A method for identifying a spectrum aliasing wireless signal

Abstract

The application discloses a kind of frequency spectrum aliasing wireless signal depth identification method, including the following steps: to the M frequency spectrum aliasing signal monitored by non-negative matrix decomposition, construct objective function Q1 and solve, obtain the coefficient matrix A of the kind information that the mth frequency spectrum aliasing signal contains m , signal kind consensus matrix A * ;Objective function Q2 is constructed, and A m Further decomposition is carried out, and the position and transmitting power of each signal in the frequency spectrum aliasing signal are obtained, wherein the position of signal is the grid position obtained by grid division to monitoring area.The application integrates signal identification, signal position estimation and transmitting power estimation in a depth identification framework, and multiple parameters of frequency spectrum aliasing signal are decoupled and estimated by multiple NMFs;The application can perform depth identification on frequency spectrum aliasing signal, and identify the kind, position and transmitting power of unknown signal in frequency spectrum aliasing signal.

Description

Technical Field

[0001] This invention relates to the field of spectrum aliasing signal identification, and particularly to a method for deep identification of spectrum aliasing wireless signals. Background Technology

[0002] With the widespread use of wireless communication devices and the application of flexible radio spectrum strategies, such as unlicensed 2.4 GHz spectrum, spectrum sharing, and dynamic frequency selection, aliasing signals are occurring more and more frequently, and some illegal aliasing signals can cause more serious interference.

[0003] Current research on spectral aliasing signal identification typically employs blind source separation techniques such as Independent Component Analysis (ICA), Principal Component Analysis (PCA), and Singular Value Decomposition (SVD). However, these techniques cannot accurately determine the number of mixed signals, maintain the original signal scale, or subsequently perform signal localization and transmit power estimation. In contrast, Non-negative Matrix Factorization (NMF) algorithms can effectively decouple aliased signals through matrix decomposition, simultaneously identifying signal types and determining the number of mixed signals. Due to their superior separation structure, multi-layer NMF algorithms, composed of multiple NMF algorithms, can perform multi-parameter decoupled estimation, enabling signal localization and transmit power estimation based on signal type identification. Furthermore, most blind source separation research focuses on static scenarios. Fixed radio monitoring systems are limited by their static location and signal reception coverage; therefore, radio monitoring systems require flexible deployment capabilities.

[0004] For example, the structured NMF model [1] can separate the spectrum of overlapping signals transmitted on different carriers without knowing the number of unknown signal sources. It is a blind source separation technique, but it cannot identify multiple types of signals with spectral aliasing.

[0005] Unmanned aerial vehicles (UAVs) equipped with spectrum analyzers offer a more flexible approach to dynamically analyzing radio signals in designated areas. Currently, UAV-based radio signal identification and localization technologies primarily rely on the characteristics of received and transmitted signals. UAVs leverage their mobility to move extensively, using spectrum analyzers to analyze signal types and perform direction finding based on parameters such as time, angle, and signal strength. However, most UAV-based surveillance system designs and research focus on locating single-type signals, which presents challenges in environments containing multiple types of unknown signals.

[0006] For example, study [2] proposed a direct positioning method based on simultaneous positioning of multiple transmitters monitored by mobile UAVs, which locates the radiation source through the cooperative direction finding of multiple UAVs. However, this scheme also cannot identify various types of signals with spectral aliasing.

[0007] [1]

[0008] [2]Li, Y.He, Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings and deficiencies of the prior art and provide a method for deep identification of spectrum aliasing wireless signals.

[0010] The objective of this invention is achieved through the following technical solution:

[0011] A method for deep identification of spectral aliasing wireless signals includes the following steps:

[0012] (1) The objective function Q1 is constructed and solved by nonnegative matrix decomposition for the detected M spectral aliasing signals to obtain the coefficient matrix A containing the type information of the m-th spectral aliasing signal. m Consensus Matrix A for Signal Types * ;where A * Used to synthesize all A m ;

[0013] The objective function Q1 is as follows:

[0014]

[0015] In the formula, R m It is the m-th spectral aliasing signal detected, 1≤m≤M; B s It is a radio signal spectrum dictionary. It is the signal type determination matrix for the m-th spectral aliasing signal, which uses 0 and 1 to indicate whether it is that type of signal;

[0016] (2) Construct the objective function Q2, and apply it to A. mFurther decomposition yields the position and transmission power of each signal in the spectral aliasing signal, where the signal position is the grid position obtained by dividing the monitoring area into grids;

[0017] The objective function Q2 is as follows:

[0018]

[0019] In the formula, P m L is a diagonal matrix representing the m-th spectral aliasing signal, where the values ​​on the diagonal correspond to the transmit power of different signals; m This is the position indication matrix for the m-th spectral aliasing signal, where 0 and 1 indicate whether the n-th signal exists at the g-th grid. W m L is the path loss fading matrix of the m-th spectral aliasing signal. * P * They are respectively for L m P m The consensus matrix.

[0020] In step (1), the objective function Q1 is solved alternately by A. m With A * It is convex, respectively with respect to A. m With A * The update rule is obtained by taking the derivative:

[0021]

[0022] In the formula, T represents the matrix transpose operation; This is the transposed radio signal spectrum dictionary;

[0023] Obtain an estimate of the number of unknown signal sources. for:

[0024]

[0025] in It is matrix A * The nth element.

[0026] In step (2), objective functions Q3 and Q4 are constructed, and the objective function Q2 is modified before being solved:

[0027] To address the diagonal matrix constraints and 0-1 integer constraints in Q2, let H... m =P m L m The constraint in Q2 is transformed into: ||(H m ) n ||1=||(H m ) n ||2,(Hm ) n It is H m The nth row vector gives the objective function Q3 as:

[0028]

[0029] In the formula, H m H is the matrix representing the position and transmit power of the unknown signal estimated from the m-th viewpoint; * A consensus matrix that integrates the position and transmission power of unknown signals obtained from all perspectives;

[0030] By transforming the non-convex constraint using a penalty term, we obtain the objective function Q4:

[0031]

[0032] In the formula, λ is H * The penalty factor; ρ is the penalty coefficient; This represents the matrix transpose of a matrix whose elements are all 1s.

[0033] Since Q4 satisfies the KL condition, it can be solved using the proximal alternating linearization minimization algorithm. m With H * The update rules are as follows:

[0034]

[0035] Solving for H yields... * H * The nth row represents the position and transmit power of the nth aliased signal, where the index of the largest element in this row is the grid number where the nth aliased signal is located, and the largest element is the transmit power of the nth aliased signal; T is the matrix transpose operation; 1 G×G It is a G×G matrix of all 1s.

[0036] Define the mixed spectral signal received by the m-th monitoring node in the t-th spectral sampling time slot as s. m (t):

[0037]

[0038] in J m This represents the number of frequency points sampled by the m-th monitoring node within this frequency band. For the n0th unknown signal in J m The standard power spectral density function at each frequency point, and A standard spectrum dictionary for signals that edge nodes can recognize; Let be the transmission power of the n0th unknown signal source in time slot t; Let z be the channel gain coefficient from the n0th unknown signal source to the mth monitoring node. m (t) represents the noise at the m-th monitoring node and follows the rules. The Gaussian distribution.

[0039] The Defined as:

[0040]

[0041] Among them, the channel constant f is the operating center frequency of the unknown signal, c is the speed of light, and E is the speed of light. r,m and These are the antenna gains for the m-th monitoring node and the n0-th unknown signal, respectively. and These represent the channel response and distance between the m-th monitoring node and the n0-th unknown signal, respectively.

[0042] The Defined as:

[0043]

[0044] in and (x) r,m ,y r,m ,h m The coordinates of the n0th unknown signal and the mth monitoring node are respectively located in a coordinate system with the edge node as the origin; the maximum x-coordinate and maximum y-coordinate of the signal monitoring range of this monitoring network are ±x. max ±y max ; In T s The internal value remains unchanged; the link loss from the g-th grid to the m-th monitoring node remains unchanged.

[0045] Meanwhile, this invention provides:

[0046] A server includes a processor and a memory, the memory storing at least one program that is loaded and executed by the processor to implement the above-described method for deep identification of spectral aliasing wireless signals.

[0047] A computer-readable storage medium storing at least one program that is loaded and executed by a processor to implement the above-described method for deep identification of spectral aliasing wireless signals.

[0048] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0049] This invention integrates signal identification, signal location estimation, and transmit power estimation into a deep identification framework. It uses multi-layered NMF to decouple and estimate multiple parameters of spectral aliasing signals. This invention can perform deep identification of spectral aliasing signals and identify the type, location, and transmit power of unknown signals in the spectral aliasing signals. Attached Figure Description

[0050] Figure 1 This is a flowchart of a method for deep identification of spectral aliasing wireless signals.

[0051] Figure 2 A comparison chart showing the signal type identification results under different algorithms.

[0052] Figure 3 A comparison chart showing the signal location estimation results under different algorithms. Detailed Implementation

[0053] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0054] Assume M monitoring nodes are uniformly distributed within the communication range, and the spectrum data generated by each monitoring node's spectrum scan is wirelessly transmitted back to the edge computing node. The detection time for each monitoring node to complete one frequency band is T. s The frequency band spectrum data return time is T. d The total time to complete a frequency band monitoring task is T0 = T s +T d Assume that there are N0 unknown wireless signal sources in the same frequency band within this monitoring area, and the monitoring nodes have no prior information about their signal type, location, transmission power, etc.

[0055] The mixed spectral signal received by the m-th monitoring node in the t-th spectral sampling time slot is:

[0056]

[0057] in J m This represents the number of frequency points sampled by the m-th monitoring node within this frequency band. For the n0th unknown signal in J m The standard power spectral density function at each frequency point, and A standard spectrum dictionary for signals that edge nodes can recognize. Let be the transmission power of the n0th unknown signal source in time slot t. Let z be the channel gain from the n0th unknown signal source to the mth monitoring node. m (t) represents the noise at the m-th monitoring node and follows the rules. The Gaussian distribution. We mainly consider the large-scale fading of wireless propagation in outdoor, scatter-rich environments; therefore, the channel gain coefficient is defined as:

[0058]

[0059] Among them, the channel constant f is the operating center frequency of the unknown signal, c is the speed of light, and E is the speed of light. r,m and These are the antenna gains for the m-th monitoring node and the n0-th unknown signal, respectively. and The channel response and distance between the m-th monitoring node and the n0-th unknown signal are respectively given.

[0060]

[0061] in and (x) r,m ,y r,m ,h m Let x and y be the coordinates of the nth unknown signal and the mth monitoring node in a coordinate system with the edge node as the origin, respectively. The maximum x-coordinate and maximum y-coordinate of the signal monitoring range of this monitoring network are (±x) max ,±y max ). In T s The internal value remains unchanged. The link loss from the g-th grid to the m-th monitoring node remains constant. Define the signal fading dictionary within the region And W m =[q 1,m ,q 2,m ……q G,m ],in It is the signal fading template matrix of the m-th monitoring node.

[0062] like Figure 1 A method for deep identification of spectral aliasing wireless signals includes the following steps:

[0063] (1) The objective function Q1 is constructed and solved by nonnegative matrix decomposition for the detected M spectral aliasing signals to obtain the coefficient matrix A containing the type information of the m-th spectral aliasing signal. m Consensus Matrix A for Signal Types * ;where A * Used to synthesize all A m ;

[0064] The objective function Q1 is as follows:

[0065]

[0066] In the formula, Rm It is the m-th spectral aliasing signal detected, 1≤m≤M; B s It is a radio signal spectrum dictionary. It is the signal type determination matrix for the m-th spectral aliasing signal, which uses 0 and 1 to indicate whether it is that type of signal;

[0067] (2) Construct the objective function Q2, and apply it to A. m Further decomposition yields the position and transmission power of each signal in the spectral aliasing signal, where the signal position is the grid position obtained by dividing the monitoring area into grids;

[0068] The objective function Q2 is as follows:

[0069]

[0070] In the formula, P m L is a diagonal matrix representing the m-th spectral aliasing signal, where the values ​​on the diagonal correspond to the transmit power of different signals; m This is the position indication matrix for the m-th spectral aliasing signal, where 0 and 1 indicate whether the n-th signal exists at the g-th grid. W m L is the path loss fading matrix of the m-th spectral aliasing signal. * P * They are respectively for L m P m The consensus matrix.

[0071] In step (1), the objective function Q1 is solved alternately by A. m With A * It is convex, respectively with respect to A. m With A * The update rule is obtained by taking the derivative:

[0072]

[0073] In the formula, T represents the matrix transpose operation; This is the transposed radio signal spectrum dictionary;

[0074] Obtain an estimate of the number of unknown signal sources. for:

[0075]

[0076] in It is matrix A * The nth element.

[0077] In step (2), objective functions Q3 and Q4 are constructed, and the objective function Q2 is modified before being solved:

[0078] To address the diagonal matrix constraints and 0-1 integer constraints in Q2, let H... m =P m L m The constraint in Q2 is transformed into: ||(H m ) n ||1=||(H m ) n ||2,(H m ) n It is H m The nth row vector gives the objective function Q3 as:

[0079]

[0080] In the formula, H m H is the matrix representing the position and transmit power of the unknown signal estimated from the m-th viewpoint; * A consensus matrix that integrates the position and transmission power of unknown signals obtained from all perspectives;

[0081] By transforming the non-convex constraint using a penalty term, we obtain the objective function Q4:

[0082]

[0083] In the formula, λ is H * The penalty factor; ρ is the penalty coefficient; This represents the matrix transpose of a matrix whose elements are all 1s.

[0084] Since Q4 satisfies the KL (Kurdyka-Lojasiewicz) condition, it is solved using the Proximal Alternating Linearization Minimization (PLAM) algorithm. m With H * The update rules are as follows:

[0085]

[0086] Solving for H yields... * H * The nth row represents the position and transmit power of the nth aliased signal, where the index of the largest element in this row is the grid number where the nth aliased signal is located, and the largest element is the transmit power of the nth aliased signal; T is the matrix transpose operation; 1 G×G It is a G×G matrix of all 1s.

[0087] The objective function Q2 is difficult to solve, but the difficulty is reduced after transformation by Q3 and Q4.

[0088] Meanwhile, this invention provides:

[0089] A server includes a processor and a memory, the memory storing at least one program that is loaded and executed by the processor to implement the above-described method for deep identification of spectral aliasing wireless signals.

[0090] A computer-readable storage medium storing at least one program that is loaded and executed by a processor to implement the above-described method for deep identification of spectral aliasing wireless signals.

[0091] like Figure 2 In signal type identification tasks, the CMNR method proposed in this invention is superior to the traditional BSS method.

[0092] like Figure 3 In signal location estimation tasks, the CMNR method proposed in this invention outperforms the traditional trilateration algorithm.

[0093] Figure 2 and Figure 3 In Chinese, CMNR stands for Collaborative Multilevel Non-negative Matrix Factorization Recognition, JMNR stands for Joint Multilevel Non-negative Matrix Factorization Recognition, CMNR (no consensus) stands for Collaborative Multilevel Non-negative Matrix Factorization Recognition without consensus, JMNR (no consensus) stands for Joint Multilevel Non-negative Matrix Factorization Recognition without consensus, FastICA stands for Fast Independent Component Analysis, PCA stands for Principal Component Analysis, TSVD stands for Truncation Singular Value Decomposition, trilateration stands for Trilateration, and PSO stands for Particle Swarm Optimization.

[0094] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for deep identification of spectral aliasing wireless signals, characterized in that, Includes the following steps: (1) The objective function Q1 is constructed and solved by nonnegative matrix decomposition for the M detected spectral aliasing signals to obtain the coefficient matrix A containing the type information of the m-th spectral aliasing signal. m Consensus Matrix A for Signal Types * ;where A * Used to synthesize all A m ; The objective function Q1 is as follows: In the formula, R m It is the m-th spectral aliasing signal detected, 1≤m≤M; B s It is a radio signal spectrum dictionary. It is the signal type determination matrix for the m-th spectral aliasing signal, which uses 0 and 1 to indicate whether it is that type of signal; The objective function Q1 is obtained by alternately solving A. m With A * It is convex, respectively with respect to A. m With A * The update rule is obtained by taking the derivative: In the formula, T represents the matrix transpose operation; This is the transposed radio signal spectrum dictionary; Obtain an estimate of the number of unknown signal sources. for: in It is matrix A * The nth element; (2) Construct the objective function Q2, and apply it to A. m Further decomposition yields the position and transmission power of each signal in the spectral aliasing signal, where the signal position is the grid position obtained by dividing the monitoring area into grids; The objective function Q2 is as follows: In the formula, P m L is a diagonal matrix representing the m-th spectral aliasing signal, where the values ​​on the diagonal correspond to the transmit power of different signals; m This is the position indication matrix for the m-th spectral aliasing signal, where 0 and 1 indicate whether the n-th signal exists at the g-th grid. W m L is the path loss fading matrix of the m-th spectral aliasing signal. * P * They are respectively for L m P m The consensus matrix.

2. The method for deep identification of spectral aliasing wireless signals according to claim 1, characterized in that, In step (2), objective functions Q3 and Q4 are constructed, and the objective function Q2 is modified before being solved: To address the diagonal matrix constraints and 0-1 integer constraints in Q2, let H... m =P m L m The constraint in Q2 is transformed into: ||(H m ) n ||1=||(H m ) n ||2,(H m ) n It is H m The nth row vector gives the objective function Q3 as: In the formula, H m H is the matrix representing the position and transmit power of the unknown signal estimated from the m-th viewpoint; * A consensus matrix that integrates the position and transmission power of unknown signals obtained from all perspectives; By transforming the non-convex constraint using a penalty term, we obtain the objective function Q4: In the formula, λ is H * The penalty factor; ρ is the penalty coefficient; This represents the matrix transpose of a matrix whose elements are all 1s. Since Q4 satisfies the KL condition, it can be solved using the proximal alternating linearization minimization algorithm. m With H * The update rules are as follows: Solving for H yields... * H * The nth row represents the position and transmit power of the nth aliased signal, where the index of the largest element in this row is the grid number where the nth aliased signal is located, and the largest element is the transmit power of the nth aliased signal; T is the matrix transpose operation; 1 G×G It is a G×G matrix of all 1s.

3. The method for deep identification of spectral aliasing wireless signals according to claim 1, characterized in that, Define the mixed spectral signal received by the m-th monitoring node in the t-th spectral sampling time slot as s. m (t): in J m This represents the number of frequency points sampled by the m-th monitoring node within this frequency band. For the n0th unknown signal in J m The standard power spectral density function at each frequency point, and A standard spectrum dictionary for signals that edge nodes can recognize; Let be the transmission power of the n0th unknown signal source in time slot t; Let z be the channel gain coefficient from the n0th unknown signal source to the mth monitoring node. m (t) represents the noise at the m-th monitoring node and follows the rules. The Gaussian distribution.

4. The method for deep identification of spectral aliasing wireless signals according to claim 3, characterized in that, The Defined as: Among them, the channel constant f is the operating center frequency of the unknown signal, c is the speed of light, and E is the speed of light. r,m and These are the antenna gains for the m-th monitoring node and the n0-th unknown signal, respectively. and These represent the channel response and distance between the m-th monitoring node and the n0-th unknown signal, respectively.

5. The method for deep identification of spectral aliasing wireless signals according to claim 4, characterized in that, The Defined as: in and Let x and y be the coordinates of the n0th unknown signal and the mth monitoring node in a coordinate system with the edge node as the origin, respectively; the maximum x and y coordinates of the signal monitoring range of this monitoring network are ±x. max ±y max ; In T s The internal value remains unchanged; the link loss from the g-th grid to the m-th monitoring node remains unchanged.

6. A server, the server comprising a processor and a memory, characterized in that, The memory stores at least one program, which is loaded and executed by the processor to implement the spectral aliasing wireless signal deep identification method according to any one of claims 1 to 5.

7. A computer-readable storage medium storing at least one program, characterized in that, The program is loaded and executed by a processor to implement the method for deep identification of spectral aliasing wireless signals as described in any one of claims 1 to 5.

Images