Interference cancellation method and device for network oriented to interworking network and network equipment

Abstract

Embodiments of the present application provide a method and device for interference cancellation for a network integrated with sensing, and a network device. The method comprises: receiving a first signal; the first signal comprising: a back echo signal of a sensing signal transmitted by a first network device and an interference signal of the first network device transmitted by at least one second network device; determining an initial estimated value of a target variable according to the first signal; the target variable comprising: a sensing channel, sensing data, an interference channel and interference data; taking the first signal and the initial estimated value of the target variable as inputs of a model-driven deep unfolding neural network to obtain an iteratively optimized target variable; and performing interference cancellation on the first signal according to the iteratively optimized target variable to obtain a target sensing signal. The method for interference cancellation for a network integrated with sensing solves the problems of low efficiency and low precision of the existing method for cross-link interference cancellation.

Description

Technical Field

[0001] This application relates to the field of integrated communication and sensing technology, specifically to an interference cancellation method, apparatus, and network equipment for integrated communication and sensing networks. Background Technology

[0002] In an integrated sensing and communication (ISAC) networking environment, sensing performance faces a severe challenge from strong cross-link interference (CLI) from neighboring communication base stations.

[0003] Existing Successive Interference Cancellation (SIC) methods are mostly single-stage cascaded processes. The estimation error of the interference signal in each stage inevitably propagates to the next stage, causing a significant decrease in sensing accuracy due to error accumulation and propagation effects. While traditional statistical modeling methods (such as Sparse Bayesian Learning (SBL) or Gaussian Mixture Model (GMM)) offer some interpretability, they face difficulties in hyperparameter tuning in complex and variable network environments and are prone to getting trapped in local optima in high-dimensional non-convex optimization problems, resulting in slow convergence and poor robustness. Therefore, methods for eliminating cross-link interference suffer from low efficiency and low accuracy. Summary of the Invention

[0004] At least one embodiment of this application provides an interference cancellation method, apparatus, and network device for integrated inductive and sensor networks, which solves the problems of low efficiency and low accuracy in existing methods for eliminating cross-link interference.

[0005] To solve the above-mentioned technical problems, this application is implemented as follows:

[0006] In a first aspect, embodiments of this application provide an interference cancellation method for a sensor-integrated network, applied to a first network device, comprising:

[0007] Receive a first signal; the first signal includes: an echo signal of a sensing signal sent by the first network device and an interference signal from at least one second network device to the first network device;

[0008] Based on the first signal, an initial estimate of the target variable is determined; the target variable includes: the sensing channel, the sensing data, the interference channel, and the interference data.

[0009] The first signal and the initial estimate of the target variable are used as inputs to a model-driven deep unfolded neural network to obtain the target variable after iterative optimization.

[0010] Based on the target variable after iterative optimization, interference is eliminated on the first signal to obtain the target perception signal.

[0011] Optionally, based on the first signal, determining an initial estimate of the target variable includes:

[0012] Based on the first signal, determine the pilot sequence of the first signal;

[0013] The initial estimate of the target variable is obtained by performing variable initialization processing based on the pilot sequence.

[0014] Optionally, the interference signal includes: interference signal transmitted by the communication signal sent by the second network device via an interference channel.

[0015] Optionally, the first signal and the initial estimate of the target variable are used as inputs to a model-driven deep unfolded neural network to obtain the iteratively optimized target variable, including:

[0016] Based on the initial estimates of the target variables, the initial estimates of the sensing channel, the initial estimates of the sensing data, the initial estimates of the interference channel, and the initial estimates of the interference data are obtained.

[0017] The estimation value of the interference channel is updated and optimized based on the initial estimation value of the sensing channel, and the estimation value of the sensing channel is updated and optimized based on the updated estimation value of the interference data, until the target estimation values ​​of the sensing channel and the interference data are determined.

[0018] The estimation value of the interference data is updated and optimized based on the initial estimation value of the sensing data, and the estimation value of the sensing data is updated and optimized based on the updated estimation value of the interference channel, until the target estimation values ​​of the sensing data and the interference channel are determined.

[0019] The target estimate of the interference signal is determined based on the target estimate of the interference data and the target estimate of the interference channel.

[0020] Optionally, the model-driven deep unfolded neural network includes:

[0021] The sensing channel update module is used to update the estimated value of the sensing channel based on the echo signal and the estimated value of the sensing data, so as to obtain the reconstructed estimated value of the sensing channel.

[0022] The interference data estimation module is used to update the estimated value of the interference data based on the reconstructed estimated value of the sensing channel and the estimated value of the interference channel, so as to obtain the reconstructed estimated value of the interference data.

[0023] The sensing data update module is used to transform the reconstructed estimate of the sensing channel to obtain the updated estimate of the sensing channel.

[0024] An interference channel estimation module is used to determine an updated estimate of the interference channel based on an initial estimate of the interference data and a reconstructed estimate of the interference channel.

[0025] The sensing channel update module, the interference data estimation module, the sensing data update module, and the interference channel estimation module are sequentially passed and interactively fed back to each other, and the estimated value of the target variable is iteratively updated alternately.

[0026] Optionally, the method further includes:

[0027] A sparse denoising operator is used to denoise the reconstructed estimate of the sensing channel to obtain the denoised estimate of the sensing channel.

[0028] The reconstructed estimate of the interference data is continuously mapped to the constellation points to obtain a smoothed estimate of the interference data.

[0029] The updated estimated value of the sensing channel is fused with the reconstructed estimated value of the sensing channel to obtain the fused estimated value of the sensing channel.

[0030] The updated estimate of the interference channel is fused with the reconstructed estimate of the interference channel to obtain the fused estimate of the interference channel.

[0031] Secondly, embodiments of this application provide an interference cancellation device for an integrated sensing network, applied to a first network device, comprising:

[0032] A receiving module is configured to receive a first signal; the first signal includes: an echo signal of a sensing signal sent by the first network device and an interference signal generated by at least one second network device;

[0033] An initialization module is used to determine an initial estimate of a target variable based on the first signal; the target variable includes: a sensing channel, sensing data, an interference channel, and interference data;

[0034] The processing module is used to take the first signal and the initial estimate of the target variable as input to a model-driven deep unfolded neural network to obtain the target variable after iterative optimization.

[0035] The interference elimination module is used to eliminate interference in the first signal based on the iteratively optimized target variable to obtain the target perception signal.

[0036] Thirdly, embodiments of this application provide a network device including: a transceiver, a processor, a memory, and a program or instructions stored in the memory and executable on the processor; the processor executes the program or instructions to implement the steps of the method described above.

[0037] Fourthly, embodiments of this application provide a computer-readable storage medium storing a program that, when executed by a processor, implements the steps of the method described above.

[0038] Fifthly, embodiments of this application provide a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the method described above.

[0039] Compared with existing technologies, the interference cancellation method, apparatus, and network device for integrated sensing networks provided in this application initialize a first signal, including a sensing signal and an interference signal, to obtain an initial estimate of a target variable. Then, using a model-driven deep unfolded neural network, the initial estimate of the target variable is used to separate and reconstruct the sensing signal and the interference signal, resulting in an iteratively optimized target variable. Finally, based on the iteratively optimized target variable, the interference signal in the first signal is eliminated, and the target sensing signal is ultimately determined. The interference cancellation method for integrated sensing networks in this application solves the problems of low efficiency and low accuracy in existing methods for eliminating cross-link interference. Attached Figure Description

[0040] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0041] Figure 1 This is a schematic diagram illustrating an application scenario according to an embodiment of this application;

[0042] Figure 2 This is a flowchart illustrating the interference cancellation method for an integrated sensor network according to an embodiment of this application when applied to a first network device;

[0043] Figure 3 A schematic diagram of the system architecture for implementing the interference cancellation method for integrated sensing networks according to the embodiments of this application;

[0044] Figure 4 This is a schematic diagram of the structure of a model-driven deep unfolded neural network according to an embodiment of this application;

[0045] Figure 5 This is a schematic diagram of the modules of a model-driven deep unfolded neural network according to an embodiment of this application;

[0046] Figure 6 This is one of the comparison diagrams between the interference cancellation method for integrated sensing networks according to an embodiment of this application and the prior art;

[0047] Figure 7 This is the second comparison diagram between the interference cancellation method for integrated sensing networks according to the present application and the prior art;

[0048] Figure 8 This is Figure 3, a comparison between the interference cancellation method for integrated sensing networks according to an embodiment of this application and the prior art.

[0049] Figure 9 Figure 4 shows a comparison between the interference cancellation method for integrated sensing networks according to this application and the prior art.

[0050] Figure 10 This is a schematic diagram of the interference cancellation device for an integrated sensor network according to an embodiment of this application.

[0051] Figure 11 This is a schematic diagram of the structure of a network device according to an embodiment of this application. Detailed Implementation

[0052] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, without limiting the number of objects; for example, the first object can be one or more. Furthermore, "or" in this application indicates at least one of the connected objects. For example, "A or B" covers three scenarios: Scenario 1: including A but not B; Scenario 2: including B but not A; Scenario 3: including both A and B. The character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0053] The term "instruction" in this application can be either a direct instruction (or explicit instruction) or an indirect instruction (or implicit instruction). A direct instruction can be understood as one in which the sender explicitly informs the receiver of specific information, the operation to be performed, or the requested result, etc.; an indirect instruction can be understood as one in which the receiver determines the corresponding information based on the instruction sent by the sender, or makes a judgment and determines the operation to be performed or the requested result, etc., based on the judgment result.

[0054] It is worth noting that the technologies described in this application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA), or other systems. The terms "system" and "network" in this application are often used interchangeably, and the described technologies can be used in the systems and radio technologies mentioned above, as well as in other systems and radio technologies. The following description describes New Radio (NR) systems for illustrative purposes, and the term NR is used in most of the following description; however, these technologies can also be applied to systems other than NR systems, such as 6th Generation (6G) communication systems.

[0055] Figure 1This diagram illustrates a block diagram of a sensor-integrated system applicable to embodiments of this application. The sensor-integrated system includes a terminal 11, a first network device 12, and a second network device 13. The terminal 11 can be a mobile phone, tablet computer, laptop computer, notebook computer, personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR), virtual reality (VR) device, robot, wearable device, flight vehicle, vehicle user equipment (VUE), shipboard equipment, pedestrian user equipment (PUE), smart home devices (home appliances with wireless communication capabilities, such as refrigerators, televisions, washing machines, or furniture), game consoles, personal computers (PCs), ATMs, or self-service machines, etc. Wearable devices include: smartwatches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart chains, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. Among these, in-vehicle devices can also be referred to as in-vehicle terminals, in-vehicle controllers, in-vehicle modules, in-vehicle components, in-vehicle chips, or in-vehicle units, etc. It should be noted that the specific type of terminal 11 is not limited in this application embodiment. The first network device 12 is a sensor-integrated base station. The second network device 13 may include access network equipment or core network equipment, wherein the access network equipment may also be referred to as Radio Access Network (RAN) equipment, radio access network function, or radio access network unit. Access network equipment may include base stations, Wireless Local Area Network (WLAN) access points (APs), or Wireless Fidelity (WiFi) nodes, etc.The term "base station" can be referred to as Node B (NB), Evolved Node B (eNB), Next Generation Node B (gNB), New Radio Node B (NRNode B), Access Point, Relay Base Station (RBS), Serving Base Station (SBS), Base Transceiver Station (BTS), Radio Base Station, Radio Transceiver, Basic Service Set (BSS), Extended Service Set (ESS), Home Node B (HNB), Home Evolved Node B, Transmission Reception Point (TRP), or any other suitable term in the relevant field, as long as the same technical effect is achieved. The term "base station" is not limited to specific technical terms. It should be noted that the embodiments in this application only use base stations in NR systems as examples for introduction and do not limit the specific type of base station.

[0056] Core network equipment may include, but is not limited to, at least one of the following: core network node, core network function, Mobility Management Entity (MME), Access and Mobility Management Function (AMF), Session Management Function (SMF), User Plane Function (UPF), Policy Control Function (PCF), Policy and Charging Rules Function (PCRF), Edge Application Server Discovery Function (EASDF), Unified Data Management (UDM), Unified Data Repository (UDR), Home Subscriber Server (HSS), Centralized network configuration (CNC), Network Repository Function (NRF), Network Exposure Function (NEF), Local NEF (or L-NEF), Binding Support Function (BSF), and Application Function. Function (AF), etc. It should be noted that the embodiments of this application only use the core network equipment in the NR system as an example for introduction, and do not limit the specific type of core network equipment.

[0057] To enable those skilled in the art to better understand the embodiments of this application, the following description is provided first:

[0058] Integrated sensing and communication systems play a crucial role in numerous applications, including intelligent transportation and low-altitude economy (such as drone surveillance). In recent years, with the development of hardware technology and the increasing convergence of communication and radar frequency bands, integrated sensing and communication design has received growing attention, and many practical examples of related joint systems have emerged. However, system integration still faces numerous challenges in the dynamic environment of dense networks. To achieve efficient sensing and monitoring, base stations need to extract weak target echo signals from complex electromagnetic environments. However, in dense cellular networks, sensing base stations are subject to strong cross-link interference (CLI) from neighboring communication base stations, and target echoes are extremely weak due to bidirectional path loss, easily masked by these strong interference signals, causing traditional sensing algorithms to completely fail.

[0059] Current interference cancellation techniques have significant limitations in addressing this problem. Traditional resource allocation methods (such as time-division / frequency-division orthogonalization) can avoid interference, but they reduce the efficiency of spectrum resource utilization. Techniques based on cascaded serial interference cancellation (SIC) face serious error propagation problems. Traditional optimization algorithms, when dealing with such multivariate non-convex problems, not only have high computational complexity and slow convergence, but also heavily rely on manual adjustment of hyperparameters, making them difficult to adapt to dynamically changing wireless environments. While pure data-driven deep learning methods have strong fitting capabilities, they are considered black-box models, lacking physical interpretability, and require massive amounts of labeled data for long-term training, increasing the system's implementation cost and resource overhead.

[0060] As described in the background section, existing methods for eliminating cross-link interference suffer from low efficiency and low accuracy, which severely affects the accuracy of sensed signals. To address at least one of these problems, this application provides an interference elimination method for integrated sensing networks, which can reduce or avoid the occurrence of the above situations, improve the accuracy of sensed signals, and enhance user experience.

[0061] This application provides an interference cancellation method and apparatus for integrated sensing and communication networks. The method and apparatus are based on the same concept, and since the principles underlying the problem-solving are similar, their implementations can be referred to interchangeably; repeated details will not be repeated.

[0062] Please refer to Figure 2 This application provides an interference cancellation method for integrated sensing networks, which, when applied to a first network device, includes:

[0063] Step 201, receive a first signal; the first signal includes: the echo signal of the sensing signal sent by the first network device and the interference signal of at least one second network device to the first network device;

[0064] Step 202: Determine the initial estimated value of the target variable based on the first signal; the target variable includes: sensing channel, sensing data, interference channel, and interference data;

[0065] Step 203: Use the first signal and the initial estimate of the target variable as input to the model-driven deep unfolded neural network to obtain the target variable after iterative optimization.

[0066] Step 204: Based on the iteratively optimized target variable, perform interference cancellation on the first signal to obtain the target perception signal.

[0067] Optionally, the echo signal is the echo signal received by the first network device after it sends a sensing signal to the target terminal;

[0068] The sensing data includes: the distance between the target terminal and the first network device, and the moving speed of the target terminal.

[0069] Optionally, the interference signal includes at least one of the following: communication signals transmitted by the second network device via an interference channel, sidelobe leakage, and multipath scattering interference signals.

[0070] It should be noted that the interference cancellation method for integrated sensor networks in this application adopts the following... Figure 3 The system architecture shown is implemented as follows. The first network device is an integrated sensing and communication base station, and the second network device is a traditional communication base station. Multiple traditional communication base stations exist around the integrated sensing and communication base station. When sensing target terminals such as drones, the integrated sensing and communication base station transmits orthogonal frequency division multiplexing (OFDM) signals. It not only receives target echoes but is also subject to at least one of the following interference signals: direct transmission of signals from surrounding communication base stations via interference channels, sidelobe leakage, and multipath scattering cross-link interference signals. Specifically, the digital baseband signal (the first signal) obtained by the first network device is composed of a target echo component, cross-link interference components from neighboring communication base stations, and noise components coupled together.

[0071] In this embodiment of the application, the first signal can be represented as:

[0072]

[0073] in, This indicates the orthogonal frequency division multiplexing signal transmitted by the local integrated sensing base station. Indicates data from perception The determined sensing echo channel matrix, It should include at least the target time delay and Doppler shift, corresponding to the target range and radial velocity; Indicates the first Interference channel matrix from a neighboring communication base station to an integrated sensing base station This indicates an unknown communication symbol transmitted by a nearby communication base station. This represents additive noise. The interference term includes both the direct path component between base stations and multiple reflected path components caused by environmental scattering, thus characterizing the complex cross-link interference propagation characteristics in dense network scenarios.

[0074] Optionally, the iteratively optimized target variable includes the iteratively optimized interference signal;

[0075] Based on the iteratively optimized target variable, interference cancellation is performed on the first signal to obtain the target sensing signal, including:

[0076] Based on the iteratively optimized interference signal, interference cancellation is performed on the first signal.

[0077] The interference cancellation method for integrated sensing networks according to embodiments of this application initializes a first signal, including a sensing signal and an interference signal, to obtain an initial estimate of a target variable. Then, using a model-driven deep unfolded neural network, the initial estimate of the target variable is used to separate and reconstruct the sensing signal and the interference signal, resulting in an iteratively optimized target variable. Finally, based on the iteratively optimized target variable, the interference signal in the first signal is eliminated, and the target sensing signal is ultimately determined. The interference cancellation method for integrated sensing networks according to embodiments of this application solves the problems of low efficiency and low accuracy in existing methods for eliminating cross-link interference.

[0078] Optionally, determining an initial estimate of the target variable based on the first signal includes:

[0079] Based on the first signal, determine the pilot sequence of the first signal;

[0080] The initial estimate of the target variable is obtained by performing variable initialization processing based on the pilot sequence.

[0081] In this embodiment, the system variables of the pilot sequence of the first signal are initialized to obtain the initial state values ​​required for subsequent joint iterations; the initialization process includes four parts: sensing channel initialization, sensing data initialization, interference channel initialization, and interference data initialization.

[0082] Specifically, variable initialization processing is performed based on the pilot sequence to obtain an initial estimate of the target variable, including:

[0083] Treating the interference term in the first signal as an additional disturbance, and combining it with the pilot signals of the sensed signals transmitted by the first network device, a least-squares estimation is performed on the sensed echo signal to obtain the initial estimate of the sensed channel. ;

[0084] The initial estimates of the sensing channel are processed by a two-dimensional fast Fourier transform (2D-FFT) to extract the main peak position in the range-Doppler spectrum, thereby obtaining initial estimates of the target time delay and Doppler frequency shift. This leads to initial parameter estimates of the sensing data, including the target range and velocity. ;

[0085] Based on the initial estimate of the sensing channel and initial parameter estimation of the sensing data The initial estimate of the target echo component is reconstructed, and the echo component is subtracted from the first signal in the pilot band to obtain a residual signal containing the dominant interference component.

[0086] Using the known structure of the pilot signals from neighboring communication base stations, a least-squares estimation of the interference channel is performed to obtain an initial estimate of the interference channel. ;

[0087] Based on the initial estimate of the interference channel, preliminary zero-forcing equalization is performed on the unknown interference data to obtain the initial estimate of the interference data. ;

[0088] Finally, the state vector of the initial estimate of the target variable is obtained. The initial estimate is then used as the input for subsequent cascaded iterative layers.

[0089] The interference cancellation method for integrated sensing networks in this application embodiment enables the model-driven deep unfolded neural network to start from an initial point with clear physical meaning through the pilot-assisted initialization process of the first signal, avoiding optimization from a random state directly, thereby improving the stability and convergence efficiency of subsequent joint estimation.

[0090] It should be noted that the structure of the model-driven deep unfolded neural network is as follows: Figure 4 As shown, the network consists of L cascaded iterative layers, each corresponding to one iteration update of the traditional joint optimization algorithm; for the l-th layer, its input is the state vector output by the previous layer. The output is the updated state vector. It can be written as:

[0091] ;

[0092] in, Indicates the first Mapping function for each iteration layer, This represents the learnable parameters in this layer, including update step size, momentum factor, sparse denoising threshold, and soft slice weights.

[0093] Optionally, the first signal and the initial estimate of the target variable are used as input to a model-driven deep unfolded neural network to obtain the iteratively optimized target variable, including:

[0094] Based on the initial estimates of the target variables, the initial estimates of the sensing channel, the initial estimates of the sensing data, the initial estimates of the interference channel, and the initial estimates of the interference data are obtained.

[0095] The estimation value of the interference channel is updated and optimized based on the initial estimation value of the sensing channel, and the estimation value of the sensing channel is updated and optimized based on the updated estimation value of the interference data, until the target estimation values ​​of the sensing channel and the interference data are determined.

[0096] The estimation value of the interference data is updated and optimized based on the initial estimation value of the sensing data, and the estimation value of the sensing data is updated and optimized based on the updated estimation value of the interference channel, until the target estimation values ​​of the sensing data and the interference channel are determined.

[0097] The target estimate of the interference signal is determined based on the target estimate of the interference data and the target estimate of the interference channel.

[0098] Optionally, such as Figure 5 As shown, the model-driven deep unfolded neural network includes:

[0099] The sensing channel update module is used to update the estimated value of the sensing channel based on the echo signal and the estimated value of the sensing data, so as to obtain the reconstructed estimated value of the sensing channel.

[0100] The interference data estimation module is used to update the estimated value of the interference data based on the reconstructed estimated value of the sensing channel and the estimated value of the interference channel, so as to obtain the reconstructed estimated value of the interference data.

[0101] The sensing data update module is used to transform the reconstructed estimate of the sensing channel to obtain the updated estimate of the sensing channel.

[0102] An interference channel estimation module is used to determine an updated estimate of the interference channel based on an initial estimate of the interference data and a reconstructed estimate of the interference channel.

[0103] The sensing channel update module, the interference data estimation module, the sensing data update module, and the interference channel estimation module are sequentially passed and interactively fed back to each other, and the estimated value of the target variable is iteratively updated alternately.

[0104] Optionally, the method further includes:

[0105] A sparse denoising operator is used to denoise the reconstructed estimate of the sensing channel to obtain the denoised estimate of the sensing channel.

[0106] The reconstructed estimate of the interference data is continuously mapped to the constellation points to obtain a smoothed estimate of the interference data.

[0107] The updated estimated value of the sensing channel is fused with the reconstructed estimated value of the sensing channel to obtain the fused estimated value of the sensing channel.

[0108] The updated estimate of the interference channel is fused with the reconstructed estimate of the interference channel to obtain the fused estimate of the interference channel.

[0109] In this embodiment, each iterative layer includes four functional modules: a sensing channel update module, an interference data update module, a sensing data estimation module, and an interference channel estimation module. These four modules are used to receive signals. Sending signals Based on the previous layer's state variables, and through sequential transmission and feedback interaction between modules, the joint update of perceived features and disturbance features is achieved. The four functional modules are described in detail below:

[0110] (1) Sensing Channel Update Module:

[0111] This module is located at the front of the iterative layer and is used to estimate the results based on the perceptual data output from the previous layer. and local transmission signals Reconstruct the estimated value of the sensing channel The module calculates a correction term by combining the current received residual signal. Then, it weights and fuses the predicted and corrected values ​​using learnable fusion weights to obtain a new sensing channel estimation result. To enhance robustness under low signal-to-noise ratio conditions, this module can also introduce a sparse denoising operator to utilize the sparse prior of the sensing channel in the time-delay-Doppler domain to suppress noise and artifacts. It should be noted that the output of this module will serve as one of the inputs to the subsequent interference data update module and sensing data estimation module.

[0112] (2) Interference data estimation module:

[0113] This module receives the sensing channel estimation result output by the sensing channel update module. And combine the interference channel estimation results from the previous or current round. The module extracts soft estimates of the interference data from the received residual signal. Subsequently, it maps the soft estimates to the constellation point neighborhood using a differentiable slicing function, thereby achieving the processing of communication interference data. The smooth update is achieved. Unlike traditional hard-decision methods, this differentiable slicing function performs a continuous mapping between soft estimates and candidate constellation points, avoiding gradient interruptions and enabling end-to-end training of the entire network. It should be noted that the output of this module is used for subsequent interference channel estimation and also supports further correction of the sensing data.

[0114] (3) Sensing data update module:

[0115] This module is located in the middle of the iterative layer and is used to estimate the current sensing channel results. A two-dimensional Fast Fourier Transform (FFT) is performed to generate a range-Doppler spectrum. New estimates of target time delay and Doppler frequency shift are extracted from the spectral peak positions to obtain the estimated sensing data. To reduce parameter jitter during iteration and enhance stability in dynamic scenes, this module introduces a momentum factor during parameter updates, fusing the extracted parameter values ​​with historical estimates from the previous layer to form a smoother sensing data update result. It should be noted that the output of this module is also fed back to the sensing channel update module and the interference data update module to support joint optimization in subsequent rounds.

[0116] (4) Interference channel estimation module:

[0117] This module is located at the end of the iterative layer and is used to estimate the results based on the current interference data. The interference channel data is re-estimated using the received residual signal and then weighted and fused with the interference channel estimate from the previous layer to obtain a new interference channel estimate. When necessary, this module can also incorporate sparse priors to constrain channel estimation, thereby improving modeling accuracy in complex multipath scenarios. It should be noted that the interference channel estimate output by this module will be fed back to the next iteration layer for subsequent interference data updates and full-time-domain interference reconstruction.

[0118] Through the iterative updates of the four modules mentioned above, the sensing channel, interference data, and sensing data and interference channel are jointly optimized within a unified framework. Compared with traditional serial cascaded interference cancellation methods, the embodiments of this application do not estimate the interference separately and then perform sensing separately, but rather gradually approach a consistent joint optimal solution through mutual correction and repeated refinement between variables, thereby effectively suppressing the error propagation problem.

[0119] Furthermore, this application expands the aforementioned joint iterative process into a trainable cascaded artificial neural network structure. This maps the repeated iterations with manually set parameters in traditional algorithms to a multi-layered network with learnable parameters for forward propagation. In this process, the update step size, momentum factor, sparse threshold, and soft slice control parameters are no longer manually fixed but are transformed into trainable network weights, automatically learning optimal values ​​through training data. Since each layer inherits the physical structure of the original iterative algorithm, this network maintains the interpretability of the model-driven method while using data-driven training to compensate for the shortcomings of traditional methods in high-dimensional non-convex optimization. Compared to directly running traditional iterative algorithms, the expanded cascaded neural network can complete more efficient state updates with fewer layers, thereby accelerating convergence, improving the accuracy of disturbance reconstruction, and enhancing robustness in complex scenarios.

[0120] The interference cancellation method for integrated sensing and communication networks in this application uses four interconnected modules to alternately update and refine sensing features and interference features layer by layer, and outputs a joint estimation result. The network input is the received signal and the initialization result of system variables (the initial estimated value of the target variable), and the network output is the target estimate of the sensing channel, the target estimate of the interference data, the target estimate of the sensing data, and the target estimate of the interference channel, thereby providing a basis for subsequent interference reconstruction and time-domain cancellation.

[0121] The model-driven deep unfolded neural network in this application embodiment does not directly adopt black-box neural network fitting. Instead, it takes the iterative optimization process of joint perception and interference cancellation as a prototype, and maps each update step in the traditional iterative algorithm to a cascaded network layer structure, so that the network has data-driven adaptive optimization capability while maintaining physical interpretability.

[0122] In this embodiment, the interference signal of the full-time cross link (the target estimate of the interference signal) is reconstructed using the final output result of the model-driven deep unfolded neural network, and the interference signal is subtracted from the original received signal to realize the target echo extraction and distance and velocity parameter estimation under interference suppression conditions.

[0123] Specifically, when After each cascaded iterative layer completes the processing, the network outputs the final interference channel estimate. and interference data estimation Therefore, cross-link interference digital baseband signals can be reconstructed across the entire time domain. And subtract it point by point from the original received signal, that is:

[0124]

[0125] in, This indicates the received signal after interference has been eliminated.

[0126] The interference signal estimation process in this application covers the entire time-domain signal block. Therefore, compared to local elimination methods that only target pilot bands or some resource units, this application can achieve more thorough cross-link interference suppression. This is achieved by obtaining the de-interference signal. Subsequently, conventional sensing processing can be performed on the target, such as sensing channel estimation, 2D-FFT range-Doppler spectrum analysis, and peak detection, to extract high-precision range and velocity parameters of the target. After full-time-domain interference cancellation, the effective signal-to-interference-plus-noise ratio of the target echo is improved, thus enabling more reliable target detection and higher-precision parameter estimation under strong interference conditions.

[0127] like Figure 6 and Figure 7 As shown in the figure, the distance and velocity estimation errors in the perceived data change with the number of iterations. The horizontal axis represents the number of iterations, and the vertical axis represents the root mean square error. In the figure, "Trad" represents the traditional iterative interference cancellation algorithm, and "MDL" represents the deep unfolded cascaded network method used in this embodiment. It can be seen from the figure that under different signal-to-interference-plus-noise ratios (SNR), the error curves of the method in this application decrease faster than those of the traditional iterative method, and can reach a stable state within fewer iteration layers.

[0128] Specifically, under low signal-to-interference-plus-noise ratio (SNR) conditions, the error descent process of traditional iterative algorithms is slow and tends to plateau at higher error levels, indicating that they are more susceptible to error propagation and local stagnation in non-convex optimization under strong interference conditions. In contrast, the scheme proposed in this application maintains a faster descent trend even under low SNR conditions and eventually converges to a lower error level, demonstrating its stronger robustness to strong cross-link interference. This result shows that by introducing learnable updateable step size, momentum factor, sparse denoising threshold, and soft slicing control parameters in each unfolding layer, this application achieves adaptive optimization of traditional iterative strategies, effectively overcoming the problems of slow convergence and easy trapping in suboptimal solutions in traditional methods.

[0129] like Figure 8 and Figure 9The figure shows a performance comparison between this application and traditional 2D-FFT methods, ISAC-Net methods, Orthogonal Approximate Message Passing (OAMP) network methods, and traditional iterative interference cancellation methods. The horizontal axis represents the signal-to-interference-plus-noise ratio (SIR / N), and the vertical axis represents the root mean square error (RMSE), using a logarithmic scale to more clearly demonstrate the performance differences of different methods over a wide dynamic range. The "Perfect-SIC" in the figure represents the upper bound of performance under ideal interference cancellation conditions, serving as a reference benchmark for the method of this invention. As the SIR / N gradually increases, the distance estimation error and velocity estimation error of various methods generally show a decreasing trend, but the inflection point and final error level differ significantly among different methods. The ICSensing-Net method used in the embodiments of this application exhibits an overall error curve that is significantly better than traditional 2D-FFT methods, traditional iterative algorithms, ISAC-Net methods, and OAMP-Net methods, and consistently maintains performance close to the ideal Perfect-SIC upper bound over a wide SIR / N range. This indicates that the proposed solution can more fully reconstruct and remove cross-link interference, making the echo signal after interference removal closer to the observation results under ideal interference-free conditions.

[0130] The various methods of the embodiments of this application have been described above. Apparatus for implementing the above methods will now be provided.

[0131] like Figure 10 As shown, this application embodiment also provides an interference cancellation device 1000 for a sensor-integrated network, applied to a first network device, comprising:

[0132] The receiving module 1001 is used to receive a first signal; the first signal includes: an echo signal of a sensing signal sent by the first network device and an interference signal generated by at least one second network device;

[0133] Initialization module 1002 is used to determine an initial estimated value of a target variable based on the first signal; the target variable includes: sensing channel, sensing data, interference channel, and interference data;

[0134] The processing module 1003 is used to take the first signal and the initial estimate of the target variable as input to a model-driven deep unfolded neural network to obtain the target variable after iterative optimization.

[0135] The interference elimination module 1004 is used to perform interference elimination on the first signal based on the iteratively optimized target variable to obtain the target perception signal.

[0136] The interference cancellation device for integrated sensing networks according to embodiments of this application initializes a first signal, including a sensing signal and an interference signal, to obtain an initial estimate of a target variable. Then, using a model-driven deep unfolded neural network, the sensing signal and the interference signal are stripped and reconstructed based on the initial estimate of the target variable to obtain an iteratively optimized target variable. Finally, based on the iteratively optimized target variable, the interference signal in the first signal is eliminated, and the target sensing signal is ultimately determined. The interference cancellation method for integrated sensing networks according to embodiments of this application solves the problems of low efficiency and low accuracy in existing methods for eliminating cross-link interference.

[0137] Another embodiment of the network device in this application, such as Figure 11 As shown, it includes a transceiver 1110, a processor 1100, a memory 1120, and a program or instructions stored in the memory 1120 and executable on the processor 1100; when the processor 1100 executes the program or instructions, it implements the various processes of the above-described interference cancellation method embodiment for integrated sensing networks and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0138] The transceiver 1110 is used to receive and send data under the control of the processor 1100.

[0139] Among them, Figure 11 In this context, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 1100) and memory (memory 1120). The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. The transceiver 1110 may be multiple elements, including transmitters and receivers, providing a unit for communicating with various other devices over a transmission medium. The processor 1100 is responsible for managing the bus architecture and general processing, and the memory 1120 may store data used by the processor 1100 during operation.

[0140] It should be noted that the implementation methods in the above embodiments are all applicable to the embodiments of the network device and can achieve the same technical effect. The device provided in this application can implement all the method steps implemented in the above method embodiments and can achieve the same technical effect. Therefore, the parts that are the same as those in the method embodiments and the beneficial effects will not be described in detail here.

[0141] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the above-described interference cancellation method embodiments for integrated sensing networks, achieving the same technical effects. To avoid repetition, these will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0142] This application also provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, they implement the various processes of the above-described interference cancellation method embodiments for integrated sensing networks and achieve the same technical effects. To avoid repetition, they will not be described again here.

[0143] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0144] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0145] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. An interference cancellation method for integrated sensing and communication networks, applied to a first network device, characterized in that, include: Receive the first signal; The first signal includes: the echo signal of the sensing signal sent by the first network device and the interference signal of at least one second network device to the first network device; Based on the first signal, an initial estimate of the target variable is determined; the target variable includes: the sensing channel, the sensing data, the interference channel, and the interference data. The first signal and the initial estimate of the target variable are used as inputs to a model-driven deep unfolded neural network to obtain the target variable after iterative optimization. Based on the target variable after iterative optimization, interference is eliminated on the first signal to obtain the target perception signal.

2. The method according to claim 1, characterized in that, Based on the first signal, determine the initial estimate of the target variable, including: Based on the first signal, determine the pilot sequence of the first signal; The initial estimate of the target variable is obtained by performing variable initialization processing based on the pilot sequence.

3. The method according to claim 1 or 2, characterized in that, The interference signal includes: the interference signal transmitted by the communication signal sent by the second network device through the interference channel.

4. The method according to claim 1, characterized in that, Using the first signal and the initial estimate of the target variable as input to a model-driven deep unfolded neural network, the iteratively optimized target variable is obtained, including: Based on the initial estimates of the target variables, the initial estimates of the sensing channel, the initial estimates of the sensing data, the initial estimates of the interference channel, and the initial estimates of the interference data are obtained. The estimation value of the interference channel is updated and optimized based on the initial estimation value of the sensing channel, and the estimation value of the sensing channel is updated and optimized based on the updated estimation value of the interference data, until the target estimation values ​​of the sensing channel and the interference data are determined. The estimation value of the interference data is updated and optimized based on the initial estimation value of the sensing data, and the estimation value of the sensing data is updated and optimized based on the updated estimation value of the interference channel, until the target estimation values ​​of the sensing data and the interference channel are determined. The target estimate of the interference signal is determined based on the target estimate of the interference data and the target estimate of the interference channel.

5. The method according to claim 1, characterized in that, The model-driven deep unfolded neural network includes: The sensing channel update module is used to update the estimated value of the sensing channel based on the echo signal and the estimated value of the sensing data, so as to obtain the reconstructed estimated value of the sensing channel. The interference data estimation module is used to update the estimated value of the interference data based on the reconstructed estimated value of the sensing channel and the estimated value of the interference channel, so as to obtain the reconstructed estimated value of the interference data. The sensing data update module is used to transform the reconstructed estimate of the sensing channel to obtain the updated estimate of the sensing channel. An interference channel estimation module is used to determine an updated estimate of the interference channel based on an initial estimate of the interference data and a reconstructed estimate of the interference channel. The sensing channel update module, the interference data estimation module, the sensing data update module, and the interference channel estimation module are sequentially passed and interactively fed back to each other, and the estimated value of the target variable is iteratively updated alternately.

6. The method according to claim 5, characterized in that, The method further includes: A sparse denoising operator is used to denoise the reconstructed estimate of the sensing channel to obtain the denoised estimate of the sensing channel. The reconstructed estimate of the interference data is continuously mapped to the constellation points to obtain a smoothed estimate of the interference data. The updated estimated value of the sensing channel is fused with the reconstructed estimated value of the sensing channel to obtain the fused estimated value of the sensing channel. The updated estimate of the interference channel is fused with the reconstructed estimate of the interference channel to obtain the fused estimate of the interference channel.

7. An interference cancellation device for a sensor-integrated network, applied to a first network device, characterized in that, include: The receiving module is used to receive the first signal; The first signal includes: the echo signal of the sensing signal sent by the first network device and the interference signal generated by at least one second network device; An initialization module is used to determine an initial estimate of a target variable based on the first signal; the target variable includes: a sensing channel, sensing data, an interference channel, and interference data; The processing module is used to take the first signal and the initial estimate of the target variable as input to a model-driven deep unfolded neural network to obtain the target variable after iterative optimization. The interference elimination module is used to eliminate interference in the first signal based on the iteratively optimized target variable to obtain the target perception signal.

8. A network device, characterized in that, include: Transceiver, processor, memory, and programs or instructions stored in the memory and executable on the processor; When the processor executes the program or instructions, it implements the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes computer instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 6.

Images