Communication sensing integrated channel estimation reconstruction method based on compressed sensing

Abstract

The invention discloses a compressed sensing-based communication sensing integrated channel estimation reconstruction method, which comprises the following steps of: cleaning, de-noising and normalizing original data from different modes to ensure the quality and consistency of the data; extracting features useful for channel estimation from data of different modes by using a feature extraction algorithm, and fusing the features from the different modes by using a fusion algorithm to form comprehensive description of a channel; and channel estimation is realized by using a compressed sensing theory. According to the method, the complementarity and redundancy of the multi-modal data are utilized, an efficient compressed sampling strategy is designed, and comprehensive description of channel characteristics is achieved; meanwhile, an adaptive channel reconstruction algorithm is provided, reconstruction parameters and strategies are dynamically adjusted according to channel characteristics and data quality, and the reconstruction performance and accuracy of the algorithm in complex and changeable scenes are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of signal processing, and in particular relates to a communication-aware integrated channel estimation and reconstruction method based on compressed sensing. Background Art

[0002] The development of the next-generation 6G integrated communication and perception technology urgently requires high-end, comprehensive testing technologies. Integrated Communication and Perception (ISAC) is a new derivative capability and core technology of 6G technology. Channel characteristic measurement and simulation research are fundamental to the evaluation, deployment, and optimization of ISAC systems. Because ISAC systems possess both communication and perception capabilities, their channel characteristic measurement requires integrating existing mobile communication channel models with radar echo detection models. Therefore, channel characteristic measurement and simulation in integrated communication and perception scenarios also face new challenges. On the one hand, traditional communication and perception channel modeling are independent research directions, each focusing on its own application scenarios. On the other hand, the performance evaluation indicators for communication and perception systems are also different. Therefore, a comprehensive channel estimation and detection method is needed for integrated communication and perception systems.

[0003] Currently, the mainstream channel estimation method is mainly based on pilot estimation. In essence, it uses the pilot matrix in the transmitted signal to map the channel to the pilot matrix of the received signal. The difference between the transmitted and received pilots is then analyzed and calculated to obtain the channel information. However, the proportion of pilot signals in the transmitted signal reduces the utilization of spectrum resources, and the sampling accuracy is easily low due to the influence of multipath effects and noise. In addition, traditional channel estimation methods follow the Nyquist sampling theorem and require the collection of a large amount of data, resulting in low sampling efficiency. When faced with complex and changing synaesthesia fusion scenarios, a large amount of computing resources are required to process multimodal data in complex environments, resulting in low system operation efficiency. Therefore, it is necessary to conduct research on composite channel estimation and detection technologies for synaesthesia fusion scenarios.

[0004] Compressed sensing (CS) is a novel sampling technique that breaks the limitations of the Nyquist sampling theorem. It maps signals into a low-dimensional space and then recovers them through reconstruction algorithms. This property has been widely applied in fields such as image compression, information acquisition, and channel estimation. Research has shown that channels are typically sparse, which satisfies the sparse signal requirement of CS. CS's synchronous sampling and compression of signals is highly compatible with pilot-based estimation. CS can not only reduce the pilot ratio but also achieve higher-accuracy channel estimation while maintaining the same pilot ratio. Existing CS-based channel estimation and reconstruction methods rely solely on single-modal data, ignoring the complementarity and redundancy between multimodal data. This method fails to fully characterize channel characteristics. Furthermore, their reconstruction algorithms often use fixed parameter strategies and cannot adapt to changes in channel characteristics and data quality. Consequently, reconstruction performance is unstable and inaccurate in complex and changing synaesthesia fusion scenarios, making them unsuitable for the dynamic channel environments of synaesthesia integration. Summary of the Invention

[0005] In response to the above-mentioned problems existing in the prior art, the present invention proposes a communication perception integrated channel estimation and reconstruction method based on compressed sensing, which can effectively improve the accuracy and performance of channel parameter estimation in complex environments and meet the real-time channel estimation needs of the communication perception system.

[0006] The communication-aware integrated channel estimation and reconstruction method based on compressed sensing includes the following steps:

[0007] Step 1: Clean, denoise, and normalize the raw data from different modalities to ensure data quality and consistency;

[0008] Step 2: Use feature extraction algorithms to extract features useful for channel estimation from data of different modalities, and use fusion algorithms to fuse features from different modalities to form a comprehensive description of the channel;

[0009] Step 3: Use compressed sensing theory to achieve channel estimation.

[0010] Furthermore, step 1 includes the following sub-steps:

[0011] Step 1.1: Receive multi-modal raw data streams from radar signals and communication signals, including I / Q signals, RSSI strength, and delay measurement data;

[0012] Step 1.2: De-noise the radar signal and eliminate clutter interference using wavelet thresholding, and use an adaptive filter to suppress inter-symbol interference caused by the multipath effect of the communication signal.

[0013] Step 1.3: Align the sampling rates and dimensions of the different modal data, and resample the radar data according to the communication signal time slot.

[0014] Furthermore, the step 2 is specifically as follows:

[0015] First, single-mode feature extraction is performed on the communication signal and radar signal respectively. For radar data, the target distance and speed features are extracted; for communication data, the signal amplitude and phase features are extracted.

[0016] Secondly, the multimodal features are mapped to a unified sparse space through orthogonal basis transformation to construct a joint sparse vector.

[0017] Furthermore, step 3 includes the following sub-steps:

[0018] Step 3.1: The mathematical model of compressed sensing is expressed as:

[0019] y=Φx;(1)

[0020] Where y is the observation vector, i.e., the measurement value directly obtained by the sensor, Φ is the observation matrix, which satisfies the constraint equidistance condition, and x is the original signal, which is sparse in a specific transformation domain;

[0021] Step 3.2: Based on the sparsity of the channel, construct a suitable sparse representation model, which is expressed as:

[0022] x=ψs;(2)

[0023] Where s is a sparse vector with only K non-zero values, and ψ is an orthogonal basis matrix;

[0024] Substituting formula (2) into formula (1) yields:

[0025] y=Φx=Φψs=Θs;(3)

[0026] Where Θ = Φψ is called the sensing matrix, and the sensing matrix and the sparse basis matrix need to satisfy non-correlation;

[0027] Step 3.3: The observation matrix needs to satisfy the RIP condition, that is, for any K coefficient signal s, we have:

[0028]

[0029] Among them, δ K ∈(0,1) is an isometric constant;

[0030] Step 3.4: Convert signal reconstruction into an optimization problem for solving the sparse coefficient s:

[0031] min||s||1subjectto||y-Θs||2≤ε; (5)

[0032] Step 3.5: Adaptively adjust the reconstruction parameters and strategies based on the channel characteristics and data quality. When the channel changes rapidly, increase the number of reconstruction iterations; when the data quality is poor, adjust the regularization parameters to improve the stability of the reconstruction.

[0033] Beneficial technical effects brought about by the present invention:

[0034] Compared with the existing technology, the present invention uses a compressed sensing algorithm for data sampling, which does not require compliance with the Nyquist sampling theorem, reduces the amount of data collected, and improves sampling efficiency. By utilizing the complementarity and redundancy of multimodal data, data from different modalities are processed and features are fused, improving data utilization and enabling a more comprehensive and accurate description of channel characteristics. Through an adaptive channel reconstruction algorithm, the present invention can dynamically adjust reconstruction parameters and strategies based on channel characteristics and data quality, maintaining stable reconstruction performance and accuracy in complex and changing synaesthesia fusion scenarios, and has greater adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a schematic diagram of the communication-sensing integrated channel estimation and reconstruction principle based on compressed sensing in the present invention;

[0036] Figure 2 This is a flow chart of the communication-sensing integrated channel estimation and reconstruction method based on compressed sensing in the present invention; DETAILED DESCRIPTION

[0037] The specific implementation of the present invention will be further described below with reference to specific embodiments:

[0038] The communication perception integrated channel estimation and reconstruction method based on compressed sensing mainly includes two parts, such as Figure 1 As shown in the figure, there is a multimodal data-driven channel compression sampling module and an adaptive channel reconstruction module. Among them, the multimodal data-driven channel compression sampling module is responsible for processing and extracting features from the original data from different modalities, and fusing the extracted features; the adaptive channel reconstruction module uses the sparse representation and reconstruction technology of compressed sensing based on the data obtained by compression sampling to reconstruct the channel characteristics. The multimodal data-driven channel compression sampling module first processes and extracts features from the original data, and passes the extracted features as input to the adaptive channel reconstruction module. The adaptive channel reconstruction module uses a sparse representation model and an adaptive adjustment strategy based on the input features to reconstruct the channel characteristics.

[0039] The present invention is mainly divided into three steps, namely multimodal data cleaning, feature extraction and fusion, and compressed sensing channel estimation. Figure 2 As shown, specifically:

[0040] Step 1: In the communication and perception fusion scenario, channel estimation typically requires the collection of a large amount of sensory data, including vision, sound, temperature, etc., and the transmission of a large amount of communication data. Therefore, to ensure the accuracy of channel estimation, the raw data from different modalities needs to be cleaned, denoised, and normalized to ensure data quality and consistency.

[0041] Step 1 includes the following sub-steps:

[0042] Step 1.1: Receive multi-modal raw data streams from radar signals and communication signals, including I / Q signals, RSSI strength, and delay measurement data;

[0043] Step 1.2: De-noise the radar signal and eliminate clutter interference using wavelet thresholding, and use an adaptive filter to suppress inter-symbol interference caused by the multipath effect of the communication signal.

[0044] Step 1.3: Align the sampling rates and dimensions of the different modal data, and resample the radar data according to the communication signal time slot.

[0045] Step 2: Use feature extraction algorithms to extract features useful for channel estimation from data of different modalities. Use fusion algorithms to fuse features from different modalities to form a comprehensive description of the channel. The fused features can more accurately reflect the characteristics of the channel. Specifically:

[0046] First, single-mode feature extraction is performed on the communication signal and radar signal respectively. For radar data, the target distance and speed features are extracted; for communication data, the signal amplitude and phase features are extracted.

[0047] Secondly, the multimodal features are mapped to a unified sparse space through orthogonal basis transformation to construct a joint sparse vector.

[0048] Step 3: According to compressed sensing theory, if a signal is sparse in a certain transform domain, it can be reconstructed and recovered using a sampling number far fewer than the number required by the Nyquist sampling theorem. In wireless communication systems, multipath effects cause the channel to exhibit sparseness. Therefore, compressed sensing theory can achieve high-performance channel estimation with a small pilot signal ratio and high spectrum utilization.

[0049] Step 3 includes the following sub-steps:

[0050] Step 3.1: The mathematical model of compressed sensing is expressed as:

[0051] y=Φx;(1)

[0052] Where y is the observation vector, i.e., the measurement value directly obtained by the sensor, Φ is the measurement matrix, which satisfies the constrained isometry property (RIP) condition, and x is the original signal, which is sparse in a specific transform domain.

[0053] Step 3.2: Based on the sparsity of the channel, construct a suitable sparse representation model. This model can represent the channel as a sparse vector, which can then be reconstructed using the theory of compressed sensing. Project the signal into the sparse domain so that most of its coefficients are close to zero. If the signal itself is not sparse, it needs to be projected into the sparse domain using the sparse basis matrix ψ, which is expressed as:

[0054] x=ψs;(2)

[0055] Where s is a sparse vector with only K non-zero values. Common sparse bases include Fourier basis, wavelet basis, DCT basis, etc., and ψ is an orthogonal basis matrix;

[0056] Substituting formula (2) into formula (1) yields:

[0057] y=Φx=Φψs=Θs;(3)

[0058] Where Θ = Φψ is called the sensing matrix, and the sensing matrix and the sparse basis matrix need to satisfy non-correlation;

[0059] Step 3.3: The observation matrix needs to satisfy the RIP condition, that is, for any K coefficient signal s, we have:

[0060]

[0061] Among them, δ K ∈(0,1) is an isometric constant;

[0062] Step 3.4: When restoring the compressed signal, since s is sparse and the sensing matrix satisfies the RIP condition, it provides a mathematical guarantee for accurately restoring the original signal using the sampled signal. Signal reconstruction is transformed into an optimization problem for solving the sparse coefficient s:

[0063] min||s||0subjectto||y-Θs||2≤ε; (5)

[0064] The L0 norm is directly non-convex and NP-hard, so it is relaxed using the L1 norm. L1 optimization can be efficiently solved using a convex optimization algorithm:

[0065] min||s||1subjectto||y-Θs||2≤ε; (6)

[0066] After solving for s, the final channel response in actual applications can be calculated according to formula (2).

[0067] Step 3.5: Adaptively adjust the reconstruction parameters and strategies based on the channel characteristics and data quality. When the channel changes rapidly, increase the number of reconstruction iterations; when the data quality is poor, adjust the regularization parameters to improve the stability of the reconstruction.

[0068] This paper exploits the complementarity and redundancy of multimodal data to design an efficient compressed sampling strategy. First, the raw data from different modalities is cleaned, denoised, and normalized to ensure data quality and consistency. Feature extraction algorithms are then used to extract features useful for channel estimation from the data from different modalities. Finally, a fusion algorithm is used to fuse the features from different modalities to form a comprehensive description of the channel characteristics.

[0069] An adaptive channel reconstruction method is proposed. Based on sparse representation and reconstruction techniques from compressed sensing, this method leverages the sparsity of the channel and constructs a suitable sparse representation model to efficiently describe and reconstruct channel characteristics. It can adaptively adjust reconstruction parameters and strategies based on channel characteristics and data quality. This adaptive mechanism enables the algorithm to maintain stable reconstruction performance and accuracy in complex and changing synaesthesia fusion scenarios.

[0070] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the scope of protection of the present invention.

Claims

1. A communication-aware integrated channel estimation and reconstruction method based on compressed sensing, characterized in that: The following steps are involved: Step 1: Clean, denoise, and normalize the raw data from different modalities to ensure data quality and consistency; Step 2: Use feature extraction algorithms to extract features useful for channel estimation from data of different modalities, and use fusion algorithms to fuse features from different modalities to form a comprehensive description of the channel; Step 3: Use compressed sensing theory to achieve channel estimation.

2. The communication-aware integrated channel estimation and reconstruction method based on compressed sensing according to claim 1, characterized in that: The step 1 includes the following sub-steps: Step 1.1: Receive multi-modal raw data streams from radar signals and communication signals, including I / Q signals, RSSI strength, and delay measurement data; Step 1.2: De-noise the radar signal and eliminate clutter interference using wavelet thresholding, and use an adaptive filter to suppress inter-symbol interference caused by the multipath effect of the communication signal; Step 1.3: Align the sampling rates and dimensions of data from different modalities, and resample the radar data according to the communication signal time slot.

3. The communication-aware integrated channel estimation and reconstruction method based on compressed sensing according to claim 2, characterized in that: The step 2 is specifically as follows: First, single-mode feature extraction is performed on the communication signal and radar signal respectively. For radar data, the target distance and speed features are extracted; for communication data, the signal amplitude and phase features are extracted. Secondly, the multimodal features are mapped to a unified sparse space through orthogonal basis transformation to construct a joint sparse vector.

4. The communication-aware integrated channel estimation and reconstruction method based on compressed sensing according to claim 3, characterized in that: Step 3 includes the following sub-steps: Step 3.1: The mathematical model of compressed sensing is expressed as: y=Φx;(1) Where y is the observation vector, i.e., the measurement value directly obtained by the sensor, Φ is the observation matrix, which satisfies the constraint equidistance condition, and x is the original signal, which is sparse in a specific transformation domain; Step 3.2: Based on the sparsity of the channel, construct a suitable sparse representation model, which is expressed as: x=ψs;(2) Where s is a sparse vector with only K non-zero values, and ψ is an orthogonal basis matrix; Substituting formula (2) into formula (1) yields: y=Φx=Φψs=Θs;(3) Where Θ = Φψ is called the sensing matrix, and the sensing matrix and the sparse basis matrix need to satisfy non-correlation; Step 3.3: The observation matrix needs to satisfy the RIP condition, that is, for any K coefficient signal s, we have: Among them, δ K ∈(0,1) is an isometric constant; Step 3.4: Convert signal reconstruction into an optimization problem for solving the sparse coefficient s: min||s||1subjectto||y-Θs||2≤ε; (5) Step 3.5: Adaptively adjust the reconstruction parameters and strategies based on the channel characteristics and data quality. When the channel changes rapidly, increase the number of reconstruction iterations; when the data quality is poor, adjust the regularization parameters to improve the stability of the reconstruction.

Images