Cognitive radio spectrum sensing method based on VG transformation and graph quadratic form, medium and equipment

Abstract

The invention provides a cognitive radio spectrum sensing method based on VG transformation and graph quadratic form, a medium and equipment. The method comprises the following steps: firstly, calculating an autocorrelation function of an observation signal and segmenting the observation signal in equal length; then, VG transformation is carried out on each section of the autocorrelation function, so that a corresponding adjacency matrix A and a degree matrix D are obtained, and a standardized adjacency matrix is calculated according to the adjacency matrix A and the degree matrix D; and then, the quadratic forms of each standardized adjacent matrix An are calculated and summed, the sum Xs of the quadratic forms is defined as a detection statistic, a judgment threshold is set, and the detection statistic is compared with the judgment threshold, so that whether signals exist in the observation signals or not is judged. Compared with a method for directly carrying out VG transformation on the time domain signal, the method has the advantage that the detection performance is better improved. Meanwhile, besides the topological characteristics, the statistical characteristics of the observation signals are fully utilized and fused with the topological characteristics, and the utilization rate of the observation signals is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of signal processing, and in particular relates to a cognitive radio spectrum sensing method, medium and device based on VG transformation and graph quadratic form. Background Art

[0002] With the large-scale commercial use of 5G and the evolution of the next generation of mobile communication technology (6G), the number of wireless devices and their application services has increased rapidly, and the problem of spectrum resource depletion has become increasingly serious. Cognitive Radio (CR) fully utilizes the idle spectrum of primary users by introducing opportunistic spectrum access mechanism and spectrum sensing technology, providing an effective way to improve the utilization rate of spectrum resources. Sensing technology is the cornerstone of spectrum sensing. Its high-precision and high-sensitivity sensors can accurately capture weak signals in the wireless environment and provide effective data input for spectrum sensing. As the primary link in realizing cognitive radio, spectrum sensing is tasked with defining effective detection statistics with the help of observed signals and making statistical judgments on the existence of primary user signals based on hypothesis testing models. Obviously, the study of spectrum sensing algorithms is a key issue in realizing the effective, safe and reliable operation of cognitive radio systems.

[0003] Existing spectrum sensing algorithms can be divided into two categories based on the way of obtaining detection statistics: statistical signal processing and topological signal processing. The former usually represents the signal collected by the sensor in the form of Euclidean space such as time domain and frequency domain and extracts its statistical features. The main representative algorithms are: energy detection, matched filter, cyclostationary feature detection, machine learning detection algorithm, etc. The latter completes the spectrum sensing task by mapping the signal into non-Euclidean space forms such as graphs and simplicial complexes, and extracting its topological features as detection statistics. This type of algorithm can effectively mine the potential structural characteristics of the signal without increasing the number of samples, provide additional information for the original statistical characteristics, and provide important clues for defining new detection statistics. It has gradually become a hot topic in the field of cognitive radio spectrum sensing research. Existing topological signal processing spectrum sensing algorithms mainly rely on graph domain transformation, such as quantization graph transformation, visibility graph transformation method, etc. The problems existing in various algorithms are as follows:

[0004] (1) In algorithms based on statistical signal processing, the separability and stability of feature quantities often rely on large samples or high-quality data sets. Under conditions such as insufficient sample numbers or missing training data sets, their performance is difficult to guarantee. Therefore, how to improve algorithm performance and control complexity under data-constrained conditions is a key point that needs urgent research.

[0005] (2) Various noise or background interference. As the electromagnetic environment becomes more and more complex, signals are easily interfered by various noises.

[0006] (3) Poor performance at low signal-to-noise ratios. The signal-to-noise ratio range applicable to most current algorithms is not wide enough, and the ROC curve at low signal-to-noise ratios is poor.

[0007] (4) Excessive computational complexity. This problem mainly occurs in spectrum sensing algorithms based on visibility graphs. In the process of transforming a random sequence into a visibility graph, each sample needs to be transformed into a visibility graph, and its complexity is the square of the number of sample points. Therefore, the more samples involved in the visibility graph transformation, the higher the computational complexity. Summary of the invention

[0008] In view of the deficiencies in the prior art, the present invention provides a cognitive radio spectrum sensing method, medium and device based on VG transformation and graph quadratic form.

[0009] To achieve the above object, the present invention adopts the following technical solutions:

[0010] In a first aspect, the present invention provides a cognitive radio spectrum sensing method based on VG transformation and graph quadratic form, comprising:

[0011] Step 1: Obtain an observation signal, calculate the autocorrelation function of the observation signal, and take half of it by using its symmetry;

[0012] Step 2: Divide the halved autocorrelation function samples into equal length segments to generate a series of short signals; perform VG transformation on each short signal to obtain a series of adjacency matrices and degree matrices, and use them to calculate the corresponding standardized adjacency matrix;

[0013] Step 3: Use each short signal and the corresponding standardized adjacency matrix to calculate the corresponding quadratic form, sum a series of quadratic forms, and use the sum of the quadratic forms as the detection statistic to detect the observed signal.

[0014] Optionally, in step 1, the observation signal is binary phase shift keying superimposed with Gaussian white noise.

[0015] Optionally, in step 1, the autocorrelation function of the observed signal is:

[0016]

[0017] In the formula, r(n) represents the observed signal, C(m) is the autocorrelation function of r(n), m represents the independent variable of the autocorrelation function, and N is the number of observed signal samples;

[0018] Using the symmetry of the autocorrelation function, we can get half of it:

[0019]

[0020] Optionally, in step 2, the halved autocorrelation function samples are divided into equal-length segments to generate a series of short signals, specifically:

[0021] The autocorrelation function samples C(m) after halving are divided into segments every λ, and a total of L = [N / λ] segments, N is the number of samples, and each segment is regarded as a short signal to obtain a new sequence group Z i (k), 1≤i≤L, 1≤k≤λ, segmented as follows:

[0022]

[0023] Optionally, in step 2, the visibility rule of the VG transformation is:

[0024]

[0025] In the formula, A(x a ,y a ), B(x b ,y b ), C(x c ,y c ) are three arbitrary points in the sequence, and satisfy x a <x c <x b .

[0026] Optionally, in step 2, the standardized adjacency matrix A n Defined as:

[0027]

[0028] Where D is the degree matrix of the adjacency matrix A.

[0029] Optionally, in step 3, the corresponding quadratic form is calculated using each short signal and the corresponding standardized adjacency matrix, and the calculated quadratic form X is as follows:

[0030] X=C(m) T ·A n C(m);

[0031] Where C(m) is the autocorrelation function, A n is the normalized adjacency matrix.

[0032] Optionally, in step 3, a series of quadratic forms are summed, and the sum of the quadratic forms is used as a detection statistic to detect the observed signal, as follows:

[0033] Calculate each normalized adjacency matrix A n The quadratic form X and sum to get X s ;

[0034] Set the following threshold value η:

[0035]

[0036] Where P fa is the false alarm probability, σ0 is s The corresponding Gaussian distribution standard deviation, μ0 is X s Corresponding Gaussian distribution mean, complementary error function x and y represent random variables X s The sample value of erfc -1 is its inverse function;

[0037] Compare the sum of quadratic forms X s and threshold value η, if X s >η, it means that there is a primary user signal in the detection signal; if it is not satisfied, it means that there is no primary user signal in the detection signal.

[0038] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as described in the first aspect.

[0039] In a third aspect, the present invention provides 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 cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as described in the first aspect is implemented.

[0040] The beneficial effects of the present invention are as follows: the present invention studies the detection of cognitive radio signals under low signal-to-noise ratio. The present invention adopts a graph domain signal detection method in an irregular domain, transforms the observed signal into a visual graph, obtains its standardized adjacency matrix, and uses the sum of the graph quadratic form of the standardized adjacency matrix as a detection statistic to detect the signal and noise. The method has the following advantages:

[0041] (1) The sum of the quadratic form of the graph normalized adjacency matrix is ​​selected as the detection statistic, which fully utilizes and integrates the statistical and topological characteristics of the observed signal, improves the detection performance at low signal-to-noise ratio, and does not require eigendecomposition of the graph adjacency matrix or Laplacian matrix, which has low computational complexity.

[0042] (2) The piecewise autocorrelation function of the observed signal is used as the input of the visible graph transformation, which reduces the number of samples involved in the graph transformation and the computational complexity; at the same time, the offset coefficient of the sum of the quadratic form of the graph can be increased, thereby improving the detection efficiency of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a flow chart of a cognitive radio spectrum sensing method based on VG transformation and graph quadratic form.

[0044] Figure 2a It is the graph structure obtained by performing a series of VG transformation on the observed signal under the H0 hypothesis.

[0045] Figure 2b It is the graph structure obtained by performing a series of processing on the observed signal under the H1 hypothesis and then performing VG transformation.

[0046] Figure 3 It is a comparison chart of the mean of the sum of quadratic forms obtained after 1000 simulations under the H0 and H1 assumptions.

[0047] Figure 4a It is a detection probability curve diagram of the method of the present invention.

[0048] Figure 4b It is a false alarm probability curve diagram of the method of the present invention.

[0049] Figure 5 It is a comparative diagram of receiver operating characteristic curves (ROC) between the method of the present invention and other comparative methods. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0051] In one embodiment, if Figure 1 As shown, the present invention proposes a cognitive radio spectrum sensing method based on VG transformation and graph quadratic form, which specifically includes the following steps.

[0052] Step 1: Get the observed signal, calculate the autocorrelation function of the observed signal, and take half of it by using its symmetry.

[0053] In this embodiment, the observation signal used is Binary Phase Shift Keying (BPSK) superimposed with Gaussian white noise.

[0054] Assume that the observed signal r(t) contaminated by additive white Gaussian noise (AWGN) can be expressed as:

[0055] r(t)=s(t)+w(t);

[0056] Where s(t) is the signal component and w(t) is the noise component. After sampling, the observed signal is obtained in discrete form:

[0057] r(n)=s(n)+w(n), n=1, 2,...,N;

[0058] Where N is the number of observed signal samples. Therefore, the spectrum sensing problem can be reduced to the following hypothesis test:

[0059]

[0060] The formula shows that when H0, the spectrum is not occupied (that is, there is no signal in the channel); when H1, the spectrum is occupied (that is, there is a signal in the channel).

[0061] Assume that the autocorrelation function of the observed signal is C(m), then:

[0062]

[0063] Where r(n) represents the observed signal, C(m) is the autocorrelation function of r(n), and m represents the independent variable of the autocorrelation function.

[0064] Taking advantage of the symmetry of the autocorrelation function and halving it, we have:

[0065]

[0066] Step 2: Divide the desired autocorrelation function into segments of equal length to generate a series of short signals; perform VG transformation on each segment to obtain a series of adjacency matrices A and degree matrices D, and use them to calculate the standardized adjacency matrix A n .

[0067] In this embodiment, the specific steps of step 2 are as follows:

[0068] Step 2.1: Divide the desired autocorrelation function into segments of equal length to generate a series of short signals, namely:

[0069] Divide the autocorrelation function samples C(m) into segments every λ, and divide them into L = [N / λ] segments in total, and regard each segment as a short signal to obtain a new sequence group Z i (k), 1≤i≤L, 1≤k≤λ, that is:

[0070]

[0071] Step 2.2: Perform VG transformation on each signal segment to obtain a series of adjacency matrices A and degree matrices D. The visibility rule of VG is:

[0072]

[0073] In the formula, A(x a ,y a ), B(x b ,yb ), C(x c ,y c ) are three arbitrary points in the sequence, and satisfy x a <x c <x b . And use this to calculate the standardized adjacency matrix A n , A n Defined as:

[0074]

[0075] Where D is the degree matrix of the adjacency matrix A.

[0076] Step 3: Use each short signal and the corresponding standardized adjacency matrix A n Calculate the corresponding quadratic form; sum the obtained series of quadratic forms, use the sum of the quadratic forms X as the detection statistic, and detect the observed signal.

[0077] In this embodiment, the specific steps of step 3 are as follows:

[0078] Step 3.1: Normalize the adjacency matrix A according to the obtained visibility graph n . Calculate the corresponding quadratic form, which is:

[0079] X=C(m) T ·A n C(m);

[0080] Where C(m) is the input autocorrelation function.

[0081] Step 3.2: Define the sum of quadratic forms X s , calculate each standardized adjacency matrix A n The quadratic form X and sum to get X s , set the threshold value η, and determine whether the observed signal is H0 or H1 hypothesis:

[0082]

[0083]

[0084] Where P fa is the false alarm probability, σ0 is the standard deviation of Gaussian distribution, μ0 is the mean of Gaussian distribution, and complementary error function x and y represent random variables X s The sample value of erfc -1 is its inverse function.

[0085] Figure 2a and Figure 2b It is a graph structure obtained by performing a series of processing on a segment of the observed signal and then performing VG transformation. Figure 2a It is the structure of the visual graph of the observed signal under the H0 hypothesis. Figure 2b It is the structure of the observed signal VG under the H1 hypothesis. By observing the VG structures of H0 and H1, it can be found that the visibility graph of the H1 hypothesis is denser, has greater connectivity, and has fewer scattered points, while the visibility graph of the H0 hypothesis is relatively scattered and has lower connectivity.

[0086] Figure 3 The comparison is made between the average values ​​of the sum of quadratic forms obtained after 1000 simulations under the assumptions H0 and H1. It can be observed that the average value of the sum of quadratic forms under the assumption H0 is not affected by the signal-to-noise ratio and has been stable at 1.8×10 6 The mean of the sum of quadratic forms under the H1 hypothesis increases with the increase of the signal-to-noise ratio, and the difference with the mean of the sum of quadratic forms under the H0 hypothesis increases.

[0087] Figure 4a and Figure 4b It is a curve diagram of detection probability and false alarm probability of the method of the present invention. Figure 4a It is the detection probability curve. When SNR=-10dB, the detection probability reaches 95.4%. Figure 4b It is a false alarm probability curve. When the detection probability is calculated by the method of the present invention, the false alarm probability needs to be controlled below 0.01.

[0088] Figure 5 It is a comparative schematic diagram of the ROC curves of the method of the present invention and the energy detection method, the cyclostationary detection method, the second largest eigenvalue method of the quantization graph, the wavelet denoising autocorrelation method, the differential spectrum method, and the aggregate operator quantization graph method. In order to make full use of the data and avoid experimental contingency, the simulation conditions are: 1000 simulations are performed, and the signal-to-noise ratio SNR is set to -10dB. It can be seen that the ROC curve of the method of the present invention is better than other methods. Therefore, the method of the present invention can better detect BPSK modulated signals, and the accuracy of detection is greatly improved.

[0089] In another embodiment, the present invention proposes a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as described in the above embodiment.

[0090] In another embodiment, the present invention proposes 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 cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as described in the aforementioned embodiment is implemented.

[0091] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0092] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0093] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A cognitive radio spectrum sensing method based on VG transformation and graph quadratic form, characterized in that: include: Step 1: Obtain an observation signal, calculate the autocorrelation function of the observation signal, and take half of it by using its symmetry; Step 2: Divide the halved autocorrelation function samples into equal length segments to generate a series of short signals; perform VG transformation on each short signal to obtain a series of adjacency matrices and degree matrices, and use them to calculate the corresponding standardized adjacency matrix; Step 3: Use each short signal and the corresponding standardized adjacency matrix to calculate the corresponding quadratic form, sum a series of quadratic forms, and use the sum of the quadratic forms as the detection statistic to detect the observed signal.

2. The cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as claimed in claim 1, characterized in that: In the step 1, the observation signal is binary phase shift keying superimposed with Gaussian white noise.

3. The cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as claimed in claim 2, characterized in that: In step 1, the autocorrelation function of the observed signal is: In the formula, r(n) represents the observed signal, C(m) is the autocorrelation function of r(n), m represents the independent variable of the autocorrelation function, and N is the number of observed signal samples; Using the symmetry of the autocorrelation function, we can get half of it:

4. The cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as claimed in claim 1, characterized in that: In step 2, the halved autocorrelation function samples are divided into equal length segments to generate a series of short signals, specifically: The autocorrelation function samples C(m) after halving are divided into segments every λ, and a total of L = [N / λ] segments, N is the number of samples, and each segment is regarded as a short signal to obtain a new sequence group Z i (k), 1≤i≤L, 1≤k≤λ, segmented as follows:

5. The cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as claimed in claim 4, characterized in that: In step 2, the visibility rule of the VG transformation is: In the formula, A(x a ,y a ), B(x b ,y b ), C(x c ,y c ) are three arbitrary points in the sequence, and satisfy x a <x c <x b .

6. The cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as claimed in claim 4, characterized in that: In step 2, the standardized adjacency matrix A n Defined as: Where D is the degree matrix of the adjacency matrix A.

7. The cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as claimed in claim 1, characterized in that: In step 3, the corresponding quadratic form is calculated using each short signal and the corresponding standardized adjacency matrix, and the calculated quadratic form X is as follows: X=C(m) T ·A n ·C(m); Where C(m) is the autocorrelation function, A n is the normalized adjacency matrix.

8. The cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as claimed in claim 1, characterized in that: In step 3, a series of quadratic forms are summed, and the sum of the quadratic forms is used as a detection statistic to detect the observed signal, as follows: Calculate each normalized adjacency matrix A n The quadratic form X and sum to get X s ; Set the following threshold value η: Where P fa is the false alarm probability, σ0 is s The corresponding Gaussian distribution standard deviation, μ0 is X s Corresponding Gaussian distribution mean, complementary error function x and y represent random variables X s The sample value of erfc -1 is its inverse function; Compare the sum of quadratic forms X s and threshold value η, if X s >η, it means that there is a primary user signal in the detection signal; if it is not satisfied, it means that there is no primary user signal in the detection signal.

9. A computer-readable storage medium storing a computer program, characterized in that: The computer program enables the computer to execute the cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as described in any one of claims 1 to 8.

10. An electronic device, characterized in that: include: 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 cognitive radio spectrum sensing method based on VG transformation and graph quadratic form as described in any one of claims 1 to 8 is implemented.

Images