Broadband sensing method of MWC under-Nyquist sampling structure based on ridge regression

Abstract

The invention relates to a broadband sensing method of an MWC under-Nyquist sampling structure based on ridge regression. The method comprises the following steps: constructing a basic signal model by using an MWC sampling structure as a front end; constructing an under-sampling framework V by adopting a CTF module based on the basic signal model, and solving an MMV problem by utilizing the under-sampling framework V to obtain a support set of broadband signals; during solving, loop error detection is carried out, when the number of cycles reaches a preset target, the updating iteration process is stopped, and during loop error detection, ridge regression analysis is utilized to introduce ridge term deviation to obtain the error of each update; and outputting a spectrum sensing result according to the support set of the broadband signals. The sensing accuracy can be improved.

Description

Technical Field

[0001] The present invention relates to the field of broadband perception technology, and in particular to a broadband perception method based on a ridge regression-based MWC sub-Nyquist sampling structure. Background Art

[0002] With the rapid development of 5G and 6G technologies, as well as high-frequency and wideband technologies such as millimeter-wave radar, the demand for wideband sensing technologies is increasing. Traditional wideband sensing algorithms often directly utilize high-speed analog-to-digital converters (ADCs) for Nyquist sampling. However, due to factors such as high cost and power consumption, this performance is less than ideal. Therefore, wideband solutions based on sub-Nyquist (SNS) sampling have attracted attention. Multi-rate sampling and multi-set sampling, as early classic SNS structures, cannot effectively meet sensing requirements due to issues such as time delay and loss of key signals. MWC, as the first commercial RF front-end, not only overcomes these issues but also offers improved hardware scalability. Furthermore, MWC's continuous-to-finite CTF module can convert SNS signals into multiple measurement vectors (MMVs).

[0003] To recover the broadband signal support set from MMVs for PU perception, current recovery algorithms primarily focus on machine learning, MUSIC, and SOMP. Machine learning algorithms, such as the k-nearest neighbor algorithm (KNN), convolutional neural networks (CNN), and deep neural networks (DNN), offer excellent perception performance and robustness against noise, but they place certain demands on the node's computing power. The compressed sensing-based MUSIC algorithm separates signal and noise based on eigenvalues ​​when the signal-to-noise ratio is high. However, as the signal-to-noise ratio decreases, signal and noise become difficult to distinguish, leading to a decrease in performance. The SOMP algorithm, on the other hand, optimizes the MMV problem through iteration, using the least squares method to update the error and iteratively update the error to achieve perception.

[0004] While there have been a growing number of improvements to the SOMP algorithm in recent years, most of them involve introducing various restrictions or improved conditions before optimizing the objective function. However, the estimation method for the update process all uses the least squares method. However, in multivariate linear regression theory, the least squares method is considered an unbiased estimator, which can degrade when the independent variables have some correlation. In the MWC architecture, since each channel compresses the same broadband signal, this can lead to a certain degree of correlation in the autocorrelation function. Therefore, the perceptual accuracy of the SOMP algorithm cannot be guaranteed.

[0005] To further improve perception accuracy and ensure compatibility with other SOMP algorithms, we propose a SOMP algorithm based on ridge regression. During each error calculation, ridge regression estimates the solution to the MMV problem, correcting the update direction for more accurate recovery. This algorithm is compatible with other similar SOMP algorithms without significantly increasing complexity. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a wideband perception method based on ridge regression and MWC sub-Nyquist sampling structure, which can improve perception accuracy.

[0007] The technical solution adopted by the present invention to solve the technical problem is to provide a wideband perception method based on MWC sub-Nyquist sampling structure of ridge regression, including the following steps:

[0008] Use the MWC sampling structure as the front end to build the basic signal model;

[0009] Based on the basic signal model, the CTF module is used to construct an undersampling framework V. The undersampling framework V is used to solve the MMV problem and obtain the support set of the wideband signal. During the solution, cyclic error detection is performed. When the number of cycles reaches a preset target, the update iteration process is stopped. During the cyclic error detection, ridge regression analysis is used to introduce ridge term deviation to obtain the error of each update.

[0010] The spectrum sensing result is output based on the support set of the wideband signal.

[0011] The cyclic error detection uses ridge regression analysis to introduce ridge term deviation to obtain the error of each update, specifically including:

[0012] The estimated solution of the least squares method is calculated based on the compressed broadband perception signal and the compression matrix under the MWC structure;

[0013] Calculating an observation error based on an estimated solution of the least squares method;

[0014] Calculating a ridge term deviation based on the estimated solution of the least squares method and the observation error;

[0015] The error for each update is calculated based on the ridge term deviation.

[0016] The estimated solution of the least squares method is obtained by Calculated, where w LS is the estimated solution of the least squares method, A S is the compression matrix under the MWC structure, and y is the compressed broadband perception signal.

[0017] The observation error is expressed as Calculated, where is the observation error, m is the number of channels of the MWC sampling structure, L is the length of the signal after discrete fast Fourier transform, y is the compressed broadband perception signal, and w LS is the estimated solution of the least squares method, A S is the compression matrix under the MWC structure.

[0018] The ridge deviation is Calculated, where k is the ridge deviation, w LS is the estimated solution of the least squares method, is the observation error.

[0019] The error of each update is Calculated, where R i is the error of each update, V is the under-adopted framework constructed by the CTF module at each update, and A S is the compression matrix under the MWC structure, and k is the ridge deviation.

[0020] The basic signal model is expressed as: y(f)=A S z(f)+n, where y(f) is the narrowband signal after SNS, A S is the compression matrix under the MWC structure, z(f) is the original broadband signal, n is the noise signal, and f is the broadband frequency domain set.

[0021] The undersampling framework V is expressed as: Where V is the undersampling frame, y(f) is the narrowband signal after SNS, f is the wideband frequency domain set, y(n) represents the SNS narrowband signal after ADC sampling, and n represents the number of samples.

[0022] The technical solution adopted by the present invention to solve its technical problem is: providing an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein when the processor executes the computer program, the steps of the wide-band perception method based on the ridge regression-based MWC sub-Nyquist sampling structure are implemented.

[0023] The technical solution adopted by the present invention to solve its technical problem is: providing a computer-readable storage medium on which a computer program is stored, characterized in that when the computer program is executed by a processor, the steps of the wide-band perception method of the MWC sub-Nyquist sampling structure based on ridge regression are implemented.

[0024] Beneficial effects

[0025] Due to the adoption of the above-mentioned technical solution, the present invention has the following advantages and positive effects compared with the existing technology: in each error calculation, the present invention uses ridge regression to estimate the solution to the MMV problem, corrects the update direction and thus performs more accurate recovery, which not only improves the perception accuracy but also does not significantly increase the complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flow chart of the SOMP algorithm based on ridge regression according to the first embodiment of the present invention;

[0027] Figure 2 is a relationship diagram between SNR and detection probability of the SOMP algorithm based on ridge regression according to the first embodiment of the present invention;

[0028] Figure 3 4 is a relationship diagram between SNR and false alarm probability of the SOMP algorithm based on ridge regression according to the first embodiment of the present invention. DETAILED DESCRIPTION

[0029] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall equally within the scope limited by the appended claims of the application.

[0030] The first embodiment of the present invention relates to a wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure. The method is based on the SOMP algorithm and improves the least squares estimation solution by modifying the error update method, thereby obtaining better algorithm performance. Figure 1 As shown, the following steps are included:

[0031] Step 1: Use the MWC sampling structure as the front end of the sampling structure to build a basic signal model.

[0032] Consider a situation where K users are simultaneously present in a certain area. In the Hilbert space, there is a multi-band signal s, which is the superposition of K PU narrowband signals and can be expressed as:

[0033]

[0034] The signal at the receiving end contains not only the sum of the signals of K PUs, but also the noise signal. Therefore, the signal at the receiving end can be expressed as: y(f) = A S z(f)+n.

[0035] Where y(f)∈C m×P is the narrowband signal after SNS, A Sis the compression matrix under the MWC structure, z(f) is the original broadband signal, n is the noise signal, f∈F S is a wideband frequency domain set.

[0036] Step 2: Use the CTF module for recovery. The constructed framework V can be expressed as:

[0037] Wherein, y(n) represents the SNS narrowband signal after ADC sampling, and n represents the number of samples.

[0038] Therefore, the matrix V is obtained by Q = VV H To obtain the support set S = supp(z(F S )), using the constructed framework V, solve the constructed MMV problem: in, is the compressed matrix A S The pseudo-inverse matrix of .

[0039] Start the error detection loop and stop the update iteration process when the number of loops reaches a certain level.

[0040] In the loop process, since the ridge term deviation in ridge regression is an approximate solution based on the least squares method, the estimated solution of the least squares method is first calculated as follows:

[0041]

[0042] Among them, w LS is the estimated solution of the least squares method, and y is the compressed broadband perception signal.

[0043] Calculate the observation error of the estimated solution of the least squares method to prepare for the subsequent deviation calculation. The calculation method is:

[0044]

[0045] in, is the observation error, m is the number of channels of the MWC sampling structure, and L is the length of the signal after discrete fast Fourier transform.

[0046] The ridge term deviation is calculated using the observation error and the estimated solution of the least squares method. The calculation method is:

[0047]

[0048] Among them, k is the ridge deviation, and min{} is the minimum value.

[0049] Ridge regression analysis is used to introduce ridge term deviation, so the error update scheme in this embodiment is:

[0050]

[0051] Among them, R i is the error of each update.

[0052] Estimate the negative frequency support set S1=min(c i ,K), the support set of the positive frequency part of the unknown wideband signal is calculated as S2 = L+1-S1, and the estimated complete support set is S = S1∪S2, where min() is the minimum value and ∪ represents the union. This shows that this embodiment only calculates (L+1) / 2 vectors Z and does not need to calculate the other N support sets.

[0053] Step 3: Output spectrum sensing results based on the support set of the wideband signal.

[0054] In order to demonstrate the performance of this embodiment, basic numerical simulation results are given. The main reference indicators here are detection probability and false alarm probability. In the simulation experiment, it is assumed that there are three PUs (prior users), that is, K = 3. The sub-band bandwidth of the PUs is assumed to be B k ={50,50,50}MHz, energy is set to E k = {1, 2, 3}. The carrier frequency band is randomly generated within the range [0, 5] GHz. The number of sampling points per cycle of the front-end MWC is P = L = 195. The number of random +-1s in the mixing signal per cycle is also M0 = 195, and the number of channels is m = 20. The added Gaussian noise is independent and identically distributed. For each setting, the experimental results are the average simulation results of 500 experiments.

[0055] Figure 2 The results and details of the detection probability of different recovery algorithms at different SNRs are shown. The classic SOMP, StOMP, StDOMP, SOMPS, ORMP algorithms and their improved ridge regression algorithm (RP-SOMP) are simulated. At the same time, the CS-MUSIC algorithm is also compared for a more comprehensive evaluation of the algorithm performance. Figure 2 As shown in the figure, the large figure on the left shows the overall experimental results from SNR = -25dB to 25dB. The bar chart on the right shows the details from SNR = -15dB to 10dB. From left to right, they correspond to the algorithms in the legend. The same color represents the same type of algorithm. The left one is the improved algorithm after ridge regression, and the right one is the original algorithm. Figure 2 The detailed histogram shows that for the five different OMP algorithms, the proposed ridge regression SOMP improved method has improved performance from low to high signal-to-noise ratios. Figure 3The false alarm probability results and details are given in [1]. Similar to the detection probability, it can be seen that the false alarm probability gradually decreases with increasing signal-to-noise ratio. Furthermore, the proposed ridge regression-based improved SOMP algorithm improves the false alarm probability at different signal-to-noise ratios compared to other algorithms, demonstrating the superiority of the ridge regression-based improved SOMP algorithm of this embodiment.

[0056] It is not difficult to find that the improved SOMP algorithm based on ridge regression in this embodiment not only ensures the advantages of blind detection and has good perception performance compared with other mainstream algorithms for wide-band perception, but also has better compatibility with other improved SOMP algorithms.

[0057] A second embodiment of the present invention relates to an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the wideband perception method of the MWC sub-Nyquist sampling structure based on ridge regression of the first embodiment are implemented.

[0058] A third embodiment of the present invention relates to a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the wideband perception method of the ridge regression-based MWC sub-Nyquist sampling structure of the first embodiment are implemented.

[0059] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage, etc.) that contain computer-usable program code.

[0060] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0061] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction method, which is implemented in the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0062] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0063] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure, characterized in that: The following steps are involved: Use the MWC sampling structure as the front end to build the basic signal model; Based on the basic signal model, the CTF module is used to construct an undersampling framework V. The undersampling framework V is used to solve the MMV problem and obtain the support set of the wideband signal. During the solution, cyclic error detection is performed. When the number of cycles reaches a preset target, the update iteration process is stopped. During the cyclic error detection, ridge regression analysis is used to introduce ridge term deviation to obtain the error of each update. The spectrum sensing result is output based on the support set of the wideband signal.

2. The wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure according to claim 1 is characterized in that: The cyclic error detection uses ridge regression analysis to introduce ridge term deviation to obtain the error of each update. Specifically include: The estimated solution of the least squares method is calculated based on the compressed broadband perception signal and the compression matrix under the MWC structure; Calculating an observation error based on an estimated solution of the least squares method; Calculating a ridge term deviation based on the estimated solution of the least squares method and the observation error; The error for each update is calculated based on the ridge term deviation.

3. The wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure according to claim 2, characterized in that: The estimated solution of the least squares method is obtained by Calculated, where w LS is the estimated solution of the least squares method, A S is the compression matrix under the MWC structure, and y is the compressed broadband perception signal.

4. The wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure according to claim 2, characterized in that: The observation error is expressed as Calculated, where is the observation error, m is the number of channels of the MWC sampling structure, L is the length of the signal after discrete fast Fourier transform, y is the compressed broadband perception signal, and w LS is the estimated solution of the least squares method, A S is the compression matrix under the MWC structure.

5. The wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure according to claim 2, characterized in that: The ridge deviation is Calculated, where k is the ridge deviation, w LS is the estimated solution of the least squares method, is the observation error.

6. The wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure according to claim 2, characterized in that: The error of each update is Calculated, where R i is the error of each update, V is the under-adopted framework constructed by the CTF module at each update, and A S is the compression matrix under the MWC structure, and k is the ridge deviation.

7. The wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure according to claim 1, characterized in that: The basic signal model is expressed as: y(f)=A S z(f)+n, where y(f) is the narrowband signal after SNS, A S is the compression matrix under the MWC structure, z(f) is the original broadband signal, n is the noise signal, and f is the broadband frequency domain set.

8. The wideband sensing method based on ridge regression and MWC sub-Nyquist sampling structure according to claim 1, characterized in that: The undersampling framework V is expressed as: Where V is the undersampling frame, y(f) is the narrowband signal after SNS, f is the wideband frequency domain set, y(n) represents the SNS narrowband signal after ADC sampling, and n represents the number of samples.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the wideband sensing method based on the MWC sub-Nyquist sampling structure based on ridge regression are implemented as claimed in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the wideband sensing method of the MWC sub-Nyquist sampling structure based on ridge regression are implemented as claimed in any one of claims 1 to 8.

Images