Method for Eliminating Index Ambiguity of Aliased Signals

Abstract

The present application relates to the field of communication technology, and in particular to a method for eliminating index ambiguity of an aliased signal, the method comprising: performing time-frequency positioning based on an acquired time-frequency spectrum of a received signal to obtain time domain information and frequency domain information of a signal of each time-frequency block, and performing time domain conversion on the signal of the time-frequency block to obtain a time domain signal; extracting rank features of the time domain signal of the time-frequency block in each window based on a sliding window algorithm and a rank estimation algorithm to obtain a rank feature vector; calculating a difference point vector based on the rank feature vector and its right shift vector, and determining a demarcation point vector based on the difference point vector; performing signal segmentation on the signal of the time-frequency block using the demarcation point vector to determine the time domain information and frequency domain information of each signal segment, and constructing a signal segmentation information matrix; performing DOA estimation using a MUSIC algorithm to obtain DOA information of each signal segment, thereby effectively eliminating the index ambiguity of the aliased signal and further improving spectrum utilization.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a method for eliminating index ambiguity of an aliased signal. Background Art

[0002] In recent years, increasingly scarce spectrum resources, under traditional static spectrum allocation strategies, have been unable to meet the demand for higher data rates and improved quality of service. In this context, cognitive radio technology has emerged to further improve spectrum efficiency. The essence of cognitive radio technology is spectrum reuse, allowing secondary users (SUs) to use licensed frequency bands when primary users (PUs) are idle or their normal communications are not affected. Therefore, cognitive radio technology requires frequent spectrum sensing to monitor spectrum occupancy. This means that efficient and accurate spectrum sensing is essential for the reuse of spectrum resources in cognitive radio technology.

[0003] In the past few years, scholars at home and abroad have proposed many model-driven spectrum sensing methods and deep learning-based spectrum sensing methods. Generally speaking, these methods are designed for narrowband spectrum sensing, that is, they are only used to determine the presence or absence of signals. For broadband signals, we need to perceive the specific information of the time-frequency resources used by the PU in order to perform more accurate spectrum sharing and access. The reason is that in some application scenarios, the center frequency and sub-channel division of the PU will change adaptively with business needs and channel status, and this prior knowledge is difficult to obtain. In view of this, scholars at home and abroad have studied many time-frequency positioning methods based on deep learning. These time-frequency positioning methods based on deep learning can well obtain the time-frequency information (TFI) of each time-frequency block (TFB) in broadband spectrum sensing.

[0004] However, the above-mentioned time-frequency positioning method based on deep learning cannot know the specific TFI and arrival angle (DOA) of each component signal contained in the TFB. This phenomenon is defined as index ambiguity. There has been no in-depth and effective research on eliminating the index ambiguity of aliased signals. Summary of the Invention

[0005] The main purpose of the embodiments of the present application is to propose a method for eliminating the index ambiguity of the aliased signal, aiming to effectively eliminate the index ambiguity of the aliased signal and obtain the three-dimensional time-frequency-space information of each component in the aliased signal, which is conducive to more refined interference management and further improves spectrum utilization.

[0006] To achieve the above-mentioned purpose, an embodiment of the present application provides a method for eliminating the index ambiguity of an aliasing signal, comprising the following steps: obtaining a time-frequency spectrum of a received signal, performing time-frequency positioning based on the time-frequency spectrum of the received signal, obtaining the time domain information and frequency domain information of the signal of each time-frequency block, and performing time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering and ISTFT algorithm to obtain the time domain signal of the time-frequency block; wherein, the received signal is an aliasing signal; based on a preset sliding window algorithm and a preset rank estimation algorithm, extracting the rank feature of the time domain signal of the time-frequency block in each window to obtain a rank feature vector corresponding to the time-frequency block; based on the rank feature vector and its right shift vector, calculating a difference point vector, and based on the difference point vector, determining a demarcation point vector corresponding to the time-frequency block; using the demarcation point vector to segment the signal of the time-frequency block, determining the time domain information and frequency domain information of each signal segment, and constructing a signal segmentation information matrix; using the MUSIC algorithm to perform DOA estimation on each signal segment to obtain the DOA information of each signal segment.

[0007] To achieve the above-mentioned purpose, an embodiment of the present application further provides a system for eliminating the index ambiguity of an aliased signal, the system comprising: a coarse time-frequency positioning module, a rank feature vector calculation module, a demarcation point vector calculation module, a signal segment segmentation module and a fine time-frequency positioning module; the coarse time-frequency positioning module is used to obtain the time-frequency spectrum of the received signal, perform time-frequency positioning based on the time-frequency spectrum of the received signal, obtain the time domain information and frequency domain information of the signal of each time-frequency block, and perform time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering and ISTFT algorithm to obtain the time domain signal of the time-frequency block, wherein the received signal is an aliased signal; the rank feature vector calculation module is used to obtain the time domain signal of the time-frequency block based on the time-frequency spectrum of the received signal. The preset sliding window algorithm and the preset rank estimation algorithm extract the rank features of the time domain signal of the time-frequency block in each window to obtain the rank feature vector corresponding to the time-frequency block; the demarcation point vector calculation module is used to calculate the difference point vector based on the rank feature vector and its right shift vector, and determine the demarcation point vector corresponding to the time-frequency block based on the difference point vector; the signal segmentation module is used to use the demarcation point vector to perform signal segmentation on the signal of the time-frequency block; the fine time-frequency positioning module is used to determine the time domain information and frequency domain information of each signal segment, construct a signal segmentation information matrix, and use the MUSIC algorithm to perform DOA estimation on each signal segment to obtain the DOA information of each signal segment.

[0008] To achieve the above-mentioned purpose, an embodiment of the present application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for eliminating the index ambiguity of the aliasing signal.

[0009] To achieve the above objectives, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned method for eliminating index ambiguity of an aliased signal when executed by a processor.

[0010] The embodiment of the present application proposes a method for eliminating the index ambiguity of the aliased signal, which first uses the time-frequency spectrum of the received signal to perform coarse time-frequency positioning to determine the time domain information and frequency domain information of the signal of each time-frequency block. Since the received signal is an aliased signal, the coarse time-frequency positioning can only obtain the time domain information and frequency domain information of the entire time-frequency block, and cannot know the time domain information, frequency domain information, and DOA information of each signal segment in the time-frequency block. At this time, fine time-frequency positioning is required, and this process is implemented based on signal segmentation. First, based on the preset sliding window algorithm and the preset rank estimation algorithm, the rank features of the time domain signal of the time-frequency block are extracted in each window to obtain the rank feature vector corresponding to the time-frequency block. Then, based on the rank feature vector and its right shift vector, the difference point vector is calculated, and based on the difference point vector, the demarcation point vector corresponding to the time-frequency block is determined. The demarcation point vector is used to segment the signal of the time-frequency block. This effectively eliminates the index ambiguity of the aliasing signal, and the time domain information and frequency domain information of each signal segment are known. At this time, the MUSIC algorithm is used to estimate the DOA of each signal segment to obtain the DOA information of each signal segment. That is, the time-frequency and space-time three-dimensional information of each component in the aliasing signal is accurately obtained, which helps to carry out more refined and accurate interference management and further improve the spectrum utilization.

[0011] Optionally, the method extracts the rank feature of the time domain signal of the time-frequency block in each window based on the preset sliding window algorithm and the preset rank estimation algorithm to obtain the rank feature vector corresponding to the time-frequency block, including: dividing the time domain signal of the time-frequency block into several windows of equal length based on the time domain information of the signal of the time-frequency block and the preset sliding window algorithm; performing rank estimation on each window in turn based on the preset rank estimation algorithm and the forward backward spatial smoothing algorithm (FBSS) to obtain the rank feature corresponding to each window; and forming the rank feature vector corresponding to the time-frequency block based on the rank features corresponding to each window; wherein the rank feature vector is expressed by the following formula:

[0012]

[0013] Where Q represents the total number of windows, Represents the result of rank estimation for the qth window, F rank Represents the rank feature vector corresponding to the determined time-frequency block. Using the forward-backward spatial smoothing algorithm in the rank estimation process can effectively improve the accuracy of rank estimation and thus improve the accuracy of signal segmentation.

[0014] Optionally, the difference point vector is calculated based on the rank feature vector and its right shift vector, and the demarcation point vector corresponding to the time-frequency block is determined based on the difference point vector, specifically including: removing the last element in the rank feature vector, shifting each element one position to the right, obtaining the right shift vector of the rank feature vector, and subtracting the right shift vector of the rank feature vector from the rank feature vector to calculate the difference point vector; using the subsequence majority comparison method, the difference point vector is preliminarily screened to obtain the initial demarcation point vector; based on the characteristics of the received signal, a threshold parameter is determined, and the threshold parameter is used to filter out the false demarcation points in the initial demarcation point vector to obtain the demarcation point vector corresponding to the time-frequency block. In the process from the difference point vector to the demarcation point vector, two screenings are performed. The first screening filters out all the difference points that can be used as demarcation points. The second screening carefully considers the characteristics that each signal component in the aliasing signal usually lasts for a period of time, and the overlapping part between the two signal components will not be too short, and removes the false demarcation points, further improving the accuracy of signal segmentation, thereby improving the accuracy of obtaining three-dimensional information in time, frequency and space.

[0015] Optionally, the subsequence majority comparison method is used to perform a preliminary screening of the difference point vector to obtain an initial demarcation point vector, including: traversing each difference point in the difference point vector, and according to a preset comparison window length, demarcating a first comparison window to the left with the current difference point as the end point, and demarcating a second comparison window to the right with the current difference point as the starting point; calculating the mode of each element in the first comparison window, and calculating the mode of each element in the second comparison window; if the mode corresponding to the first comparison window is different from the mode corresponding to the second comparison window, retaining the current difference point, otherwise, discarding the current difference point; generating an initial demarcation point vector based on the difference points retained in the difference point vector.

[0016] Optionally, the threshold parameter is used to filter out false demarcation points in the initial demarcation point vector by the following formula to obtain a demarcation point vector corresponding to the time-frequency block:

[0017]

[0018] in, represents the initial demarcation point vector, represents the length of the initial demarcation point vector, δ represents the threshold parameter, represents the kth element in the initial demarcation point vector, represents the k-1th element in the initial demarcation point vector, Represents the dividing point vector corresponding to the time-frequency block.

[0019] Optionally, the DOA information of each signal segment is obtained by using the MUSIC algorithm to estimate the DOA of each signal segment using the following formula:

[0020]

[0021] Where B represents the angle search space, Indicates the steering vector corresponding to the b-th angle, and the superscript H indicates the conjugate transpose operation. represents the noise subspace corresponding to the current signal segment, P MUSIC (θ b ) represents the spatial spectrum estimation result of the b-th angle using the MUSIC algorithm, Indicates the estimated DOA information.

[0022] Optionally, the received signal is obtained by a SU equipped with a uniform linear array consisting of M antennas sensing the current spectrum, and P single-antenna PUs use orthogonal frequency division multiplexing to share the current frequency band and transmit signals, and the P PUs and SU are on the same horizontal plane; wherein M and P are both integers greater than 1; obtaining the time-frequency spectrum of the received signal includes: using an STFT algorithm to determine the time-frequency spectrum of the received signals of each antenna of the SU; averaging the time-frequency spectra corresponding to different antennas to obtain the time-frequency spectrum of the received signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flowchart of a method for eliminating index ambiguity of an aliased signal provided by an embodiment of the present application;

[0024] Figure 2 This is a schematic diagram of multi-dimensional information extraction of aliased signals provided by an embodiment of the present application;

[0025] Figure 3 is a schematic diagram of a method for eliminating index ambiguity of an aliased signal provided by an embodiment of the present application;

[0026] Figure 4 This is a schematic diagram of constructing a rank feature vector provided by an embodiment of the present application;

[0027] Figure 5This is a flowchart of an embodiment of the present application, which provides a method for calculating a difference point vector based on a rank feature vector and its right shift vector, and determining a demarcation point vector corresponding to a time-frequency block based on the difference point vector;

[0028] Figure 6 This is a schematic diagram of the principle of the subsequence mode comparison method provided by one embodiment of the present application;

[0029] Figure 7 is a structural diagram of a system for eliminating index ambiguity of an aliased signal provided by another embodiment of the present application;

[0030] Figure 8 Another embodiment of the present application provides a time-frequency spectrum of a received signal;

[0031] Figure 9 is a schematic diagram of a rank estimation result of a signal R provided by another embodiment of the present application;

[0032] Figure 10 Another embodiment of the present invention provides a signal Schematic diagram of the rank estimation results;

[0033] Figure 11 Another embodiment of the present invention provides a signal Schematic diagram of the rank estimation results after sliding window;

[0034] Figure 12 Another embodiment of the present application provides a signal Schematic diagram of the rank estimation results after FBSS;

[0035] Figure 13 Another embodiment of the present invention provides a signal Schematic diagram of the rank estimation results after sliding window and FBSS;

[0036] Figure 14 is a schematic diagram of the effect of the comparison window U on the segmentation performance provided by another embodiment of the present application;

[0037] Figure 15 is a schematic diagram of the effect of the threshold parameter δ on segmentation performance provided by another embodiment of the present application;

[0038] Figure 16 This is a structural diagram of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, each embodiment of the present application will be described in detail below with reference to the accompanying drawings. However, it will be understood by those skilled in the art that in each embodiment of the present application, many technical details are proposed to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in the present application can be implemented. The division of the following embodiments is for convenience of description and should not constitute any limitation on the specific implementation of the present application. The various embodiments can be combined and referenced with each other under the premise of no contradiction.

[0040] An embodiment of the present application proposes a method for eliminating the index ambiguity of an aliased signal, which is applied to a server. The implementation details of the method for eliminating the index ambiguity of an aliased signal proposed in this embodiment are described in detail below. The following content is only the implementation details provided for ease of understanding and is not necessary for implementing this solution.

[0041] The process of the method for eliminating the index ambiguity of the aliased signal proposed in this embodiment can be as follows: Figure 1 Shown, including:

[0042] Step 101: obtain the time-frequency spectrum of the received signal, perform time-frequency positioning based on the time-frequency spectrum of the received signal, obtain the time domain information and frequency domain information of the signal of each time-frequency block, and perform time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering and ISTFT algorithm to obtain the time domain signal of the time-frequency block. The received signal is an aliased signal.

[0043] In the specific implementation, the server first obtains the time-frequency spectrum of the received signal, performs time-frequency positioning (coarse time-frequency positioning) based on the time-frequency spectrum of the received signal, and obtains the time domain information and frequency domain information of the signal of each time-frequency block. Since the received signal is an aliased signal, coarse time-frequency positioning can only obtain the time domain information and frequency domain information of the entire time-frequency block, and cannot know the time domain information, frequency domain information, and DOA information of each signal segment in the time-frequency block. In order to perform more refined interference management and achieve efficient spectrum reuse, the server needs to perform fine time-frequency positioning, that is, to segment the signal contained in the time-frequency block. The key is to find the accurate segmentation point to determine the correspondence between the signal segment and the PU. The server can perform time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering and ISTFT algorithm to obtain the time domain signal of the time-frequency block in preparation for segmentation.

[0044] In one example, the multi-dimensional information extraction process of the aliased signal can be as follows: Figure 2 As shown, the information to be extracted includes time domain information, frequency domain information and DOA information.

[0045] The system model considered in this embodiment can be as follows Figure 3 As shown in the figure, P single-antenna PUs share the current frequency band and transmit signals using orthogonal frequency division multiplexing. Their digital modulation type can be randomly selected from BPSK, QPSK, and 16QAM. A single SU equipped with a uniform linear array consisting of M antennas perceives the current spectrum environment. The PU signals impact the SU's uniform linear array from the far field, placing all PUs and SUs on the same horizontal plane. Both M and P are integers greater than 1.

[0046] The sample of the i-th snapshot received by SU can be expressed as:

[0047]

[0048] in, represents an independent and identically distributed complex Gaussian noise vector with a mean of 0 and a variance of s p (i) represents the transmitted signal of the p-th PU, It represents the received signal of SU. The upper right corner T represents the transposition operation. p represents the channel of the p-th PU, ζ p and denote the complex channel gain and array response vector of the channel of the p-th PU, respectively. represents the DOA of each PU, d represents the element spacing of the uniform linear array, λ represents the wavelength, and L represents the sampling signal length.

[0049] The matrix form of the SU received signal can be expressed as: R = HS + W.

[0050] In order to predict the TFI of each PU, the server can use the STFT algorithm to represent the time-frequency spectrum of the received signal, and transform the signal R received by the mth antenna into m Reconstructed into matrix Y m , Y m The specific expression is:

[0051]

[0052] Among them, r m (i) represents the sample of the ith snapshot of the mth antenna, N represents the number of sampling points, and V = L / N. By stacking the spectrum of each column, the time-frequency spectrum of the signal received by the mth antenna is X m It can be expressed as

[0053] The time-frequency spectrum X of the received signal can be obtained by averaging the time-frequency spectrum corresponding to different antennas.

[0054] At this point, the server can perform coarse time-frequency positioning and obtain the time domain information (tstart , t end ) and frequency domain information (f lower , f upper ), t start Indicates the starting time point, t end Indicates the end time point, f lower Indicates the lowest frequency point, f upper Indicates the highest frequency point. Figure 3 Fine time-frequency positioning shown.

[0055] In one example, the server performs time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering, and the ISFT algorithm to obtain the time domain signal of the time-frequency block, which can be achieved by the following formula:

[0056]

[0057] Where T(·) represents time domain truncation, h(·) represents frequency domain filtering, and ISTFT(·) represents the ISTFT algorithm. Represents the time-domain signal of the time-frequency block.

[0058] Step 102: Based on a preset sliding window algorithm and a preset rank estimation algorithm, extract the rank feature of the time domain signal of the time-frequency block in each window to obtain a rank feature vector corresponding to the time-frequency block.

[0059] In a specific implementation, after obtaining the time domain signal of the time-frequency block, the server can extract the rank feature of the time domain signal of the time-frequency block in each window based on the preset sliding window algorithm and the preset rank estimation algorithm to obtain the rank feature vector corresponding to the time-frequency block. Taking into account the presence of aliasing signals, the differences of certain signal features will be weakened. Therefore, in order to determine the correspondence between signal segmentation and PU, it is necessary to analyze and select features with significant differences. As is well known, the covariance matrix rank of the received signal is a specific representation of the number of PUs in the aliasing signal. Therefore, this embodiment selects rank features (covariance matrix rank) as features with significant differences.

[0060] In one example, the process of constructing the rank feature vector can be as follows Figure 4As shown. The server first divides the time domain signal of the time-frequency block into several windows of equal length based on the time domain information of the signal of the time-frequency block and the preset sliding window algorithm, and then performs rank estimation on each window in turn based on the preset rank estimation algorithm and forward and backward spatial smoothing algorithm to obtain the rank feature corresponding to each window. Finally, the rank feature vector corresponding to the time-frequency block is composed based on the rank feature corresponding to each window. The traditional rank estimation method is to perform rank estimation on the entire received signal, which can only obtain the number of signal sources contained in the entire received signal, and cannot reflect the changes in the characteristics of the aliased signal. To this end, the server combines the sliding window algorithm for rank estimation, uses the sliding window method to extract the rank features of the extracted time-frequency block signal, constructs the rank feature vector, and segments the signal by analyzing the changes in the rank estimation results in the rank feature vector. It is also worth noting that after time domain truncation and frequency domain filtering, the independence between different PU signals will be destroyed, resulting in a decrease in rank estimation performance. Therefore, the server uses the forward and backward spatial smoothing algorithm in the rank estimation process, which can effectively improve the accuracy of rank estimation, thereby improving the accuracy of signal segmentation.

[0061] In one example, the rank eigenvector is represented by the following formula:

[0062]

[0063] Where Q represents the total number of windows, Represents the result of rank estimation for the qth window, F rank Represents the rank feature vector corresponding to the determined time-frequency block.

[0064] In one example, the optional rank estimation algorithms include: Akaike Information Criterion (AIC), Minimum Description Length (MDL), Wishart-matrix's largest eigenvalue (WME), and Gerschgorin Disk Estimator (GDE).

[0065] Step 103: Based on the rank eigenvector and its right shift vector, a difference point vector is calculated, and based on the difference point vector, a demarcation point vector corresponding to the time-frequency block is determined.

[0066] In a specific implementation, after calculating the rank feature vector, the server can calculate the difference point vector based on the rank feature vector and its right shift vector, and determine the demarcation point vector corresponding to the time-frequency block based on the difference point vector.

[0067] In one example, the server calculates the difference point vector based on the rank feature vector and its right shift vector, and determines the demarcation point vector corresponding to the time-frequency block based on the difference point vector. Figure 5 The steps shown in the figure are as follows:

[0068] Step 1031 , remove the last element in the rank feature vector, shift each element right by one position to obtain a right-shifted vector of the rank feature vector, and subtract the right-shifted vector of the rank feature vector from the rank feature vector to calculate a difference point vector.

[0069] In the specific implementation, the server first obtains the rank feature vector F rank The server removes the last element in the rank feature vector and shifts the remaining elements one position to the right to obtain the right shift vector of the rank feature vector. Server with F rank minus The difference point vector can be calculated

[0070] In one example, Figure 6 shows a rank feature vector F rank and its right shift vector Subtracting the two gives the difference point vector

[0071] Step 1032: Use the subsequence majority comparison method to preliminarily screen the difference point vectors to obtain the initial demarcation point vector.

[0072] In a specific implementation, after calculating the difference point vector, the server can traverse each difference point in the difference point vector, and according to the preset comparison window length, use the current difference point as the end point to delimit the first comparison window to the left, and use the current difference point as the starting point to delimit the second comparison window to the right. Then, the mode of each element in the first comparison window is calculated, and the mode of each element in the second comparison window is calculated, and compared. If the mode corresponding to the first comparison window is different from the mode corresponding to the second comparison window, the current difference point is retained. Otherwise, the current difference point is discarded. Finally, based on the difference points retained in the difference point vector, the initial demarcation point vector can be generated.

[0073] In one example, the principle of the subsequence mode comparison method can be as follows Figure 6 shown.

[0074] Step 1033: Determine a threshold parameter based on the characteristics of the received signal, and use the threshold parameter to filter out false demarcation points in the initial demarcation point vector to obtain a demarcation point vector corresponding to the time-frequency block.

[0075] In the specific implementation, although the above subsequence majority comparison method can effectively filter out most of the false demarcation points, there are still some false demarcation points that are not filtered out. To address this problem, by analyzing the characteristics of the received aliased signal, it is found that each signal component usually lasts for a period of time, and the overlap between the two signal components is not too short. Based on this, the server determines the threshold parameter δ based on the characteristics of the received signal, and uses the threshold parameter δ to filter out all false demarcation points in the initial demarcation point vector, and obtains the demarcation point vector corresponding to the time-frequency block.

[0076] In one example, the server uses the following formula to filter out false demarcation points in the initial demarcation point vector using a threshold parameter to obtain a demarcation point vector corresponding to the time-frequency block:

[0077]

[0078] in, represents the initial dividing point vector, represents the length of the initial dividing point vector, δ represents the threshold parameter, represents the kth element in the initial demarcation point vector, represents the k-1th element in the initial demarcation point vector, Represents the dividing point vector corresponding to the time-frequency block.

[0079] It is worth noting that in the process from the difference point vector to the demarcation point vector, the server performed two screenings. The first screening screened out all the difference points that could be used as demarcation points. The second screening carefully considered the characteristics that each signal component in the aliasing signal usually lasts for a period of time and the overlapping part between the two signal components will not be too short, and removed the false demarcation points, further improving the accuracy of signal segmentation, and thus improving the accuracy of obtaining three-dimensional information in time, frequency and space.

[0080] Step 104: Use the demarcation point vector to segment the signal of the time-frequency block, determine the time domain information and frequency domain information of each signal segment, and construct a signal segmentation information matrix.

[0081] In a specific implementation, after obtaining the demarcation point vector, the server can use the demarcation point vector to segment the signal of the time-frequency block, determine the time domain information and frequency domain information of each signal segment, and construct a signal segmentation information matrix G.

[0082] In one example, based on the above analysis of the signal feature change mechanism, the server can segment the aliased signal in the time-frequency block through the designed signal segmentation algorithm. By converting the subscript q∈[1,Q] corresponding to the sliding window into the time range t∈[1,V] corresponding to the time-frequency spectrum X, the time-frequency information of each signal segment can be determined. Then, the server can construct the signal segmentation information matrix G, which includes the number of signals corresponding to each signal segment. Time domain information (t start ,t end ) and frequency domain information (f lower ,f upper ).

[0083] Step 105: Use the MUSIC algorithm to estimate the DOA of each signal segment to obtain the DOA information of each signal segment.

[0084] In a specific implementation, after the server determines the time domain information and frequency domain information of each signal segment, it can use the MUSIC algorithm to estimate the DOA of each signal segment to obtain the DOA information of each signal segment, thereby obtaining the time-frequency-space three-dimensional information of each component in the aliased signal.

[0085] In one example, the server uses the MUSIC algorithm to estimate the DOA of each signal segment using the following formula to obtain the DOA information of each signal segment:

[0086]

[0087] Where B represents the angle search space, Indicates the steering vector corresponding to the b-th angle, and the superscript H indicates the conjugate transpose operation. represents the noise subspace corresponding to the current signal segment, P MUSIC (θ b ) represents the spatial spectrum estimation result of the b-th angle using the MUSIC algorithm, Indicates the estimated DOA information.

[0088] In this embodiment, coarse time-frequency positioning is first performed using the time-frequency spectrum of the received signal to determine the time domain information and frequency domain information of the signal of each time-frequency block. Since the received signal is an aliased signal, coarse time-frequency positioning can only obtain the time domain information and frequency domain information of the entire time-frequency block, and cannot know the time domain information, frequency domain information, and DOA information of each signal segment in the time-frequency block. At this time, fine time-frequency positioning is required, and this process is implemented based on signal segmentation. First, based on the preset sliding window algorithm and the preset rank estimation algorithm, the rank features of the time domain signal of the time-frequency block are extracted in each window to obtain the rank feature vector corresponding to the time-frequency block. Then, based on the rank feature vector and its right shift vector, the difference point vector is calculated, and based on the difference point vector, the demarcation point vector corresponding to the time-frequency block is determined. The demarcation point vector is used to segment the signal of the time-frequency block. This effectively eliminates the index ambiguity of the aliasing signal, and the time domain information and frequency domain information of each signal segment are known. At this time, the MUSIC algorithm is used to estimate the DOA of each signal segment to obtain the DOA information of each signal segment. That is, the time-frequency and space-time three-dimensional information of each component in the aliasing signal is accurately obtained, which helps to carry out more refined and accurate interference management and further improve the spectrum utilization.

[0089] Another embodiment of the present application provides a system for eliminating index ambiguity of an aliased signal. The details of the system for eliminating index ambiguity of an aliased signal proposed in this embodiment are described in detail below. The following content is only implementation details provided for ease of understanding and is not required for implementing this embodiment. Figure 7 This is a schematic diagram of the system for eliminating the index ambiguity of the aliased signal proposed in this embodiment, including: a coarse time-frequency positioning module 201, a rank feature vector calculation module 202, a demarcation point vector calculation module 203, a signal segment segmentation module 204 and a fine time-frequency positioning module 205.

[0090] The coarse time-frequency positioning module 201 is used to obtain the time-frequency spectrum of the received signal, perform time-frequency positioning based on the time-frequency spectrum of the received signal, obtain the time domain information and frequency domain information of the signal of each time-frequency block, and perform time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering and ISTFT algorithm to obtain the time domain signal of the time-frequency block, wherein the received signal is an aliased signal.

[0091] The rank feature vector calculation module 202 is used to extract the rank feature of the time domain signal of the time-frequency block in each window based on a preset sliding window algorithm and a preset rank estimation algorithm to obtain the rank feature vector corresponding to the time-frequency block.

[0092] The demarcation point vector calculation module 203 is configured to calculate a difference point vector based on the rank feature vector and its right shift vector, and determine a demarcation point vector corresponding to the time-frequency block based on the difference point vector.

[0093] The signal segmentation module 204 is configured to segment the signal of the time-frequency block using the demarcation point vector.

[0094] The fine time-frequency positioning module 205 is used to determine the time domain information and frequency domain information of each signal segment, construct a signal segmentation information matrix, and use the MUSIC algorithm to perform DOA estimation on each signal segment to obtain the DOA information of each signal segment.

[0095] It is not difficult to find that this embodiment is a system embodiment corresponding to the above-mentioned method embodiment, and this embodiment can be implemented in conjunction with the above-mentioned method embodiment. The relevant technical details and technical effects mentioned in the above-mentioned embodiments are still valid in this embodiment, and to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above-mentioned embodiments.

[0096] It is worth mentioning that all modules involved in this embodiment are logical modules. In actual applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, to highlight the innovation of this application, this embodiment does not include units that are not closely related to solving the technical problem proposed by this application. However, this does not mean that other units do not exist in this embodiment.

[0097] In order to verify the effectiveness of the method for eliminating the index ambiguity of the aliased signal proposed in the present application, relevant situations of simulation experiments are demonstrated in another embodiment of the present application.

[0098] Three PUs (P = 3) share the current frequency band, with PU1 and PU2 using the same band, and PU3 occupying a different band from PU1 and PU2. The SU uses a uniform linear array with M = 12 for signal reception. The carrier frequency is 5.8 GHz, the sampling rate is 16 MHz, the sampling time is 3.4 ms, and the received signal length is 54,400. The number of discrete Fourier transform points in the short-time Fourier transform is 256, and the size of the time-frequency spectrum X is 256 × 212. In the sliding window-based feature vector construction, the window length and step size are set to 500 and 100, respectively. Furthermore, the DOAs of different PUs are randomly generated in the range [-90°, 90°] and are unique to each other.

[0099] Figure 8 An example of time-frequency positioning is given. Figure 8 is the time-frequency spectrum of the received signal, where time-frequency block 1 is the mixed signal of PU1 and PU2, and time-frequency block 2 is the signal of PU3. The time-frequency information corresponding to each PU is shown in Table 1, and the time-frequency information corresponding to each time-frequency block is shown in Table 2.

[0100] Table 1: Time-frequency information corresponding to each PU

[0101] <![CDATA[f lower ]]> <![CDATA[f upper ]]> <![CDATA[t start ]]> <![CDATA[t end ]]> PU1 76 108 41 212 PU2 76 108 0 110 PU3 138 202 0 164

[0102] Table 2: Time-frequency information corresponding to each time-frequency block

[0103] <![CDATA[f lower ]]> <![CDATA[f upper ]]> <![CDATA[t start ]]> <![CDATA[t end ]]> Time-frequency block 1 76 108 0 212 Time-frequency block 2 138 202 0 164

[0104] Based on the time-frequency positioning results, the time domain signal corresponding to each time-frequency block can be extracted Since accurate rank feature extraction is the prerequisite for subsequent analysis of signal feature changes, the performance of different rank estimation criteria is analyzed here. Figure 9 is the rank estimation result of signal R, Figure 10 For signal The rank estimation results of Figure 11 For signal The rank estimation result after sliding window is: Figure 12 For signal The rank estimation result after FBSS is: Figure 13 For signal The rank estimation results after sliding window and FBSS. Figure 9 It can be seen from the figure that for the received signal R, except for AIC, the rank estimation accuracy of the other three methods can reach more than 95%, and as the signal-to-noise ratio increases, the accuracy of MDL and WME gradually approaches 1. For example, Figure 10 As shown in , except for GDE, the performance of the other three methods has been seriously deteriorated. This is because after time domain truncation and frequency domain filtering, the independence between the received signals is destroyed, resulting in the deterioration of the performance of the AIC, MDL and WME algorithms. Figure 11 As shown, the sliding window method is used to During the analysis, except for GDE, the estimation accuracy of the other three methods decays to 0. This is because when using a sliding window, the signal samples used for rank feature extraction are only the samples within each window.

[0105] In order to reduce the impact of time domain interception and frequency domain filtering, we use FBSS technology to filter the signal Perform analysis, such as Figure 12 and Figure 13 As shown in the figure, we can see that after FBSS processing, the rank estimation accuracy is improved. In summary, the GDE method combined with FBSS can show the best performance in the rank estimation process, so this method is used to construct the rank feature vector.

[0106] In the designed signal segmentation algorithm, the comparison window (window set) U and the threshold parameter δ have a significant impact on the accuracy of the demarcation point vector. Therefore, the server defines the following two indicators, segmentation accuracy ρ and segmentation error ε, and conducts a simulation analysis on the impact of these two parameters. Specifically, Figure 8 As shown in the time-frequency spectrum, time-frequency block 1 contains three segments, each with a rank of 1, 2, and 1, while time-frequency block 2 contains a segment with a rank of 1. Therefore, when the segmentation result of time-frequency block 1 is 3 and the ranks of each signal segment are 1, 2, and 1, the segmentation result is considered correct; otherwise, the segmentation result is considered incorrect. Based on the correct segmentation, the segmentation error is measured as follows:

[0107]

[0108] in, and v true are the estimated cutoff point vector and the true cutoff point vector, respectively, and ‖·‖2 represents the L2 norm.

[0109] Figure 14 This article presents an analysis of segmentation performance for five different window sets. The threshold parameter δ is set to 10. Overall, for different window sets, segmentation accuracy increases and stabilizes with increasing signal-to-noise ratio, while segmentation error decreases and stabilizes with increasing signal-to-noise ratio. Furthermore, when U = [3, 4, 5], segmentation accuracy outperforms other window sets, while segmentation error is similar to that of other window sets.

[0110] Figure 15 This article shows an analysis of segmentation performance under different threshold parameters. The window set is set to U = [3, 4, 5]. It can be seen that when δ = 10, the segmentation accuracy is superior to other settings, approaching 90%. While the segmentation error is larger than that of other settings, it is still within an acceptable range. Given that accurate segmentation is a prerequisite, δ = 10 was chosen.

[0111] From the above, we can see that the signal segmentation algorithm designed in this application is effective in segmenting the aliased signal. Based on the above segmentation results, the signal segmentation information matrix G can be obtained by converting the index of the sliding window into the time domain index of the time-frequency spectrum X. Based on G, the signal corresponding to each segment can be extracted, and the corresponding DOA can be estimated using the MUSIC algorithm. Table 1 shows an example of the signal segmentation information matrix G and the corresponding DOA, where: is the signal number estimation result for the corresponding signal segment. It is worth noting that according to the time information of the signal segment in G, a good angle estimation result can be obtained by using only part of the samples, so the segmentation error has little impact on the DOA estimation.

[0112] Table 3: Signal segmentation information matrix G and DOA corresponding to each signal segment

[0113]

[0114] Generally speaking, specific information of each component signal in the received aliased signal, including time-frequency information and DOA information, can be obtained from Table 3, so as to eliminate the index ambiguity and provide more prior information for accurate interference management in the subsequent spectrum reuse process.

[0115] Time-frequency block 1 contains 2 signals. The frequency-domain and time-domain information of the first signal are (f lower , f upper ) = (76, 108) and (t start , t end ) = (0, 109) respectively, and the DOA is 27°. The frequency-domain and time-domain information of the second signal are (f lower , f upper ) = (76, 108) and (t start , t end ) = (43, 212) respectively, and the DOA is -14°.

[0116] Time-frequency block 2 contains 1 signal. Its frequency-domain information and time-domain information are (f lower , f upper ) = (138, 202) and (t start , t end ) = (0, 164) respectively, and the DOA is 53°.

[0117] Another embodiment of this application relates to an electronic device, as Figure 16 shown, including: at least one processor 301; and a memory 302 communicatively connected to the at least one processor 301; wherein, the memory 302 stores instructions executable by the at least one processor 301, and the instructions are executed by the at least one processor 301 to enable the at least one processor 301 to execute a method for eliminating the index ambiguity of an aliased signal described in each of the above method embodiments.

[0118] The memory and processor are connected using a bus, which can include any number of interconnected buses and bridges. The bus connects various circuits of one or more processors and memories. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits. These are all well known in the art and therefore will not be described further in this article. The bus interface is used to provide an interface between the bus and the transceiver. The transceiver can be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. Data processed by the processor is transmitted on a wireless medium via an antenna. Furthermore, the antenna also receives data and transmits it to the processor.

[0119] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0120] Another embodiment of the present application relates to a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for eliminating index ambiguity of an aliased signal as described in the above method embodiments.

[0121] That is, those skilled in the art will understand that all or part of the steps in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a program, which is stored in a storage medium and includes a number of instructions for causing a device (which may be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes, but is not limited to, various media that can store program code, such as read-only memory (ROM), random access memory (RAM), USB flash drive, mobile hard disk, magnetic disk or optical disk.

[0122] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present application, and that in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present application.

Claims

1. A method for eliminating index ambiguity of an aliased signal, characterized in that: include: Obtain a time-frequency spectrum of the received signal, perform time-frequency positioning based on the time-frequency spectrum of the received signal, obtain time domain information and frequency domain information of the signal of each time-frequency block, and perform time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering and I S TFT algorithm to obtain the time domain signal of the time-frequency block; wherein the received signal is an aliased signal; Based on a preset sliding window algorithm and a preset rank estimation algorithm, the rank feature of the time domain signal of the time-frequency block is extracted in each window to obtain a rank feature vector corresponding to the time-frequency block; Based on the rank feature vector and its right shift vector, a difference point vector is calculated, and based on the difference point vector, a demarcation point vector corresponding to the time-frequency block is determined; Using the demarcation point vector to segment the signal of the time-frequency block, determine the time domain information and frequency domain information of each signal segment, and construct a signal segmentation information matrix; The DOA information of each signal segment is obtained by using the MUSIC algorithm to estimate the DOA of each signal segment.

2. The method for eliminating index ambiguity of an aliased signal according to claim 1, wherein: The method of extracting the rank feature of the time domain signal of the time-frequency block in each window based on the preset sliding window algorithm and the preset rank estimation algorithm to obtain the rank feature vector corresponding to the time-frequency block includes: Based on the time domain information of the signal of the time-frequency block and a preset sliding window algorithm, the time domain signal of the time-frequency block is divided into several windows of equal length; Based on the preset rank estimation algorithm and forward and backward spatial smoothing algorithm, the rank of each window is estimated in turn to obtain the rank feature corresponding to each window; The rank feature vector corresponding to the time-frequency block is formed based on the rank feature corresponding to each window; wherein the rank feature vector is expressed by the following formula: Where Q represents the total number of windows, Represents the result of rank estimation for the qth window, F rank Represents the rank feature vector corresponding to the determined time-frequency block.

3. The method for eliminating index ambiguity of an aliased signal according to claim 2, wherein: The calculating a difference point vector based on the rank feature vector and its right shift vector, and determining a demarcation point vector corresponding to the time-frequency block based on the difference point vector, includes: Remove the last element in the rank feature vector, shift each element right by one position to obtain a right-shifted vector of the rank feature vector, and subtract the right-shifted vector of the rank feature vector from the rank feature vector to calculate a difference point vector; Using the subsequence majority comparison method, the difference point vectors are preliminarily screened to obtain the initial demarcation point vector; A threshold parameter is determined based on the characteristics of the received signal, and the false demarcation points in the initial demarcation point vector are filtered out using the threshold parameter to obtain a demarcation point vector corresponding to the time-frequency block.

4. The method for eliminating index ambiguity of an aliased signal according to claim 3, wherein: The method of using the subsequence mode comparison method to preliminarily screen the difference point vectors to obtain the initial demarcation point vectors includes: Traversing each difference point in the difference point vector, and defining a first comparison window to the left with the current difference point as the end point and a second comparison window to the right with the current difference point as the starting point according to a preset comparison window length; Calculating the mode of each element in the first comparison window, and calculating the mode of each element in the second comparison window; If the mode corresponding to the first comparison window is different from the mode corresponding to the second comparison window, the current difference point is retained; otherwise, the current difference point is discarded; An initial demarcation point vector is generated based on the difference points retained in the difference point vector.

5. The method for eliminating index ambiguity of an aliased signal according to claim 4, wherein: The threshold parameter is used to filter out false demarcation points in the initial demarcation point vector by the following formula to obtain the demarcation point vector corresponding to the time-frequency block: in, represents the initial demarcation point vector, represents the length of the initial demarcation point vector, δ represents the threshold parameter, represents the kth element in the initial demarcation point vector, represents the k-1th element in the initial demarcation point vector, Represents the dividing point vector corresponding to the time-frequency block.

6. The method for eliminating index ambiguity of an aliased signal according to any one of claims 1 to 5, characterized in that: The DOA information of each signal segment is obtained by using the MUSIC algorithm to estimate the DOA of each signal segment using the following formula: Where B represents the angle search space, Indicates the steering vector corresponding to the b-th angle, and the superscript H indicates the conjugate transpose operation. represents the noise subspace corresponding to the current signal segment, P MUSIC (θ b ) represents the spatial spectrum estimation result of the b-th angle using the MUSIC algorithm, Indicates the estimated DOA information.

7. The method for eliminating index ambiguity of an aliased signal according to any one of claims 1 to 5, characterized in that: The received signal is obtained by sensing the current spectrum using a SU equipped with a uniform linear array consisting of M antennas. P single-antenna PUs use orthogonal frequency division multiplexing to share the current frequency band and transmit signals. The P PUs and the SU are on the same horizontal plane. Wherein, M and P are both integers greater than 1. The obtaining of the time-frequency spectrum of the received signal includes: Using the STFT algorithm, determine the time-frequency spectrum of the signal received by each antenna of the SU; The time-frequency spectra corresponding to different antennas are averaged to obtain the time-frequency spectra of the received signal.

8. A system for eliminating index ambiguity of an aliased signal, characterized in that: include: Coarse time-frequency positioning module, rank feature vector calculation module, demarcation point vector calculation module, signal segmentation module and fine time-frequency positioning module; The coarse time-frequency positioning module is used to obtain the time-frequency spectrum of the received signal, perform time-frequency positioning based on the time-frequency spectrum of the received signal, obtain the time domain information and frequency domain information of the signal of each time-frequency block, and perform time domain conversion on the signal of the time-frequency block through time domain truncation, frequency domain filtering and ISTFT algorithm to obtain the time domain signal of the time-frequency block, wherein the received signal is an aliased signal; The rank feature vector calculation module is used to extract the rank feature of the time domain signal of the time-frequency block in each window based on a preset sliding window algorithm and a preset rank estimation algorithm to obtain the rank feature vector corresponding to the time-frequency block; The demarcation point vector calculation module is used to calculate a difference point vector based on the rank feature vector and its right shift vector, and determine the demarcation point vector corresponding to the time-frequency block based on the difference point vector; The signal segment segmentation module is used to segment the signal of the time-frequency block using the demarcation point vector; The fine time-frequency positioning module is used to determine the time domain information and frequency domain information of each signal segment, construct a signal segmentation information matrix, and use the MUSIC algorithm to perform DOA estimation on each signal segment to obtain the DOA information of each signal segment.

9. An electronic device, characterized in that: include: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method for eliminating index ambiguity of an aliased signal according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for eliminating index ambiguity of an aliased signal according to any one of claims 1 to 7 is implemented.

Images