and send each CIR matrix to the perception model together to obtain the speed estimation result of the perceived target; however, the CIR matrix of each time unit can independently serve as the input of a perception model that functions as a target position perception, and obtain an estimation result about the target position. Therefore, in the supervision process of multi-model collaborative supervision, a cumulative time unit amount and a preset cumulative threshold can be set in the supervised receiving node to determine the amount of accumulated CIR matrix in the supervised receiving node, and the preset cumulative threshold is used to determine whether the accumulated CIR matrix can be input into the supervised perception model for processing to complete supervision. It can be understood that the time unit in the embodiment of the present application can be an orthogonal frequency division multiplexing symbol (OFDM symbol) or an orthogonal frequency division multiplexing time slot (OFDM slot). The supervision auxiliary information can be specifically understood as the input of the supervised perception model, which is input into the supervision model for processing and converted into information of the same type as the output of the supervised perception model to determine whether the supervised perception model is working properly.

[0113] Specifically, after receiving the sensing configuration information, the receiving node can retrieve the model supervision mode signaling contained in the sensing configuration information. Upon determining that its model supervision mode is multi-model collaborative supervision based on the model supervision mode signaling, the receiving node can determine the identity information of the model used to supervise its own configured sensing model from the sensing configuration information, thereby determining the corresponding supervisory model. At this point, the node configured with the supervisory model in the ISAC system can be identified as the supervisory node corresponding to the current receiving node. During model supervision, the receiving node continuously receives target reflection signals and processes them into a CIR matrix. Each time a CIR matrix is ​​generated, the receiving node sends the CIR matrix to the supervisory node, which then inputs the received CIR matrix into the supervisory model for processing, thereby obtaining a supervisory model output corresponding to each CIR matrix. Furthermore, each time a CIR matrix is ​​generated, the receiving node accumulates the CIR matrix and increments the corresponding accumulated time unit by one. When the accumulated time unit reaches a preset accumulation threshold, the receiving node can then input all accumulated CIR matrices into its own configured sensing model for processing, and determine the output of its own configured sensing model as the output to be supervised. While the receiving node generates its own supervised output, the supervisory node also processes the supervised model output to obtain auxiliary supervisory information consistent with the supervised output. The supervisory auxiliary information is then sent to the receiving node. Upon receiving the auxiliary supervisory information from the supervisory node, the receiving node compares the supervised output with the auxiliary supervisory information and uses the difference between the two as the model supervision result.

[0114] For example, assuming that the supervised perception model is a perception model whose perception model function is to perceive target speed, and the supervisory model is a perception model whose perception model function is to perceive target position, the supervised perception model will send the CIR matrix corresponding to N time units to the supervisory model for processing to obtain N target position estimation results. At the same time, after completing the accumulation of the CIR matrices of N time units, the supervised perception model will send the N CIR matrices to the supervised perception model itself to obtain the corresponding perception target speed estimation results. In order to enable the output results of the supervisory model to be used to supervise the supervised perception model, the N target position estimation results need to be processed by differential operation, Kalman filter (KF) and other speed estimation methods, and all instantaneous speeds or average speeds within the obtained N time units are sent as supervision auxiliary information to the receiving node where the supervised perception model is located to assist the receiving node in completing the determination of the model supervision results of its own configured perception model.

[0115] In one embodiment, the supervisory auxiliary information is the output result output by the supervisory model after each CIR matrix is ​​input respectively; or the supervisory auxiliary information is the information obtained after data processing of the output result output by the supervisory model after each CIR matrix is ​​input respectively, and the information is consistent with the type of the output result to be supervised.

[0116] In one embodiment, the supervisory node is another receiving node or a network-side node in the ISAC system.

[0117] In one embodiment, when the model supervision mode is radar cross section supervision, model supervision is performed on the perception model configured by the receiving node itself through the model supervision mode to determine the model supervision result, including:

[0118] Determine radar cross section value range information based on perception configuration information;

[0119] The channel impulse response matrix determined based on the target reflection signal is input into the perception model configured by the receiving node itself to determine the target radar cross section value;

[0120] The target radar cross section value is compared with the radar cross section value range information, and the model supervision result is determined based on the comparison result.

[0121] In this embodiment, the radar cross section value range information can be understood as information pre-set based on actual conditions, indicating the range of RCS values ​​for different types of perceived targets within the perception area. For example, the different types of perceived targets may be pedestrians, birds, and vehicles, etc., but this embodiment of the application is not limited to this.

[0122] Specifically, if the model supervision mode included in the model supervision mode signaling carried in the perception configuration information is radar cross-section supervision, then the perception configuration information must also carry RCS value range information. When the receiving node inputs the CIR matrix determined based on the target reflection signal into the perception model configured by the receiving node itself, the target RCS value corresponding to the CIR matrix can also be obtained. The target RCS value is compared with the RCS value range information. If the target RCS value output by the perception model exceeds the RCS value range configured by the SF, it can be considered that the perception model configured by the receiving node itself is mismatched. Otherwise, it can be considered that the perception model configured by the receiving node itself is in normal working condition, and the model supervision result can be determined based on the comparison result.

[0123] The target perception method provided in the embodiment of the present application completes the supervision of the perception model in the receiving node during its working process through multi-model collaborative supervision or radar scattering cross-section supervision, and supervises the effectiveness of the perception model currently configured by the receiving node itself through the configuration of the receiving node itself or other nodes in the ISAC system, so that when the accuracy of the perception model currently configured by the receiving node itself is insufficient, it can be processed in time based on the supervision results, thereby ensuring the effectiveness of each perception model in the ISAC system during its working process and ultimately improving the accuracy of the determined target perception results.

[0124] In one embodiment, after performing model supervision on the perception model configured by the receiving node itself through the model supervision method and determining the model supervision result, the method further includes:

[0125] Obtaining the perceived service quality condition from the perceived configuration information;

[0126] When the model supervision result does not meet the perception service quality conditions, the perception model configured by the receiving node itself is shut down or updated, and the shutdown status or update status is reported to the network side node.

[0127] In this embodiment, the perception service quality condition (Quality of Service, QoS) can be specifically understood as a condition for evaluating whether the perception model in AI perception is competent for perceiving the perception objects within the perception area.

[0128] Specifically, if the receiving node needs to implement supervision and control of its own configured perception model, the perception configuration information sent by the SF to the receiving node must also include QoS information. After the receiving node performs model supervision on its own configured perception model through the corresponding model supervision method and finally determines the model supervision result, it can compare the model supervision result with the QoS information. When it is determined that the model supervision result does not meet the QoS conditions, it can be considered that the perception model configured by itself cannot meet the accuracy requirements of the target perception processing. At this time, the receiving node itself can determine the need for adjustment operations on the configured perception model, such as shutdown operations or update operations, and after completing the shutdown or update of the perception model, the shutdown state or update state of the perception model is reported to the SF node in the ISAC system, so that the SF node can synchronize the model working status of each node in the ISAC system.

[0129] In one embodiment, after performing model supervision on the perception model configured by the receiving node itself through the model supervision method and determining the model supervision result, the method further includes:

[0130] Report the model supervision results to the network side node;

[0131] Receiving a model shutdown indication or a model update indication fed back by a network-side node;

[0132] Shutting down the perception model configured by the receiving node itself according to the model shutdown indication, or updating the perception model configured by the receiving node itself according to the model update indication;

[0133] Report the shutdown status or update status to the network side node.

[0134] Specifically, if the receiving node does not need to implement supervision and control of the perception model configured by itself, then after determining the model supervision result of the perception model configured by itself, the receiving node can report the model supervision result to the SF node in the ISAC system. Then, after the SF node compares the model supervision result based on the QoS information, if it is determined that the model supervision result does not meet the QoS conditions, it can generate a model shutdown indication or a model update indication accordingly, and feed back the model shutdown indication or the model update indication to the corresponding receiving node. When receiving the model shutdown indication, the receiving node can shut down the perception model configured by itself and report the shutdown status to the SF node for feedback; when receiving the model update indication, the receiving node updates the perception model configured by itself and reports the update status to the SF node for feedback after the update is completed, so that the SF node can synchronize the model working status of each node in the ISAC system.

[0135] It is understood that the aforementioned updates to the perception models configured on the receiving nodes themselves are generally achieved through model retraining. If the receiving node's own data processing capabilities are sufficient to meet the model retraining requirements, the receiving node can complete the retraining of its own perception model. However, since some receiving nodes are low-capability and lack the ability to collect data or train models, other nodes in the ISAC system can assist the receiving node in updating its perception model through the following two methods.

[0136] In one embodiment, updating the perception model configured by the receiving node itself includes:

[0137] Sending the perception model configured by the receiving node itself to an update auxiliary node, so as to construct a perception model to be updated that is the same as the perception model configured by the receiving node itself in the update auxiliary node;

[0138] receiving a model alignment instruction sent by the update auxiliary node;

[0139] Sending the channel impulse response matrices corresponding to different time units to the update auxiliary node, so that the update auxiliary node retrains the perception model to be updated through each of the channel impulse response matrices;

[0140] Receiving the updated model parameters fed back by the update auxiliary node, and updating the updated model parameters into the perception model configured by the receiving node itself;

[0141] The update auxiliary node is a node in the synaesthesia integration system that has the model retraining capability except the receiving node.

[0142] In this embodiment, the update auxiliary node can be specifically understood as a node in the ISAC system that can undertake model retraining data processing, except for the receiving node that needs to update the perception model. It can be other receiving nodes in the ISAC system, or it can be an SF node or a sending node. The embodiment of this application does not limit this.

[0143] In this embodiment, the model alignment indication can be specifically understood as indication information sent by the update auxiliary node to the receiving node that needs to update the perception model, which is used to synchronize the required updated perception model to the update auxiliary node.

[0144] Specifically, when the receiving node determines that it needs to update its own configured perception model, it can select a node with model retraining capabilities from the ISAC system as an update auxiliary node, and send the perception model that needs to be updated in its own configuration to the update auxiliary node, so that the update auxiliary node can construct a perception model to be updated that is completely consistent with the required updated perception model, and complete the alignment of the two models after receiving the model alignment instruction sent by the update auxiliary node. At the same time, after starting the model retraining, the receiving node will send its own CIR matrix determined in different time units to the update auxiliary node, so that the update auxiliary node can complete the retraining of the perception model to be updated after synchronization through each CIR matrix. After completing the retraining of the perception model to be updated, the update auxiliary node will feed back the parameters of the perception model to be updated after retraining as the update model parameters to the receiving node. The receiving node replaces the parameters of the corresponding position of the perception model configured by itself with the updated model parameters, thereby updating the perception model configured by itself.

[0145] In one embodiment, updating the perception model configured by the receiving node itself includes:

[0146] Sending the perception model configured by the receiving node itself to the updating auxiliary node, so as to construct the perception model to be updated that is the same as the perception model configured by the receiving node itself in the updating auxiliary node;

[0147] Receive a model alignment instruction sent by an update auxiliary node;

[0148] Sending the model output of the perception model configured by the receiving node itself to the update auxiliary node, so that the update auxiliary node performs reverse derivation on the model output to determine the updated model parameters, and updates the updated model parameters to the perception model to be updated;

[0149] receiving and updating the updated model parameters fed back by the update auxiliary node into the perception model configured by the receiving node itself, and returning to the step of sending the model output of the perception model configured by the receiving node itself to the update auxiliary node until the model supervision result of the perception model configured by the receiving node itself meets the perception service quality condition;

[0150] Among them, the update auxiliary node is a node in the synaesthesia integration system that has the ability to retrain the model except the receiving node.

[0151] Specifically, in order to reduce the amount of data transmission during the model update process, the receiving node can also select a node with model retraining capabilities in the ISAC system as an update auxiliary node when determining that the perception model configured by itself needs to be updated, and send the perception model configured by itself that needs to be updated to the update auxiliary node, so that the update auxiliary node can construct a perception model to be updated that is completely consistent with the perception model to be updated, and complete the alignment of the two models after receiving the model alignment indication sent by the update auxiliary node. After model training begins, the receiving node's own perception model infers its own CIR matrix and sends the model output to the update assistant node. Since the perception model in the receiving node and the perception model to be updated in the update assistant node have identical structures and parameters, their model output should theoretically be identical to the output of the perception model to be updated for the CIR matrix. Sending the model output directly to the update assistant node reduces data transmission and computational complexity within the update assistant node. The update assistant node then reverse-derives the received model output to determine updated model parameters, applies them to the perception model to be updated, and simultaneously feeds the updated model parameters back to the receiving node. This completes the update of the receiving node's own perception model through the updated model parameters. The receiving node then infers based on the updated perception model and the new CIR matrix. If the model supervision result corresponding to the model output fails to meet QoS requirements, the receiving node repeats the above steps of sending the model output to the update assistant node for model update until QoS requirements are met. The receiving node's perception model is deemed retrained, completing the update of its own perception model.

[0152] It can be understood that in the above two perception model update methods, if the update auxiliary node is not the SF node in the ISAC system, the model alignment indication can be transmitted between the update auxiliary node and the receiving node by transparent transmission or forwarding through the SF node.

[0153] The target perception method provided in the embodiment of the present application, when it is clear that the perception model configured by the receiving node itself needs to be updated, uses a node with model retraining capability in the ISAC system to help the receiving node retrain the perception model, so that the updated perception model can meet QoS and be competent for the perception of the perception objects in the perception area, thereby improving the accuracy of the target perception results.

[0154] The following is an exemplary description of this application, illustrating the target perception, supervision, and update process of multi-model collaboration through the following embodiments:

[0155] FIG3 is an example diagram of a data transmission process of multi-model collaborative target perception provided by an embodiment of the present application. As shown in FIG3 , the data transmission of multi-model collaborative target perception in the ISAC system may include multi-model collaboration between UEs and multi-model collaboration between UEs and base stations (BSs) when the SF nodes in the ISAC system do not participate. As shown in FIG3 , a transmitting node that sends a perception reference signal, such as BS A, and three perception signal receiving nodes, such as UE A, UE B, and UE C, or UE A, BS B, and UE C, may be set within the perception area. It can be understood that any perception signal receiving node may be a UE or a BS, and the embodiment of the present application does not limit this. FIG3 only shows two examples of data transmission processes of multi-model collaborative target perception, namely, perception signal receiving nodes that are all UEs and perception signal receiving nodes that include a BS. At the same time, each perception signal receiving node contains a perception model with at least two different perception model functions.

[0156] Continuing with the above example, FIG4 is a diagram illustrating an example of a data transmission process for multi-model collaborative target perception provided by an embodiment of the present application. A receiving node equipped with a perception model capable of estimating target position and velocity is determined as a first receiving node for the perception signal, and a receiving node equipped with a perception model capable of estimating the number of targets is determined as a second receiving node for the perception signal. The specific process of target perception is as follows:

[0157] 1. The SF node in the ISAC system first sends a signal to the sensing signal transmitting node and each sensing signal receiving node.

[0158] Among them, the perception configuration information may include information such as the range that needs to be perceived, the target type, and the upper limit of the number of perceived targets.

[0159] 2. After receiving the sensing configuration information, the sensing signal transmitting node configures the parameters and transmits the sensing reference signal.

[0160] 3. Each sensing signal receiving node measures and generates a CIR matrix after receiving the reflected signal reflected by the sensing target and the environment.

[0161] 4. The second receiving node of the sensing signal sends the CIR matrix generated by its own measurement to its corresponding sensing model, and sends the target number estimation result obtained by the perception model output as the first auxiliary information (as shown by the solid arrow in Figure 3) to the first receiving node of the sensing signal.

[0162] 5. After receiving the first auxiliary information, the first receiving node of the perception signal adjusts the number of output layer neurons of its own configured perception model according to the first auxiliary information, and sends the CIR matrix generated by its own measurement to the adjusted perception model. The perception model output, such as the target position and speed estimation results, is sent as the second auxiliary information (as shown by the dotted arrow in Figure 3) to the second receiving node of the perception signal.

[0163] 6. The first receiving node and the second receiving node integrate the perception signals to generate a target perception result, and the perception ends.

[0164] It can be understood that the first auxiliary information and the second auxiliary information are transmitted between the first sensing signal receiving node and the second sensing signal receiving node using a direct transparent transmission method.

[0165] FIG5 is another example diagram of a data transmission process for multi-model collaborative target perception provided by an embodiment of the present application. As shown in FIG5 , the data transmission for multi-model collaborative target perception in the ISAC system includes multi-model collaboration between UEs and multi-model collaboration between UEs and BSs when SF nodes in the ISAC system participate. As shown in FIG5 , a transmitting node that sends a sensing reference signal, such as BS A, and three sensing signal receiving nodes, such as UE A, UE B, and UE C, or UE A, BS B, and UE C, can be set up within the sensing area. It is understood that any sensing signal receiving node can be a UE or a BS, and the present embodiment is not limited to this. FIG5 only shows two examples of multi-model collaborative target perception data transmission processes, one in which all sensing signal receiving nodes are UEs, and the other in which a sensing signal receiving node is a BS. At the same time, each sensing signal receiving node contains at least two sensing models with different sensing model functions. In the multi-model collaborative target perception data transmission process, each sensing signal receiving node needs to first send the generated auxiliary information to the SF node on the network side. The SF node then distributes the auxiliary information and then sends it to the designated sensing signal receiving node.

[0166] Continuing with the above example, FIG6 is another exemplary diagram of a data transmission process for multi-model collaborative target perception provided by an embodiment of the present application. A receiving node equipped with a perception model capable of estimating target position and velocity is determined as a first receiving node for the perception signal, and a receiving node equipped with a perception model capable of estimating the number of targets is determined as a second receiving node for the perception signal. The specific process of target perception is as follows:

[0167] 1. The SF node in the ISAC system first sends the sensing configuration to the sensing signal transmitting node and each sensing signal receiving node.

[0168] 2. After receiving the sensing configuration information, the sensing signal transmitting node configures the parameters and transmits the sensing reference signal.

[0169] 3. Each sensing signal receiving node measures and generates a CIR matrix after receiving the reflected signal reflected by the sensing target and the environment.

[0170] 4. The second receiving node of the sensing signal sends the CIR matrix generated by its own measurement to its corresponding sensing model, and sends the target number estimation result obtained by the sensing model output as the first sensing information (as shown by the solid arrow in Figure 5) to the SF node.

[0171] 5. After receiving the first perception information, the SF node sends first auxiliary information to the second perception signal receiving node (as shown by the solid arrow in Figure 5), where the content of the first auxiliary information may include but is not limited to the first perception information.

[0172] 6. After receiving the first auxiliary information sent by the SF node, the second receiving node of the perception signal adjusts the number of output layer neurons of its own configured perception model according to the content of the first auxiliary information, and sends the CIR matrix generated by its own measurement to the adjusted perception model, and sends the target position and speed estimation results obtained by the perception model output as the second perception information to the SF node (as shown by the dotted arrow in Figure 5).

[0173] 7. After receiving the second perception information, the SF node sends second auxiliary information to the first receiving node of the perception signal (as shown by the dotted arrow in Figure 5), where the content of the second auxiliary information may include but is not limited to the second perception information.

[0174] 8. The first receiving node for sensing the signal and the second receiving node for sensing the signal integrate to generate a target sensing result, and the sensing ends.

[0175] It is understood that after the perception process begins, each perception signal receiving node must first report the perception model number or perception model function of its deployed perception model to the SF node. This allows the SF node to determine the model collaboration mode for multiple models in the ISAC system based on the perception model function of each perception signal receiving node and transmit this information to each perception signal receiving node via perception configuration information. For example, Figure 7 illustrates an example process for reporting model information deployed by a perception signal receiving node, as provided in an embodiment of the present application.

[0176] Optionally, perception model functions may include target distance estimation, angle estimation, speed estimation, and target number estimation. When reporting deployment model-related information to the SF node, the perception signal receiving node may report all or one of the perception model numbers and perception model functions. When the SF node informs the perception signal receiving node of the model collaboration mode via perception configuration information, the perception configuration information should at least include the perception model numbers and perception model functions of all participating collaborative perception models, as well as the model collaboration order of each perception model.

[0177] FIG8 is an example diagram of a data transmission process for multi-model collaborative supervision provided in an embodiment of the present application. As shown in FIG8 , for each sensing signal receiving node in the ISAC system, model supervision must be performed on the sensing model configured therein during the target sensing process. When the model supervision mode is multi-model collaborative supervision, a node in the ISAC system must be selected as the supervisory node for the supervised sensing model. The supervisory node can be an SF node or another sensing signal receiving node in the ISAC system. As shown in FIG8 , during the multi-model collaborative supervision process, assuming that a sensing reference signal transmitting node BS A is set within the sensing area, the sensing model to be supervised is set on the sensing signal receiving node UE A, and the supervisory node is a UE / BS or SF node serving as a sensing signal receiving node in the ISAC system. When performing multi-model collaborative supervision, UE A must report the CIR matrix generated by its own reflection signal measurement to the supervisory node (as shown by the solid line in FIG8 ) and receive supervision auxiliary information fed back by the supervisory node regarding the received CIR matrix (as shown by the dotted line in FIG8 ).

[0178] Continuing with the above example, FIG9 is an example diagram of a data transmission process for multi-model collaborative supervision provided in an embodiment of the present application. As shown in FIG9 , the specific process is as follows:

[0179] 1. The SF node first sends the sensing configuration information to each sensing signal transmitting node and sensing signal receiving node.

[0180] 2. After receiving the sensing configuration information, the sensing signal transmitting node configures the parameters and transmits the sensing reference signal.

[0181] 3. The sensing signal receiving node measures and generates a CIR matrix after receiving the reflected signal reflected by the sensing target and the environment.

[0182] 4. The sensing signal receiving node reports the CIR matrix generated by its own measurement to the supervisory node. At the same time, the sensing signal receiving node accumulates the CIR matrix corresponding to this data.

[0183] 5. After receiving the reported CIR matrix, the supervisory node sends the received CIR matrix to its own configured supervisory model and obtains the supervisory model output result of the supervisory model.

[0184] The supervisory model is a perception model that has functional overlap with or can perform functional conversion with the perception model configured in the perception signal receiving node.

[0185] 6. After accumulating CIR matrices for a certain number of time units, the perception model receiving node sends these CIR matrices to its own configured perception model to obtain the output results to be supervised.

[0186] 7. The supervision node will process the output results of each supervision model to obtain supervision auxiliary information of the same type as the output result to be supervised, and send the supervision auxiliary information to the perception model receiving node.

[0187] 8. The perception model receiving node compares the received supervision auxiliary information with its own supervised output results, and determines whether the perception model configured by itself is mismatched with the environment based on the comparison results, and obtains the model supervision results.

[0188] It can be understood that the outputs of the perception model and the supervision model that need to be supervised in the perception signal receiving node can be all perception results or part of the perception results. For example, the output result of the perception model that needs to be supervised can be one or all of the number, position or speed of perception targets; the output result of the supervision model can also be one or all of the number, position or speed of perception targets.

[0189] After determining the model supervision results, it can be determined based on the model supervision results whether the perception model configured in the perception signal receiving node is mismatched with the environment. When there is a mismatch with the environment, the perception model needs to be processed, such as shutting down or updating, to ensure the accuracy of the ISAC system during target perception. The shutting down or updating of the perception model can be performed by the perception signal receiving node itself, or it can be assisted by the SF node in the ISAC system. Figure 10 is an example diagram of a process for shutting down or updating a perception model provided in an embodiment of the present application. As shown in Figure 10, when the shutting down or updating of the perception model is performed by the perception signal receiving node itself, it can specifically include the following steps:

[0190] 1. The SF node sends sensing requirement information to the sensing signal receiving node. The sensing requirement information includes QoS, which may include sensing distance accuracy, angle accuracy, detection probability, and false alarm probability.

[0191] 2. The perception signal receiving node performs multi-model collaborative supervision or radar cross-section supervision to obtain the model supervision result.

[0192] 3. The perception signal receiving node shuts down the perception model or updates its parameters based on the model supervision results and QoS.

[0193] 4. The perception signal receiving node reports the shutdown or update status of the perception model to the SF node.

[0194] FIG11 is another example flow diagram of shutting down or updating a perception model according to an embodiment of the present application. As shown in FIG11 , when the shutting down or updating process of the perception model is performed with the assistance of the SF node, the process may specifically include the following steps:

[0195] 1. The perception signal receiving node reports the model supervision results determined by itself to the SF node.

[0196] 2. After receiving the model supervision result, the SF node sends a model shutdown or update instruction to the perception signal receiving node according to QoS.

[0197] 3. After receiving the model shutdown indication or model update indication, the perception signal receiving node shuts down or updates the perception model, and reports the shutdown or update status of the perception model to the SF node.

[0198] It is understandable that model updates are generally achieved by retraining the perception model, but for some perception signal receiving nodes, such as some base stations or low-capability terminals (RedCap UE), they may not have the ability to collect data or train models. Therefore, they may need to rely on other network elements in the ISAC system, such as SF nodes, to achieve model parameter updates.

[0199] Figure 12 is an example diagram of a perception model update process provided by an embodiment of the present application. As shown in Figure 12, a perception signal receiving node deployed with a perception model that needs to be updated, after receiving a model parameter update indication sent by an SF node, first sends the perception model that needs to be updated to the SF node through Operation Administration and Maintenance (OAM) or OTT service. After receiving the perception model, the SF node performs model alignment and sends a model alignment indication to the perception signal receiving node; the perception signal receiving node will then report a large number of CIR matrices corresponding to different time units; after receiving these CIR matrices, the SF node retrains the perception model that needs to be updated, and after the retraining is completed, sends the updated model parameters to the perception signal receiving node, so that the perception signal receiving node can update the updated model parameters to its own configured perception model, and the model parameter update is completed.

[0200] FIG13 is a flowchart illustrating another process for updating a perception model provided by an embodiment of the present application. As shown in FIG13 , after receiving a model parameter update indication from an SF node, the perception signal receiving node first sends the perception model to be updated to the SF node via OAM or OTT services. After receiving the perception model, the SF node performs model alignment and sends a model alignment indication to the perception signal receiving node. After receiving the model alignment indication, the perception signal receiving node continues to use its configured perception model for inference and reports the output perception result as model output 1 to the SF node. After receiving model output 1, the SF node reverse-derives the model output to obtain updated parameter 1. The SF node then sends updated parameter 1 to the perception signal receiving node. The perception signal receiving node updates the perception model using updated parameter 1 and outputs model output 2. The above steps are repeated until the model supervision result meets the QoS requirements, and the model parameter update process of the perception model receiving node ends.

[0201] For example, the schemes for multi-model collaborative target perception, model supervision, and updating can be integrated through the following steps:

[0202] 1. After the sensing process starts, the sensing signal receiving node first reports at least one of the sensing model number and sensing model function deployed on its own network element to the SF node.

[0203] 2. After receiving the perception model number or perception model function reported by each perception signal receiving node, the SF node sends the perception configuration information and QoS to the perception signal transmitting node and each perception signal receiving node.

[0204] 3. After receiving the perception configuration information and QoS, the perception signal transmitting node completes the configuration and transmits the perception reference signal.

[0205] 4. The sensing signal receiving node receives the target reflected signal and generates a CIR matrix.

[0206] 5. After receiving the perception configuration information, the perception signal receiving node determines the perception model number and model collaboration order that needs to be coordinated for inference.

[0207] 6. According to the model cooperation order, the perception signal receiving node serving as the second perception signal receiving node sends the first auxiliary information to the first perception signal receiving node, or sends the first perception content to the SF node.

[0208] 7. The first sensing signal receiving node sends the second auxiliary information to the second sensing signal receiving node or sends the second sensing content to the SF node.

[0209] 8. If the SF node receives the perception content sent by the first and second perception signal nodes, the SF node will send the first perception content as the first auxiliary information to the first perception signal receiving node, and send the second perception content as the second auxiliary information to the second perception signal receiving node.

[0210] 9. The perception signal receiving node supervises its own configured perception model according to the model supervision mode signaling contained in the QoS and perception configuration information.

[0211] 10. If the perception signal receiving node uses RCS to supervise the perception model, the perception signal receiving node will compare the RCS output by its own configured perception model with the RCS value range information in the perception configuration information, and report the model supervision result to the SF node; if the perception signal receiving node uses multi-model collaborative supervision for supervision, the perception signal receiving node will report the CIR matrix to the SF node or other supervision model deployment network element, and the SF node or other supervision model deployment network element will send supervision auxiliary information to the perception signal receiving node. The perception signal receiving node uses the supervision auxiliary information to supervise the perception model and reports the model supervision result to the SF node.

[0212] 11. The perception signal receiving node maintains, shuts down or updates the perception model based on its own QoS judgment or the signaling sent by the SF node.

[0213] 12. If the perception signal receiving node needs to update the perception model, it can update the perception model in any of the following ways:

[0214] 1) The sensing signal receiving node itself retrains the sensing model to update the sensing model;

[0215] 2) The sensing signal receiving node sends a certain number of CIR matrices to the SF node or other network elements that can update the perception model parameters. After receiving a certain number of CIR matrices, the SF node or other network element updates the perception model parameters and sends the updated parameters to the sensing signal receiving node;

[0216] 3) The perception signal receiving node sends a model output result to the SF node or other network element that can update the perception model parameters. After receiving the model output result, the SF node or other network element updates the model parameters and sends the updated parameters to the perception signal receiving node. The perception signal receiving node uses the updated parameters to obtain the model output result again and resends the model output result at this time to the SF node or other network element. The above process is repeated until the model parameter update is completed.

[0217] In an exemplary embodiment, FIG14 is a schematic diagram of the structure of a target sensing device provided in an embodiment of the present application. The target sensing device is applied to a receiving node in an ISAC system. As shown in FIG14 , the device includes:

[0218] A signal receiving module 310 is configured to receive a target reflected signal;

[0219] A perception result determination module 320 is configured to determine a target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the received auxiliary information;

[0220] The perception model function of the receiving node corresponding to the auxiliary information is different from the perception model function of the perception model configured by the receiving node itself.

[0221] The target perception device provided by the embodiment of the present application sets the perception models with different perception model functions in different receiving nodes of the synaesthesia integration system, so that after each receiving node in the synaesthesia integration system receives the target reflection signal reflected by the desired perception target, it can determine the partial perception information of the desired perception target based on the perception model configured by itself, and then determine the final desired target perception result in combination with the other types of perception information received by other receiving nodes for the desired perception target. Since the perception models for realizing different functions are respectively set on different nodes in the synaesthesia integration system, the volume of the perception model configured on each receiving node is small, which reduces the training overhead of the perception model corresponding to each receiving node. At the same time, since different receiving nodes in the synaesthesia integration system perceive the desired perception target from different angles respectively, the prior knowledge required for the perception models of different perception model functions is different, and there is no need to provide complete prior knowledge for each perception model. Even if the prior knowledge of some perception models is missing, the remaining perception models can still give relatively accurate partial perception results, so that the target perception result finally combined with the generated one can still maintain a high accuracy, thereby achieving stable perception of the target and environment within the desired perception area, reducing the requirements of AI perception for environmental prior information, and improving perception accuracy.

[0222] In one embodiment, before receiving the target reflected signal, the method further includes:

[0223] Reporting at least one of a perception model function and a perception model number of the perception model configured by the receiving node itself to the network side node;

[0224] Receives the perception configuration information sent by the network side node.

[0225] In one embodiment, determining a target perception result based on a target reflection signal, a perception model configured by a receiving node, and received auxiliary information includes:

[0226] Obtain multi-model collaboration mode signaling from the perception configuration information;

[0227] Determine the target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the multi-model cooperation mode signaling;

[0228] Among them, the multi-model collaboration mode signaling includes the model collaboration order of each perception model participating in the collaboration, and at least one of the perception model function and the perception model number.

[0229] In one embodiment, when the perception models participating in the collaboration include a perception model whose perception model function is to estimate the number of targets, determining a target perception result based on a target reflection signal, a perception model configured by the receiving node itself, and auxiliary information received based on multi-model collaboration mode signaling includes:

[0230] Determine the target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the model cooperation sequence;

[0231] The auxiliary information includes first auxiliary information and second auxiliary information.

[0232] In one embodiment, when the perception model function of the perception model configured in the receiving node itself is target number estimation, the perception model corresponding to the receiving node is placed first in the model cooperation order; accordingly, determining the target perception result based on the target reflection signal, the perception model configured in the receiving node itself, and the auxiliary information received according to the model cooperation order includes:

[0233] determining a channel impulse response matrix based on the target reflected signal;

[0234] The channel impulse response matrix is ​​input into the perception model configured by the receiving node itself to determine the target number estimation result;

[0235] Sending the target number estimation result to other receiving nodes in the synaesthesia integration system, so that the target number estimation result is used as the first auxiliary information received by each other receiving node;

[0236] receiving second auxiliary information sent by each other receiving node;

[0237] The target number estimation result and each second auxiliary information are combined to determine the target perception result.

[0238] In one embodiment, when the perception model function of the perception model configured in the receiving node itself is not target number estimation, the perception model whose perception model function is target number estimation is placed first in the model cooperation order; accordingly, determining the target perception result based on the target reflection signal, the perception model configured in the receiving node itself, and the auxiliary information received according to the model cooperation order includes:

[0239] determining a channel impulse response matrix based on the target reflected signal;

[0240] receiving first auxiliary information sent by a receiving node whose perception model function is target number estimation; wherein the first auxiliary information is a target number estimation result;

[0241] adjusting a perception model configured by the receiving node itself according to the first auxiliary information, and inputting the channel impulse response matrix into the adjusted perception model to determine a partial perception result;

[0242] Sending the partial perception results to other receiving nodes in the synaesthesia integration system so that the partial perception results are used as second auxiliary information received by each of the other receiving nodes;

[0243] receiving second auxiliary information sent by each receiving node other than the receiving node for which the perception model function is target number estimation;

[0244] The first auxiliary information, the partial perception result and each second auxiliary information are combined to determine the target perception result.

[0245] In one embodiment, when the perception models participating in the collaboration do not include a perception model whose perception model function is to estimate the number of targets, determining a target perception result based on a target reflection signal, a perception model configured by the receiving node itself, and auxiliary information received based on multi-model collaboration mode signaling includes:

[0246] determining a channel impulse response matrix based on the target reflected signal;

[0247] Input the channel impulse response matrix into the perception model configured by the receiving node itself to determine the partial perception results;

[0248] Receive auxiliary information sent by other receiving nodes in the synaesthesia integration system according to the model cooperation order in the multi-model cooperation mode signaling, and send part of the perception results to each other receiving node, so as to use the part of the perception results as auxiliary information received by each other receiving node;

[0249] The partial perception results are combined with various auxiliary information to determine the target perception results.

[0250] In one embodiment, the auxiliary information is information transparently transmitted from other receiving nodes in the synaesthesia integration system to the receiving node; or the auxiliary information is information forwarded from other receiving nodes in the synaesthesia integration system to the receiving node via network-side node management.

[0251] In one embodiment, when determining the target perception result based on the target reflection signal, the perception model configured by the receiving node itself, and the received auxiliary information, the method further includes:

[0252] Obtain model supervision mode signaling from the perception configuration information;

[0253] determining a model supervision mode of a receiving node according to the model supervision mode signaling;

[0254] Perform model supervision on the perception model configured by the receiving node itself through model supervision to determine the model supervision result;

[0255] Among them, the model supervision method includes at least one of the following: multi-model collaborative supervision and radar cross-section supervision.

[0256] In one embodiment, when the model supervision mode is multi-model collaborative supervision, model supervision is performed on the perception model configured by the receiving node itself through the model supervision mode to determine the model supervision result, including:

[0257] Determine, based on the perception configuration information, a supervision model corresponding to the perception model configured by the receiving node itself;

[0258] Send the channel impulse response matrix determined according to the target reflection signal to the supervisory node corresponding to the supervisory model, and accumulate the channel impulse response matrix;

[0259] When the cumulative time unit quantity corresponding to each channel impulse response matrix reaches a preset cumulative threshold, each channel impulse response matrix is ​​input into the perception model configured by the receiving node itself to determine the output result to be supervised;

[0260] Receive supervisory auxiliary information fed back by the supervisory node;

[0261] Compare the output result to be supervised with the auxiliary information of supervision, and determine the model supervision result based on the comparison result;

[0262] The supervisory model is a perception model that has functional overlap or functional conversion with the perception model configured by the receiving node itself.

[0263] In one embodiment, the supervisory auxiliary information is the output result output by the supervisory model after the impulse response matrix of each channel is input respectively; or the supervisory auxiliary information is the information obtained after data processing of the output result output by the supervisory model after the impulse response matrix of each channel is input respectively, and the information is consistent with the type of the output result to be supervised.

[0264] In one embodiment, the supervisory node is another receiving node or a network-side node in the synaesthesia integration system.

[0265] In one embodiment, when the model supervision mode is radar cross section supervision, model supervision is performed on the perception model configured by the receiving node itself through the model supervision mode to determine the model supervision result, including:

[0266] Determine radar cross section value range information based on perception configuration information;

[0267] The channel impulse response matrix determined based on the target reflection signal is input into the perception model configured by the receiving node itself to determine the target radar cross section value;

[0268] The target radar cross section value is compared with the radar cross section value range information, and the model supervision result is determined based on the comparison result.

[0269] In one embodiment, the radar cross section value range information includes radar cross section value ranges for different types of perception targets.

[0270] In one embodiment, after performing model supervision on the perception model configured by the receiving node itself through the model supervision method and determining the model supervision result, the method further includes:

[0271] Obtaining the perceived service quality condition from the perceived configuration information;

[0272] When the model supervision result does not meet the perception service quality conditions, the perception model configured by the receiving node itself is shut down or updated, and the shutdown status or update status is reported to the network side node.

[0273] In one embodiment, after performing model supervision on the perception model configured by the receiving node itself through the model supervision method and determining the model supervision result, the method further includes:

[0274] Report the model supervision results to the network side node;

[0275] Receiving a model shutdown indication or a model update indication fed back by a network-side node;

[0276] Shutting down the perception model configured by the receiving node itself according to the model shutdown indication, or updating the perception model configured by the receiving node itself according to the model update indication;

[0277] Report the shutdown status or update status to the network side node.

[0278] In one embodiment, updating the perception model configured by the receiving node itself includes:

[0279] Sending the perception model configured by the receiving node itself to the updating auxiliary node, so as to construct the perception model to be updated that is the same as the perception model configured by the receiving node itself in the updating auxiliary node;

[0280] Receive a model alignment instruction sent by an update auxiliary node;

[0281] Sending the channel impulse response matrices corresponding to different time units to the update auxiliary node, so that the update auxiliary node retrains the perception model to be updated using each channel impulse response matrix;

[0282] Receive the updated model parameters fed back by the update auxiliary node, and update the updated model parameters to the perception model configured by the receiving node itself;

[0283] Among them, the update auxiliary node is a node in the synaesthesia integration system that has the ability to retrain the model except the receiving node.

[0284] In one embodiment, updating the perception model configured by the receiving node itself includes:

[0285] Sending the perception model configured by the receiving node itself to the updating auxiliary node, so as to construct the perception model to be updated that is the same as the perception model configured by the receiving node itself in the updating auxiliary node;

[0286] Receive a model alignment instruction sent by an update auxiliary node;

[0287] Sending the model output of the perception model configured by the receiving node itself to the update auxiliary node, so that the update auxiliary node performs reverse derivation on the model output to determine the updated model parameters, and updates the updated model parameters to the perception model to be updated;

[0288] receiving and updating the updated model parameters fed back by the update auxiliary node into the perception model configured by the receiving node itself, and returning to the step of sending the model output of the perception model configured by the receiving node itself to the update auxiliary node until the model supervision result of the perception model configured by the receiving node itself meets the perception service quality condition;

[0289] Among them, the update auxiliary node is a node in the synaesthesia integration system that has the ability to retrain the model except the receiving node.

[0290] The target perception device proposed in this embodiment and the target perception method proposed in the above embodiment belong to the same inventive concept. Technical details not fully described in this embodiment can be referred to any of the above embodiments, and this embodiment has the same beneficial effects as executing the target perception method.

[0291] An embodiment of the present application also provides a communication node. Figure 15 is a structural diagram of a communication node provided by an embodiment of the present application. As shown in Figure 15, the communication node provided by an embodiment of the present application includes a memory 420, a processor 410, and a computer program stored in the memory and executable on the processor. When the processor 410 executes the program, the above-mentioned target perception method is implemented.

[0292] The communication node may also include a memory 420; the processor 410 in the communication node may be one or more, and Figure 15 takes one processor 410 as an example; the memory 420 is used to store one or more programs; the one or more programs are executed by the one or more processors 410, so that the one or more processors 410 implement the target perception method as described in the embodiment of the present application.

[0293] The communication node further includes: a communication device 430 , an input device 440 and an output device 450 .

[0294] The processor 410, memory 420, communication device 430, input device 440 and output device 450 in the communication node may be connected via a bus or other means. FIG15 takes the bus connection as an example.

[0295] The input device 440 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the communication node. The output device 450 may include a display device such as a display screen.

[0296] The communication device 430 may include a receiver and a transmitter. The communication device 430 is configured to perform information transmission and reception communication according to the control of the processor 410.

[0297] The memory 420, as a computer-readable storage medium, can be configured to store software programs, computer executable programs, and modules, such as the program instructions / modules corresponding to the target perception method described in the embodiment of the present application (for example, the signal receiving module 310 and the perception result determination module 320 in the target perception device). The memory 420 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the communication node, etc. In addition, the memory 420 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 420 may further include a memory remotely arranged relative to the processor 410, and these remote memories may be connected to the communication node via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0298] An embodiment of the present application also provides a storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the target perception method described in any one of the embodiments of the present application.

[0299] Optionally, the target perception method is applied to a receiving node in an ISAC system, including: receiving a target reflection signal; determining a target perception result based on the target reflection signal, a perception model configured by the receiving node itself, and received auxiliary information; wherein the perception model function of the receiving node corresponding to the auxiliary information is different from the perception model function of the perception model configured by the receiving node itself.

[0300] The computer storage medium of the embodiment of the present application can adopt any combination of one or more computer-readable media.Computer-readable media can be computer-readable signal media or computer-readable storage media.Computer-readable storage media can be, for example, but not limited to: electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination of the above.More specific examples (non-exhaustive list) of computer-readable storage media include: electrical connections with one or more wires, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM), flash memories, optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.Computer-readable storage media can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0301] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0302] The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, radio frequency (RF), etc., or any suitable combination of the foregoing.

[0303] The computer program code for performing the operations of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., using an Internet service provider to connect via the Internet).

[0304] Optionally, an embodiment of the present application further provides a computer program product, including a computer program, which, when executed by a processor, implements the target perception method provided in any embodiment of the present application.

[0305] The above description is merely an exemplary embodiment of the present application and is not intended to limit the scope of protection of the present application.

[0306] It will be appreciated by those skilled in the art that the term user terminal covers any suitable type of wireless user equipment, such as a mobile phone, a portable data processing device, a portable web browser or a vehicle-mounted mobile station.

[0307] In general, various embodiments of the present application may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. For example, some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device, although the present application is not limited thereto.

[0308] Embodiments of the present application may be implemented by executing computer program instructions by a data processor of a mobile device, for example, in a processor entity, or by hardware, or by a combination of software and hardware. The computer program instructions may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages.

[0309] The block diagram of any logical flow in the drawings of this application may represent program steps, or may represent interconnected logical circuits, modules and functions, or may represent a combination of program steps and logical circuits, modules and functions. A computer program may be stored on a memory. The memory may be of any type suitable for the local technical environment and may be implemented using any suitable data storage technology, such as but not limited to read-only memory (ROM), random access memory (RAM), optical storage devices and systems (digital versatile discs (DVD) or compact disks (CD), etc.). Computer-readable media may include non-transitory storage media. The data processor may be of any type suitable for the local technical environment, such as but not limited to a general-purpose computer, a special-purpose computer, a microprocessor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), and a processor based on a multi-core processor architecture.

[0310] The above description of exemplary embodiments of the present application has been provided by way of exemplary and non-limiting examples. However, various modifications and adaptations of the above embodiments will be apparent to those skilled in the art, when considered in conjunction with the accompanying drawings and the appended claims, without departing from the scope of the present application. Therefore, the proper scope of the present application will be determined by reference to the appended claims.

Claims

1. A target perception method, applied to a receiving node in a communication-sensing integrated system, includes: Receiving a target reflected signal; Determining a target perception result according to the target reflected signal, the perception model configured by the receiving node itself, and the received auxiliary information; Wherein, the perception model function corresponding to the auxiliary information of the receiving node is different from the perception model function of the perception model configured by the receiving node itself.

2. According to the target perception method described in claim 1, before receiving the target reflected signal, it further includes: Reporting at least one of the perception model function and the perception model number of the perception model configured by the receiving node itself to the network-side node; Receiving the perception configuration information sent by the network-side node.

3. The target perception method according to claim 2, wherein The determining the target perception result according to the target reflected signal, the perception model configured by the receiving node itself, and the received auxiliary information includes: Obtaining a multi-model cooperation mode signaling from the perception configuration information; Determining a target perception result according to the target reflected signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the multi-model cooperation mode signaling; Wherein, the multi-model cooperation mode signaling includes the model cooperation order of each perception model participating in the cooperation, and at least one of the perception model function and the perception model number.

4. The target perception method according to claim 3, wherein, When there is a perception model with a perception model function of target number estimation among the perception models participating in the cooperation, The determining the target perception result according to the target reflected signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the multi-model cooperation mode signaling includes: Determining a target perception result according to the target reflected signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the model cooperation order; Wherein, the auxiliary information includes first auxiliary information and second auxiliary information.

5. The target perception method according to claim 4, wherein, When the perception model function of the perception model configured by the receiving node itself is target number estimation, the perception model corresponding to the receiving node is at the first place in the model cooperation order; correspondingly The determining the target perception result according to the target reflected signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the model cooperation order includes: Determining a channel impulse response matrix according to the target reflected signal; Inputting the channel impulse response matrix into the perception model configured by the receiving node itself to determine a target number estimation result; Sending the target number estimation result to other receiving nodes in the communication-sensing integrated system to use the target number estimation result as the first auxiliary information received by each of the other receiving nodes; Receiving the second auxiliary information sent by each of the other receiving nodes; Combining the target number estimation result and each of the second auxiliary information to determine a target perception result.

6. The target perception method according to claim 4, wherein, When the perception model function of the perception model configured by the receiving node itself is not target number estimation, the perception model with a perception model function of target number estimation is at the first place in the model cooperation order; correspondingly Determining the target perception result according to the target reflection signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the model cooperation sequence includes: Determining a channel impulse response matrix according to the target reflection signal; Receiving first auxiliary information sent by a receiving node whose perception model function is target number estimation; wherein, the first auxiliary information is a target number estimation result; Adjusting the perception model configured by the receiving node itself according to the first auxiliary information, and inputting the channel impulse response matrix into the adjusted perception model to determine a partial perception result; Sending the partial perception result to other receiving nodes in the communication and sensing integrated system, so as to use the partial perception result as the second auxiliary information received by each of the other receiving nodes; Receiving the second auxiliary information sent by each of the other receiving nodes except the receiving node whose perception model function is target number estimation; Combining the first auxiliary information, the partial perception result, and the second auxiliary information sent by each of the other receiving nodes except the receiving node whose perception model function is target number estimation to determine the target perception result.

7. The target perception method according to claim 3, wherein, When the perception models participating in the cooperation do not include a perception model whose function is target number estimation, Determining the target perception result according to the target reflection signal, the perception model configured by the receiving node itself, and the auxiliary information received according to the multi-model cooperation mode signaling includes: Determining a channel impulse response matrix according to the target reflection signal; Inputting the channel impulse response matrix into the perception model configured by the receiving node itself to determine a partial perception result; Receiving the auxiliary information sent by other receiving nodes in the communication and sensing integrated system according to the model cooperation sequence in the multi-model cooperation mode signaling, and sending the partial perception result to each of the other receiving nodes, so as to use the partial perception result as the auxiliary information received by each of the other receiving nodes; Combining the partial perception result with the auxiliary information sent by each of the other receiving nodes to determine the target perception result.

8. The target perception method according to any one of claims 2-7, wherein, The auxiliary information is the information relayed by other receiving nodes in the communication and sensing integrated system to the receiving node; Or The auxiliary information is the information managed and forwarded by other receiving nodes in the communication and sensing integrated system to the receiving node via the network side node.

9. According to the target perception method described in claim 2, when determining the target perception result according to the target reflection signal, the perception model configured by the receiving node itself, and the received auxiliary information, it further includes: Obtaining a model supervision mode signaling from the perception configuration information; Determining the model supervision mode of the receiving node according to the model supervision mode signaling; Performing model supervision on the perception model configured by the receiving node itself through the model supervision mode to determine a model supervision result; Wherein, the model supervision mode includes at least one of the following: multi-model cooperation supervision, radar cross section supervision.

10. The target perception method according to claim 9, wherein, When the model supervision method is multi - model collaborative supervision, the model supervision of the perception model configured by the receiving node itself through the model supervision method to determine the model supervision result includes: Determine the supervision model corresponding to the perception model configured by the receiving node itself according to the perception configuration information; Send the channel impulse response matrix determined according to the target reflection signal to the supervision node corresponding to the supervision model, and accumulate the channel impulse response matrix; When the cumulative time unit quantity corresponding to each channel impulse response matrix reaches a preset cumulative threshold, input each channel impulse response matrix into the perception model configured by the receiving node itself to determine the to - be - supervised output result; Receive the supervision auxiliary information fed back by the supervision node; Compare the to - be - supervised output result with the supervision auxiliary information, and determine the model supervision result according to the comparison result; Among them, the supervision model is a perception model that has functional overlap or functional conversion with the perception model configured by the receiving node itself.

11. The target perception method according to claim 10, wherein, The supervision auxiliary information is the output result output by the supervision model after respectively inputting each channel impulse response matrix; Or The supervision auxiliary information is the information of the same type as the to - be - supervised output result obtained after data processing of the output result output by the supervision model after respectively inputting each channel impulse response matrix.

12. The target perception method according to claim 10, wherein, The supervision node is other receiving nodes or network - side nodes in the integrated communication and sensing system.

13. The target perception method according to claim 9, wherein, When the model supervision method is radar cross - section supervision, the model supervision of the perception model configured by the receiving node itself through the model supervision method to determine the model supervision result includes: Determine the radar cross - section value range information according to the perception configuration information; Input the channel impulse response matrix determined according to the target reflection signal into the perception model configured by the receiving node itself to determine the target radar cross - section value; Compare the target radar cross - section value with the radar cross - section value range information, and determine the model supervision result according to the comparison result.

14. The target perception method according to claim 13, wherein, The radar cross - section value range information includes the radar cross - section value ranges of different types of perception targets.

15. According to the target perception method described in claim 9, after the model supervision of the perception model configured by the receiving node itself through the model supervision method to determine the model supervision result, it further includes: Obtain the perception service quality condition from the perception configuration information; When the model supervision result does not meet the perception service quality condition, turn off or update the perception model configured by the receiving node itself, and report the turn - off state or update state to the network - side node.

16. According to the target perception method described in claim 9, after the model supervision of the perception model configured by the receiving node itself through the model supervision method to determine the model supervision result, it further includes: Report the model supervision result to the network - side node; Receive the model turn - off instruction or model update instruction fed back by the network - side node; Turn off the sensing model configured by the receiving node itself according to the model off indication, or update the sensing model configured by the receiving node itself according to the model update indication; Report the off state or update state to the network side node.

17. The target perception method according to claim 15 or 16, wherein, Updating the sensing model configured by the receiving node itself includes: Send the sensing model configured by the receiving node itself to the update auxiliary node to construct a to-be-updated sensing model identical to the sensing model configured by the receiving node itself in the update auxiliary node; Receive the model alignment indication sent by the update auxiliary node; Send the channel impulse response matrices corresponding to different time units to the update auxiliary node, so that the update auxiliary node retrains the to-be-updated sensing model through each of the channel impulse response matrices; Receive the updated model parameters fed back by the update auxiliary node and update the updated model parameters into the sensing model configured by the receiving node itself; Wherein, the update auxiliary node is a node in the integrated communication and sensing system that has the ability to retrain the model except the receiving node.

18. The target perception method according to claim 15 or 16, wherein, Updating the sensing model configured by the receiving node itself includes: Send the sensing model configured by the receiving node itself to the update auxiliary node to construct a to-be-updated sensing model identical to the sensing model configured by the receiving node itself in the update auxiliary node; Receive the model alignment indication sent by the update auxiliary node; Send the model output of the sensing model configured by the receiving node itself to the update auxiliary node, so that the update auxiliary node performs backpropagation to determine the updated model parameters and updates the updated model parameters into the to-be-updated sensing model; Receive and update the updated model parameters fed back by the update auxiliary node into the sensing model configured by the receiving node itself, and return to execute sending the model output of the sensing model configured by the receiving node itself to the update auxiliary node until the model supervision result of the sensing model configured by the receiving node itself meets the sensing service quality condition; Wherein, the update auxiliary node is a node in the integrated communication and sensing system that has the ability to retrain the model except the receiving node.

19. A communication node, comprising: A memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing the connection communication between the processor and the memory, wherein when the program is executed by the processor, it realizes the target sensing method according to any one of claims 1-18.

20. A storage medium for computer-readable storage, storing one or more programs, wherein the one or more programs can be executed by one or more processors to realize the target sensing method according to any one of claims 1-18.

21. A computer program product, including a computer program, wherein when the computer program is executed by a processor, it realizes the target sensing method according to any one of claims 1-18.

Images