A Spectrum Sensing Method Based on Array Spatial Spectrum

Abstract

The present invention discloses a spectrum sensing method based on array spatial spectrum. The spatial spectrum of the signal is obtained through the Conventional Beamforming (CBF) algorithm, and a test statistic is constructed by the ratio of the maximum value to the minimum value of the spatial spectrum and compared with a threshold to obtain a judgment result. If the primary user signal exists, the arrival angle corresponding to the maximum value of the spatial spectrum is output. In addition, since spectrum sensing is continuously performed, the present invention proposes to retain the spatial spectra previously detected as signal non-existence and superimpose them on the current detection, improving the detection performance and angle measurement accuracy under low signal-to-noise ratio.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and more particularly, to a spectrum sensing method based on array spatial spectrum. Background Art

[0002] In recent years, with the rapid development of wireless communication services, the available spectrum resources have become increasingly scarce. Research results show that in the traditional fixed spectrum resource allocation mode, the spectrum utilization rate is not high. To solve the contradiction between scarce spectrum resources and low utilization rate, some scholars have proposed the concept of cognitive radio. In cognitive radio, primary users are defined as users with legitimate spectrum usage rights, while secondary users can access the spectrum when primary users are idle, thus improving the spectrum utilization rate. As one of the key technologies of cognitive radio, spectrum sensing can help secondary users detect available spectrum. The main task of spectrum sensing is to detect spectrum holes, which is generally performed in the time domain, frequency domain, code domain or spatial domain.

[0003] Relatively classic spectrum sensing algorithms include the Energy Detection (ED) method, the Maximum-Minimum Eigenvalue (MME) detection method based on the covariance matrix, and the Cyclostationary detection (CSD) method. The ED algorithm compares the energy value of the sampled signal with a threshold according to the different energy levels of the received signal in the presence of a primary user signal and only noise, and obtains the detection result. This method does not require prior information about the signal and has a low computational complexity, but is affected by noise uncertainty; since the covariance matrix can distinguish the difference between the signal and the noise, the MME algorithm constructs a test statistic using the ratio of the maximum and minimum eigenvalues of the received signal covariance matrix, and the setting of the threshold is independent of the noise power, so the detection performance can still be guaranteed in an environment with uncertain noise; the CSD algorithm uses the cyclostationary characteristics of the signal for detection and has a high computational complexity.

[0004] The prior art discloses a multi - band cooperative spectrum sensing method based on clustering optimization, which establishes a cooperative sensing model composed of a spectrum sensing fusion center, N secondary users and licensed users; the spectrum sensing fusion center clusters the secondary users according to their own signal - to - noise ratios and M preset clustering signal - to - noise ratio thresholds, and within each cluster containing secondary users, selects the secondary user with the maximum signal - to - noise ratio as the cluster - head user of the corresponding cluster. Taking this cluster - head user as the fusion center of the corresponding cluster, it fuses the spectrum sensing results of other secondary users within this cluster to obtain the cooperative detection result of this cluster; finally, the spectrum sensing fusion center performs fusion detection according to the global detection probability and global false - alarm probability within the corresponding cluster sent by each cluster - head user, and takes this fusion detection result as the final multi - band cooperative spectrum detection result, so as to adapt to the change of the signal energy received by the secondary users, improve the detection performance of the secondary users, reduce the calculation amount of the spectrum sensing fusion center, and improve the cooperative detection efficiency. However, this scheme conducts detection in the time - domain and frequency - domain dimensions, and the characteristics of the signal angle dimension have not been fully utilized. Summary of the Invention

[0005] The present invention provides a spectrum sensing method based on array spatial spectrum, which senses the primary user signal through the spatial spectrum of the signal.

[0006] To solve the above - mentioned technical problems, the technical solution of the present invention is as follows:

[0007] A spectrum sensing method based on array spatial spectrum includes the following steps:

[0008] S1: Initialize the spatial spectrum P b (θ i ) and the number of times t for superimposing the spatial spectra;

[0009] S2: Receive the received data of the array elements;

[0010] S3: Calculate the sample covariance matrix according to the received data of the array elements;

[0011] S4: According to the sample covariance matrix, use the CBF method to calculate the spatial spectrum F c (θ i ) obtained from the current received data;

[0012] S5: Use P c (θ i ) to construct a test statistic T1 and solve the threshold γ1;

[0013] S6: If T1 > γ1, output the result that the primary user exists and output the estimated signal arrival angle Jump to step S9; if T1 < γ1, determine that the primary user signal does not exist, then jump to step S7;

[0014] S7: Superimpose the spatial spectrum P c (θ i ) detected this time with the spatial spectrum P b (θ i ) retained from the previous detection, construct a test statistic T2 using the superimposed spatial spectrum, and solve for the threshold γ2;

[0015] S8: If T2 > γ2, output that the primary user signal exists and output the estimated signal direction angle Clear the retained spatial spectrum and set t to 1; if T2 < γ2, output that the primary user signal does not exist, retain the currently detected spatial spectrum P c (θ i ) and superimpose it onto P b (θ i ), and increment the superimposition count t by 1;

[0016] S9: End the current sensing, output the result, and return to step S2 for the next sensing.

[0017] Preferably, in step S2, the received data of the array elements is x(n) = [x1(n), x2(n),..., x M (n)] T , where x1(n), x2(n),..., x M (n) are the received data of the 1st array element, the 2nd array element,..., the Mth array element respectively.

[0018] Preferably, in step S3, calculate the sample covariance matrix based on the received data of the array elements, specifically:

[0019]

[0020] In the formula, N is the number of sampling points.

[0021] Preferably, in step S4, calculate the spatial spectrum P c (θ i ) obtained from the current received data, specifically:

[0022] P c (θ i ) = a H (θ i )R x (N)a(θ i )

[0023] In the formula, a(θ i ) is the steering vector, and the expression is:

[0024]

[0025] where λ is the signal wavelength, d is the element spacing, and d = λ / 2;

[0026] Define the search area of the signal arrival angle as Θ = {θ|θ ∈ (-90°, 90°)}, the search interval is Δ, and the total number of search angles is K = 180° / Δ - 1. Then θ i = -90° + iΔ, i = 1, 2,..., K, and θ i is the scanning angle. By converting the scanning angle, the spatial spectrum value at each angle can be obtained.

[0027] Preferably, in step S5, use P c (θ i ) to construct the test statistic T1, specifically:

[0028]

[0029] Preferably, in step S7, use the superimposed spatial spectrum to construct the test statistic T2, specifically:

[0030]

[0031] Preferably, the solution methods for the threshold γ1 in step S5 and the threshold γ2 in S7 are specifically as follows:

[0032] In the case of only noise present, it can be considered that Prob{max{P(θ i )} > γmin{P(θ i )}} ≈ Prob{λ max (R x (N)) > γσ 2} where λ max (R x (N)) represents the largest eigenvalue of the sample covariance matrix, σ 2 represents the noise variance. When the primary user signal is absent, R x (N) can be approximately regarded as a Wishart random matrix. According to the distribution of its largest eigenvalue, the relationship between the false alarm probability P fa and the threshold γ can be obtained as:

[0033]

[0034] In the formula, F1(·) is the cumulative distribution function of the Tracy-Wisdom first-order distribution, and the threshold is obtained by presetting the false alarm probability.

[0035] Preferably, in step S5, the number of samples used is N, and the expression for the threshold γ1 can be obtained as:

[0036]

[0037] Preferably, in step S7, the spatial spectrum P retained by the secondary detection is superimposed b (θ i ) is equivalent to using t·N sampling data, so the expression of the threshold γ2 is:

[0038]

[0039] Preferably, in step S6, when it is determined that the primary user signal exists, the direction-of-arrival angle output In step S8 described above, when it is determined that the primary user signal exists, the direction-of-arrival angle output

[0040] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0041] The present invention proposes a new single-stage spectrum sensing algorithm based on array spatial spectrum. The spatial spectrum of the signal is obtained through the Conventional Beamforming (CBF) algorithm. A test statistic is constructed by the ratio of the maximum and minimum values of the spatial spectrum and compared with a threshold to obtain a judgment result. If the primary user signal exists, the direction-of-arrival angle corresponding to the maximum value of the spatial spectrum is output. In addition, since spectrum sensing is continuously performed, the present invention proposes to retain the spatial spectra previously detected and judged as signal non-existence and superimpose them on the current detection. Compared with the existing energy detection algorithm (ED), the maximum-minimum eigenvalue detection method based on the covariance matrix (MME), and the maximum-minimum delay summation spectral value ratio detection method (MMDS), the solution of the present invention has better detection performance at low signal-to-noise ratios. Additionally, compared with the MMDS method, the angle measurement accuracy is higher at low signal-to-noise ratios. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic flowchart of the method of the present invention.

[0043] Figure 2 It is a schematic diagram of the uniform linear array reception model used in the present invention.

[0044] Figure 3 It is a schematic diagram of the detection performance comparison between the new method proposed by the present invention and the energy detection algorithm (ED), the maximum-minimum eigenvalue detection method based on the covariance matrix (MME), and the maximum-minimum delay summation spectral value ratio detection method (MMDS).

[0045] Figure 4Schematic diagram of the detection probability variation of the new method proposed by the present invention, the energy detection algorithm (ED), the maximum-minimum eigenvalue detection method based on the covariance matrix (MME), and the maximum-minimum delay-added spectral value ratio detection method (MMDS) under different false alarm probabilities.

[0046] Figure 5 Schematic diagram of the comparison of the angle measurement accuracy between the method described in the present invention and the maximum-minimum delay-added spectral value ratio detection method (MMDS) under different signal-to-noise ratios. Detailed implementation manners

[0047] The accompanying drawings are only for illustrative purposes and should not be construed as limitations on this patent;

[0048] To better illustrate this embodiment, some components in the accompanying drawings are omitted, enlarged or reduced, which do not represent the dimensions of the actual product;

[0049] For those skilled in the art, it is understandable that some well-known structures and their descriptions in the accompanying drawings may be omitted.

[0050] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0051] Embodiment 1

[0052] A spectrum sensing method based on array spatial spectrum, as Figure 1 shown, includes the following steps:

[0053] S1: Initialize the spatial spectrum P b (θ i ) reserved for sensing and the number of times t of superimposing the spatial spectrum;

[0054] S2: Receive the received data of the array elements;

[0055] S3: Calculate the sample covariance matrix according to the received data of the array elements;

[0056] S4: According to the sample covariance matrix, use the CBF method to calculate the spatial spectrum P c (θ i ) obtained from the current received data;

[0057] S5: Use P c (θ i ) to construct a test statistic T1 and solve the threshold γ1;

[0058] S6: If T1 > γ1, output the result that the primary user exists and output the estimated signal arrival angle Jump to step S9; if T1 < γ1, determine that the primary user signal does not exist, then jump to step S7;

[0059] S7: Superimpose the spatial spectrum P c (θ i ) detected this time with the spatial spectrum P b (θ i ) retained from the previous detection, construct a test statistic T2 using the superimposed spatial spectrum, and solve for the threshold γ2;

[0060] S8: If T2 > γ2, output that the primary user signal exists and output the estimated signal direction angle Clear the retained spatial spectrum and set t to 1; if T2 < γ2, output that the primary user signal does not exist, retain the currently detected spatial spectrum P c (θ i ) and superimpose it onto P b (θ i ), and increment the superimposition count t by 1;

[0061] S9: End the current sensing, output the result, and return to step S2 for the next sensing.

[0062] This embodiment provides a spectrum sensing algorithm based on array spatial spectrum, which uses the difference in spatial spectra in the cases of the existence of the primary user signal and the existence of only noise for detection. The core of the new algorithm is to save the spatial spectrum detected previously for the current detection. At the same time, a set of superimposition rules is designed to make the best use of the spatial spectrum data. The new algorithm proposed in the present invention includes two judgments, and reasonable thresholds are set for both. Compared with other methods, the method of the present invention achieves higher detection performance and angle measurement accuracy at low signal-to-noise ratios.

[0063] Embodiment 2

[0064] On the basis of Embodiment 1, this embodiment further discloses the following content:

[0065] The uniform linear array reception model used in the present invention is as Figure 2 shown. The received data of the array elements in step S2 is x(n) = [x1(n), x2(n),..., x M (n)] T , where x1(n), x2(n),..., x M (n) are the received data of the 1st element, the 2nd element,..., the Mth element respectively.

[0066] In step S3, calculate the sample covariance matrix according to the received data of the array elements, specifically:

[0067]

[0068] In the formula, N is the number of sampling points.

[0069] The spatial spectrum P obtained by calculating the current received data in step S4 c (θ i ), specifically:

[0070] P c (θ i ) = a H (θ i )R x (N)a(θ i )

[0071] In the formula, a(θ i ) is the steering vector, and the expression is:

[0072]

[0073] where λ is the signal wavelength, d is the element spacing, and d = λ / 2;

[0074] Define the search area of the signal arrival angle as Θ = {θ|θ ∈ (-90°, 90°)}, the search interval is Δ, and the total number of search angles is K = 180° / Δ - 1. Then θ i = -90° + iΔ, i = 1, 2,..., K, θ i is the scanning angle, and the spatial spectrum value at each angle can be obtained by converting the scanning angle.

[0075] In step S5, use P c (θ i ) to construct the test statistic T1, specifically:

[0076]

[0077] In step S7, use the superimposed spatial spectrum to construct the test statistic T2, specifically:

[0078]

[0079] The solution methods for the threshold γ1 in step S5 and the threshold γ2 in S7 are specifically:

[0080] In the case of only noise present, it can be considered that Prob{max{P(θ i )} > γmin{P(θ i )}} ≈ Prob{λ max (R x (N)) > γσ 2} where λ max (R x (N)) represents the maximum eigenvalue of the sample covariance matrix, σ 2 represents the noise variance. When the primary user signal is absent, R x(N) can be approximately regarded as a Wishart random matrix, and the false alarm probability P can be obtained according to the distribution of its maximum eigenvalue. fa The relationship with the threshold γ is:

[0081]

[0082] In the formula, F1(·) is the cumulative distribution function of the Tracy-Wisdom first-order distribution, and the threshold is obtained through the preset false alarm probability.

[0083] In step S5, the number of samples used is N, and the expression of the threshold γ1 can be obtained as:

[0084]

[0085] In step S7, the spatial spectrum P b (θ i ) retained after t times of detection is equivalent to using t·N sampling data, so the expression of the threshold γ2 is:

[0086]

[0087] In step S6, when it is judged that the primary user signal exists, the output direction-of-arrival angle In step S8, when it is judged that the primary user signal exists, the output direction-of-arrival angle

[0088] Embodiment 3

[0089] This embodiment provides a specific embodiment of Embodiment 1 and Embodiment 2. Taking 16 array elements as an example, the primary user signal is set as an analog FM signal.

[0090] In the specific implementation process, the simulation parameters are set as follows: the primary user signal is set as a single FM modulated signal, the signal frequency is 30 MHz, the signal arrival angle is 30°, the electromagnetic wave propagation speed is 3×10 8 m / s, the sampling frequency is 2 kHz, the number of array elements is 16, the array element spacing is 5 m; the channel model is set as Nakagami-m channel (m = 1), the noise type is Gaussian white noise, the false alarm probability is 0.01, and the number of samples for a single detection is 1000; 20000 Monte Carlo experiments are set, and the two situations of the existence of the primary user signal and only noise appear randomly.

[0091] Compare the detection performance of the new methods proposed in Embodiment 1 and Embodiment 2 with the energy detection algorithm (ED), the maximum-minimum eigenvalue detection method based on the covariance matrix (MME), and the maximum-minimum delay-added spectral value ratio detection method (MMDS). Specifically, asFigure 3 As shown, the Proposed curve corresponds to the new algorithm proposed by the present invention; the ED curve corresponds to the energy detection algorithm; the MME curve corresponds to the maximum-minimum eigenvalue detection method based on the covariance matrix; the MMDS curve corresponds to the maximum-minimum delay added spectral value ratio detection method. The detection probabilities of the four algorithms all increase with the increase of the signal-to-noise ratio. The algorithms proposed in Embodiment 1 and Embodiment 2 have the best detection performance, especially at low signal-to-noise ratios. It should be noted that in the interval of signal-to-noise ratio from -30 dB to -26 dB, the detection probability of the ED algorithm seems to be higher because the actual false alarm probability of this algorithm is higher than the preset one.

[0092] The above embodiments show that when the primary user signal is an analog FM signal, the channel environment is a Nakagami-m channel (m = 1), and the signal-to-noise ratio of the secondary user's received signal is relatively low, the methods described in Embodiment 1 and Embodiment 2 can achieve better detection performance.

[0093] Based on the above embodiments, the detection probabilities are obtained for different preset false alarm probabilities, and the ROC curves are plotted. Specifically, when setting the simulation parameters, the signal-to-noise ratio is fixed at -20 dB, and ten preset false alarm probabilities are selected as 0.001, 0.004, 0.007, 0.01, 0.04, 0.07, 0.1, 0.4, 0.7, 1, and the settings of other parameters are the same as those in the above embodiments. The detection performances of the new methods proposed in Embodiment 1 and Embodiment 2 are compared with those of the energy detection algorithm (ED), the maximum-minimum eigenvalue detection method based on the covariance matrix (MME), and the maximum-minimum delay added spectral value ratio detection method (MMDS) as follows:

[0094] As Figure 4 shown, the detection probabilities of the four algorithms all increase with the increase of the false alarm probability, and at the same false alarm probability, the algorithm proposed by the present invention reaches the highest detection probability, further verifying that the proposed algorithm has the best detection performance.

[0095] The above embodiments show that under low signal-to-noise ratios and different preset false alarm probabilities, the methods described in Embodiment 1 and Embodiment 2 can both achieve better detection effects.

[0096] Similarly, the angle measurement accuracies of the methods described in Embodiment 1 and Embodiment 2 and the maximum-minimum delay added spectral value ratio detection method (MMDS) are tested under different signal-to-noise ratios, and the simulation settings are the same as those in the above embodiments. The mean square error (MSE) is selected as the measurement index, and the mean square error between the estimated arrival angle of the algorithm and the true arrival angle is obtained. The performance comparison results are as follows:

[0097] As Figure 5As shown, the angle measurement MSE of both algorithms decreases as the signal-to-noise ratio increases. At low signal-to-noise ratios, the MSE of the methods described in Example 1 and Example 2 is lower, meaning higher angle measurement accuracy. When the signal-to-noise ratio increases to -19 dB, the MSE of both algorithms is almost the same and approaches zero.

[0098] The above embodiments show that at low signal-to-noise ratios, the methods described in Example 1 and Example 2 can achieve high angle measurement accuracy while ensuring a high detection probability.

[0099] The same or similar reference numerals correspond to the same or similar components;

[0100] The terms used to describe the positional relationship in the drawings are for illustrative purposes only and should not be construed as a limitation of this patent;

[0101] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention and are not limitations on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A spectrum sensing method based on array spatial spectrum, characterized in that Including the following steps: S1: Initialize the spatial spectrum P retained by perception b (θ i ) and the number of times t of the superimposed spatial spectrum; S2: Receive the received data of the array elements; S3: Calculate the sample covariance matrix according to the received data of the array elements; S4: According to the sample covariance matrix, use the conventional beamforming CBF method to calculate the spatial spectrum P obtained from the current received data c (θ i ); S5: Use P c (θ i ) to construct a test statistic T1 and solve for the threshold γ1; S6: If T1 > γ1, output the result that the primary user exists and output the estimated signal arrival angle Jump to step S9; if T1 < γ1, determine that the primary user signal does not exist, and then jump to step S7; S7: Take the spatial spectrum P detected this time c (θ i ) and the spatial spectrum P retained from the previous detection b (θ i ) for superposition, construct a test statistic T2 using the superposed spatial spectrum, and solve for the threshold γ2; S8: If T2 > γ2, output that the primary user signal exists and output the estimated signal arrival angle. Clear the remaining spatial spectrum and set t to 1; if T2 < γ2, output that the primary user signal does not exist, and save the currently detected spatial spectrum P c (θ i ) and add it to P b (θ i ), increment the stacking count t by 1. S9: When the current sensing ends, output the result, and return to step S2 for the next sensing; The spatial spectrum P obtained by calculating the currently received data in step S4 c (θ i ), specifically: P c (θ i )=a H (θ i )R x (N)a(θ i ) where R x (N) represents the sample covariance matrix, N is the number of samplings, and a(θ i ) is the steering vector, and the expression is: where λ is the signal wavelength, d is the element spacing, and d = λ / 2; Define the search region of the signal arrival angle as Θ = {θ∣θ∈(-90°,90°)}, the search interval is Δ, and the total number of search angle is K = 180° / Δ - 1. Then θ i = -90° + iΔ, i = 1, 2, …, K, θ i is the scanning angle. By converting the scanning angle, the spatial spectrum value at each angle can be obtained; In step S5, use P c (θ i ) to construct the test statistic T1, specifically: In step S7, the test statistic T2 is constructed using the superimposed spatial spectrum, specifically: The specific methods for solving the threshold γ1 in step S5 and the threshold γ2 in S7 are as follows: In the case where only noise exists, it can be considered that Prob{max{P(θ i )}>γmin{P(θ i )}}≈Prob{λ max (R x (N))>γσ 2}, where λ max (R x (N)) represents the maximum eigenvalue of the sample covariance matrix, σ 2 represents the noise variance. When the primary user signal does not exist, R x (N) can be approximately regarded as a Wishart random matrix, and the false alarm probability P fa and the relationship with the threshold γ is: In the formula, M is the number of array elements, F1(·) is the cumulative distribution function of the Tracy-Wisdom first-order distribution, and the threshold is obtained through a preset false alarm probability.

2. The spectrum sensing method based on array spatial spectrum according to claim 1, wherein In step S2, the received data of the array elements is \(x(n)=[x_1(n),x_2(n),\cdots,x M (n)] T , where \(x_1(n),x_2(n),\cdots,x M (n)\) are the received data of the 1st array element, the 2nd array element, \(\cdots\), the Mth array element respectively.

3. The spectrum sensing method based on array spatial spectrum according to claim 2, characterized in that In step S3, the sample covariance matrix is calculated according to the received data of the array elements, specifically: In the formula, N is the number of samplings.

4. The spectrum sensing method based on array spatial spectrum according to claim 3, wherein In step S5, the number of samplings used is N, and the expression for the threshold γ1 can be obtained as:

5. The spectrum sensing method based on array spatial spectrum according to claim 4, wherein In step S7, the spatial spectrum P retained after t - time detections is superimposed b (θ i ) is equivalent to using t·N sampling data, so the expression for the threshold γ2 is:

6. The spectrum sensing method based on array spatial spectrum according to claim 5, characterized in that In step S6, when it is determined that the primary user signal exists, the output arrival angle In step S8, when it is determined that the primary user signal exists, the output arrival angle

Images