Self-adaptive code element detection method and system for spread spectrum signal

Abstract

The invention provides a spread spectrum signal adaptive code element detection method, which comprises the following steps of: acquiring a spread spectrum signal to be detected, and performing signal parameter estimation; generating or updating a signal sparse expression dictionary in a single code element period; constructing a self-adaptive compressed sensing measurement kernel; performing denoising and filtering on the spread spectrum signal to be detected, performing compressed sampling by using an adaptive compressed sensing measurement kernel, and designing a compressed measurement kernel for signal compressed sampling in the future after each compressed sampling; and performing code element detection based on the maximum posterior probability according to the compressed sampling data, the signal sparse expression dictionary and the compressed sensing measurement matrix to obtain a code element detection result. On-line self-adaptive code element detection of spread spectrum signals is realized through system construction of self-adaptive signal sampling and code element judgment logic based on cognitive enhanced compressed sensing; the accuracy of signal code element detection is ensured while the signal sampling frequency is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of spread spectrum signal detection, and in particular to a spread spectrum signal adaptive symbol detection method. Background Art

[0002] Nowadays, spread spectrum technology has been widely used in military and civilian communication fields. In spread spectrum communication, the spectrum of the baseband signal is extended to a relatively wider frequency band through spread spectrum technology. Common spread spectrum technologies include frequency hopping, direct sequence spread spectrum, time hopping, etc. Among them, frequency hopping technology uses frequency hopping to quickly jump the carrier frequency of the signal in a wide frequency band within a certain period of time, thereby achieving the expansion of the overall spectrum; direct sequence spread spectrum technology uses a pseudo-random sequence to modulate the original signal to achieve the expansion of the spectrum; and time hopping technology achieves the expansion of the signal spectrum through pseudo-random selection of signal transmission time. For the cooperative receiving end of the signal, the spread spectrum parameters of the communication are completely known, so the receiving end only needs to despread the received signal to obtain the original baseband signal and detect the signal code element. However, for the non-cooperative (such as supervision, listening) receiving end, the spread spectrum parameters of the communication cannot be accurately known. Therefore, according to the Nyquist sampling theorem, the traditional receiver needs to use a very high frequency to sample the received signal to ensure the integrity of the signal information. In this case, the cost of the system may be greatly increased.

[0003] For the non-cooperation of spread spectrum signals, sparse reconstruction can be used in the processing process; that is, if a signal can be sparsely represented based on a certain dictionary or transformation, then the signal can be reconstructed from sampling data below the Nyquist sampling number. Since at any time, the spread spectrum signal can be sparsely represented (such as frequency hopping signals are sparse in the frequency domain, direct sequence spread spectrum signals are sparse in a certain pseudo-random sequence dictionary, and time hopping signals are sparse in transmission time), the prior art has proposed a variety of technologies for spread spectrum signal sampling reconstruction, parameter estimation and codeword detection. In addition, in order to further improve the accuracy of signal analysis results, some scholars have proposed a sparse reconstruction pre-training method. Although the existing spread spectrum signal analysis method can use a lower sampling frequency to collect signals and obtain signal analysis results. However, there are still disadvantages such as high complexity in reconstructing the time domain or sparse domain of the signal.

[0004] However, in general, the practical application of these methods is limited due to the following defects:

[0005] 1. It is necessary to reconstruct the time domain or sparse domain of the signal, so it has high complexity.

[0006] 2. During the signal measurement and acquisition process, there is a lack of effective use of the collected samples and signal posterior information, which to a certain extent limits the further improvement of the accuracy of the signal analysis results. Summary of the invention

[0007] The object of the present invention is to solve at least one of the above-mentioned technical drawbacks.

[0008] To this end, an object of the present invention is to propose a method for adaptive codeword detection of spread spectrum signals, which realizes online adaptive codeword detection of spread spectrum signals by constructing a system of adaptive signal sampling and codeword judgment logic based on cognitive enhanced compressed sensing; while reducing the signal sampling frequency, the accuracy of signal codeword detection is guaranteed.

[0009] In order to achieve the above object, an embodiment of one aspect of the present invention provides a spread spectrum signal adaptive symbol detection method, comprising the following steps:

[0010] S1. Obtain the spread spectrum signal to be detected and estimate the signal parameters;

[0011] S2. Generate or update a signal sparse expression dictionary within a single symbol period according to the estimated signal parameters;

[0012] S3, constructing an adaptive compressed sensing measurement kernel using the signal sparse expression dictionary;

[0013] S4, after denoising and filtering the spread spectrum signal to be detected, compress it using the adaptive compressed sensing measurement kernel in S3 to obtain compressed sampling data; after each compressed sampling, using the signal sparse expression dictionary, the compressed sensing measurement matrix and the obtained compressed sampling data, based on the linear discrimination analysis system, design a compressed measurement kernel for future signal compression sampling;

[0014] S5. Perform codeword detection based on the maximum a posteriori probability according to the compressed sampling data, the signal sparse expression dictionary and the compressed sensing measurement matrix to obtain a codeword detection result.

[0015] Further preferably, in S1, among the spread spectrum signals to be detected, the spread spectrum signal types include: frequency hopping, direct sequence spread spectrum or time hopping; the signal parameter estimation includes at least: signal code element period, carrier frequency range of frequency hopping signal, pseudo-random sequence of direct sequence spread spectrum signal, transmission time of time hopping signal.

[0016] Further preferably, in S2, the signal sparse expression dictionary can be expressed by the following formula (1):

[0017] Ψ=[ψ 0,1 ,ψ 0,2 ,…,ψ 0,K ,ψ1,1 , ψ 1,2 , …, ψ 1,K , …, ψ R-1,1 , ψ R-1,2 , …, ψ R-1,K (1)

[0019] Among them, each sub-block ψ r,k (0 ≤ r < R, 1 ≤ k ≤ K) represents a column, that is, a possible expression of the signal in one symbol period; R represents the number of symbol types.

[0020] Further preferably, in S3, when constructing the adaptive compressive sensing measurement kernel, first calculate the scatter matrix using the following formula (2):

[0021]

[0022] Among them, l represents the l-th compressive sensing measurement kernel, is the posterior usage probability of the dictionary element ψ obtained by analyzing the measured compressive sensing data and the measurement matrix r,k ; ψ r,k,l is ψ r,k the dictionary element segment composed of the ((l - 1)CR + 1)-th element to the l×CR-th element in ψ Ψ l is the segment composed of the ((l - 1)CR + 1)-th row to the l×CR-th row of the dictionary Ψ; diag() represents the operation of converting a column vector into a diagonal matrix; () H represents the Hermitian operation of the matrix; p is an intermediate vector.

[0023] Further preferably, it also includes adopting a method of mutual cooperation of multiple design units to design the compressive measurement kernel, where the number of adaptive design units can be obtained by the following (3):

[0024]

[0025] Among them, T adp is the time consumed by the measurement kernel for one compression sampling designed by the system; T samp is the time interval between two compression samplings of the system.

[0026] ​Further preferably, in S4, the spread spectrum signal first passes through a pre-filter to filter out interference and noise outside the spread spectrum band; the time domain signal obtained by time-domain multiplication of the compressed measurement kernel and the pre-filtered signal passes through an analog-to-digital converter composed of a low-pass filter and a down-sampler to obtain compressed sensing data of the signal; wherein the low-pass filter acts as an integrator, and the down-sampler operates at a frequency lower than the Nyquist sampling rate.

[0027] Further preferably, in S5, the process of codeword detection includes: calculating the posterior usage probability of each dictionary element by using the Bayesian formula, the compressed sensing matrix and the compressed measurement data;

[0028] The dictionary elements to be used are determined by the maximum a posteriori probability criterion;

[0029] According to the code element carried by the signal in the code element period, determine the code element carried by the dictionary element

[0030] Further preferably, the posterior usage probability of each dictionary element is calculated using the compressed sensing matrix and the compressed measurement data, as shown in formula (4):

[0031]

[0032] Where y and Φ are all compressed sampling data and compressed measurement matrix in one symbol period respectively; P(ψ r,k ) is the dictionary element ψ r,k The prior probability used; p(y|Φ,ψ r,k ) is a given compressed sensing matrix and dictionary element ψ r,k , the probability density of the compressed sampling data y.

[0033] Further preferably, the dictionary elements used are determined by the maximum a posteriori probability criterion, which is expressed by formula (5):

[0034] sym=arg r maxP(ψ r,k |y,Φ) (5)

[0035] Among them, y and Φ are all compressed sampling data and compressed measurement matrix in one code element period respectively.

[0036] According to an embodiment of the present invention, a method for adaptive symbol detection of a spread spectrum signal is provided, which can measure the spread spectrum signal at a lower sampling frequency and detect the signal symbols without reconstructing the original signal. In combination with the hardware cost of the system design and using the a posteriori information of the signal measurement, the compressed sampling matrix is ​​efficiently and adaptively designed to further ensure the accuracy of the symbol detection.

[0037] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0039] Figure 1 It is the overall framework structure of the present invention;

[0040] Figure 2 It is a schematic diagram of the structure of the compressed sensing measurement matrix;

[0041] Figure 3 It is a module structure used for signal symbol detection in the present invention;

[0042] Figure 4 The system structure of the present invention when using three adaptive design units;

[0043] Figure 5 The figure is a performance comparison diagram of symbol error rate using the present invention, random compressed sensing measurement and traditional Nyquist sampling for binary Gaussian filtered frequency hopping signal (BGFSK) codeword detection when the compressed sensing compression ratio is 5;

[0044] Figure 6 The figure is a performance comparison diagram of symbol error rate using the present invention, random compressed sensing measurement and traditional Nyquist sampling for binary Gaussian filtered frequency hopping signal (BGFSK) codeword detection when the compressed sensing compression ratio is 10;

[0045] Figure 7 The figure is a performance comparison diagram of symbol error rate using the present invention, random compressed sensing measurement and traditional Nyquist sampling for binary Gaussian filtered frequency hopping signal (BGFSK) codeword detection when the compressed sensing compression ratio is 15. DETAILED DESCRIPTION

[0046] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0047] The present invention provides a method for adaptive symbol detection of spread spectrum signals, which can measure spread spectrum signals at a lower sampling frequency and detect signal symbols without reconstructing the original signal. In combination with the hardware cost of system design, the a posteriori information of signal measurement is used to efficiently and adaptively design a compressed sampling matrix, thereby further ensuring the accuracy of symbol detection.

[0048] The method for adaptive symbol detection of spread spectrum signals provided in the embodiment of the present invention constructs or updates a dictionary for sparse expression of signals according to the estimated signal symbol period and the selection range of other parameters (such as the carrier frequency range of the frequency hopping signal, the pseudo-random sequence of the direct sequence spread spectrum signal, the transmission time of the time hopping signal, etc.) for the type of spread spectrum signal to be processed (frequency hopping, direct sequence spread spectrum or time hopping). Within one symbol period, the expression sparsity of the signal to be processed on the dictionary is 1.

[0049] According to the system cost, a signal sampling system for adaptive signal compression sensing is constructed, wherein the number of samples of the signal each time is 1. The adaptive compression sensing design module uses the a posteriori information of the acquired signal sampling data, based on the constructed sparse expression dictionary and linear discriminative analysis, to design the measurement matrix for future signal compression sampling independently or in turn;

[0050] When the compressed sensing measurement of a code element period is completed, the compressed matching filter is designed based on the sparse expression dictionary, the measurement matrix and measurement results in adaptive compressed sensing, and the maximum a posteriori probability criterion to determine the code elements carried by the signal.

[0051] like Figure 1 As shown, the method for adaptive symbol detection of spread spectrum signals provided by the embodiment of the present invention comprises the following steps:

[0052] S1. Acquire a spread spectrum signal to be detected and perform signal parameter estimation; among the spread spectrum signals to be detected, the spread spectrum signal types include: frequency hopping, direct sequence spread spectrum or time hopping; the signal parameter estimation includes at least: signal symbol period, carrier frequency range of frequency hopping signal, pseudo-random sequence of direct sequence spread spectrum signal, transmission time of time hopping signal.

[0053] S2. Generate or update a signal sparse expression dictionary within a single symbol period according to the estimated signal parameters;

[0054] The signal sparse expression dictionary can be expressed by the following formula (1):

[0055] Ψ=[ψ 0,1 ,ψ 0,2 ,…,ψ 0,K ,ψ 1,1 ,ψ 1,2 ,…,ψ 1,K ,…,ψ R-1,1 ,ψ R-1,2 ,…,ψ R-1,K ] (1)

[0057] Among them, each sub-block ψ r,k (0 ≤ r < R, 1 ≤ k ≤ K) represents a column, that is, a possible expression of the signal in one symbol period; R represents the number of symbol types. For example, in BFSK modulation, R = 2, and in QPSK modulation, R = 4; K represents the number of possible expressions of the signal under each symbol type.

[0058] In order to keep the sparse representation dictionary continuously effective, the sparse representation dictionary in the present invention can be updated regularly according to the estimated range of real-time updated signal parameters.

[0059] S3. Use the signal sparse representation dictionary to construct an adaptive compressive sensing measurement kernel;

[0060] When constructing the adaptive compressive sensing measurement kernel, first calculate the scatter matrix using the following formula (2):

[0061]

[0062] Among them, l represents the l-th compressive sensing measurement kernel, is the posterior usage probability of the dictionary element ψ r,k obtained by analyzing the measured compressive sensing data and the measurement matrix; ψ r,k,l is the dictionary element segment composed of the ((l - 1)CR + 1)-th element to the l×CR-th element in ψ r,k ; Ψ l is the segment composed of the ((l - 1)CR + 1)-th row to the l×CR-th row of the dictionary Ψ; diag() represents the operation of converting a column vector into a diagonal matrix; () H represents the Hermitian operation of the matrix; p is an intermediate vector.

[0063]

[0064] Similar to the update of the posterior probability of the dictionary element in symbol detection, the system calculates the posterior probability in a similar way to the posterior usage probability of each dictionary element. Among them, in order to improve the operation speed, for different dictionary elements the system can adopt parallel computing.

[0065] For the l-th compressive measurement kernel, that is, the non-zero elements of the l-th row of the compressive sensing matrix, take the normalized most principal eigenvector of the matrix in formula (3).

[0066] The present invention uses the signal posterior probability obtained from the compressed sensing data, and through the linear differentiation analysis of the sparse expression dictionary fragment, adaptively designs the compressed measurement kernel for adaptive compressed sensing in the future, and then only adapts from the compressed sensing matrix; in order to ensure that the design process of each compressed sensing measurement kernel can be allocated enough time, the present invention adopts a multi-design unit to cooperate with each other to design the compressed measurement kernel. In this case, the shorter the design time of a single compressed measurement kernel, the stronger the computing processing capability of the adaptive design unit, and the higher the cost of software and hardware implementation.

[0067] The number of adaptive design units can be obtained as follows (3):

[0068]

[0069] Among them, T adp is the time consumed by the measurement core designed for a compressed sampling; T samp It is the time interval between two compressed samplings of the system.

[0070] S4, after denoising and filtering the spread spectrum signal to be detected, compress it using the adaptive compressed sensing measurement kernel in S3 to obtain compressed sampling data; after each compressed sampling, using the signal sparse expression dictionary, the compressed sensing measurement matrix and the obtained compressed sampling data, based on the linear discrimination analysis system, design a compressed measurement kernel for future signal compression sampling;

[0071] In the present invention, the compressed sensing of the signal within one symbol period can be expressed by the following formula:

[0072] y=Φ(s+n)=Φ(Ψ+n)=ΦΨ+n meas

[0073] Where y is the M×1 compressed sensing data; s=Ψ is the N×1 noiseless spread spectrum signal, which is a column vector with sparsity 1; n and n meas are the signal noise and the equivalent noise in the measured data, respectively. Figure 1 For the system described, the compressed sensing measurement matrix in the present invention is as follows Figure 2 The M×N block diagonal matrix structure shown in FIG. 5 is a block structure in which each sub-block is a 1×CR row vector, ie, the compressed measurement kernel of each compressed sampling, and CR is the compression ratio of compressed sensing.

[0074] S5. Perform codeword detection based on the maximum a posteriori probability according to the compressed sampling data, the signal sparse expression dictionary and the compressed sensing measurement matrix to obtain a codeword detection result.

[0075] When the system completes the sample collection of a symbol period, the signal symbol detection module starts to detect the symbol. Figure 3 The present invention firstly calculates the posterior usage probability of each dictionary element by using the Bayesian formula, the compressed sensing matrix and the compressed measurement data, as shown in formula (4):

[0076]

[0077] Where y and Φ are all compressed sampling data and compressed measurement matrix in one symbol period respectively; P(ψ r,k ) is the dictionary element ψ r,k The prior probability used; p(y|Φ,ψ r,k ) is a given compressed sensing matrix and dictionary element ψ r,k When , the probability density of the compressed sampling data y is Figure 3 As shown, in order to improve the operation speed, p(y|Φ,ψ r,k ), the system can adopt parallel computing.

[0078] The present invention determines the dictionary element to be used by the maximum a posteriori probability criterion. Finally, the code element carried by the signal in the code element period is determined to be the code element carried by the dictionary element. This process can be expressed by formula (4):

[0079] sym=arg r maxP(ψ r,k |y,Φ) (5)

[0080] In formula (5), y and Φ are all compressed sampling data and compressed measurement matrix in one code element period, respectively. Similar to the update of the posterior probability of dictionary elements in the adaptive compressed sensing process, in order to improve the operation speed, the system can use parallel computing for the calculation of the posterior probability of different dictionary elements.

[0081] When multiple adaptive design units are used to perform adaptive design of the measurement core, each unit cooperates with each other to complete the design of the overall measurement core. For example, when three adaptive design units are used, the compressed sensing system can be composed of Figure 4 It indicates that only one design unit is connected to the multiplier at a time. There is a synchronization module at the junction of the three adaptive design units and the multiplier, thereby ensuring that the connection between the multiplier and the adaptive unit is effectively rotated after each compressed sampling.

[0082] use Figure 4 When the system is designed, the workflow of each adaptive design unit is shown in Table 1.

[0083] Table 1 Schematic diagram of the workflow of each design unit of the three-unit adaptive compression sampling system

[0084]

[0085]

[0086] It should be noted that if the design time of the single measurement kernel is short enough, even equal to or less than the Nyquist sampling interval, then the measurement kernel for the m-th (1 < m ≤ M) compressive sampling can be directly designed based on the sampling data from the 1st to the (m - 1)-th within the symbol period.

[0087] As an example of the use of the present invention, Figure 5 、 Figure 6 、 Figure 7 respectively show the variation of the symbol detection error symbol rate of the frequency-hopping signal with the signal-to-noise ratio when the compressive sensing compression ratios are 5, 10, and 15 using the present invention. Among them, the definition of the compression ratio of bandpass Nyquist compressive sensing is as shown in formula (6):

[0088]

[0089] where B refers to the overall bandwidth of frequency hopping; T s is the symbol period of the frequency-hopping signal; M is the number of compressive samplings within one symbol period. Here, the signal-to-noise ratio is defined as the ratio of the received power of the frequency-hopping signal to the noise power.

[0090] The generation of the frequency-hopping signal adopts the standard of binary Gaussian filtered frequency shift keying (BGFSK) Bluetooth signal. The main key parameters include: the frequency-hopping carrier frequency is taken from 79 frequencies uniformly distributed from 2.402 GHz to 2.480 GHz with equal probability, the modulation frequency shift of the signal is 160 KHz, the symbol period is 1 μs, and one frequency-hopping period includes 625 symbol periods. In Figure 5 、 Figure 6 、 Figure 7 ,"Traditional non-compressive" refers to the result of adopting traditional bandpass Nyquist sampling; "Random measurement kernel" refers to the result when using the compressive sampling system shown in Figure 1 , but the measurement kernel is generated randomly; "Ideal adaptive" refers to the result when using the adaptive compressive sampling system shown in Figure 1 , and it is assumed that the measurement kernel design unit can design the compressive measurement kernel within the Nyquist sampling interval after obtaining the compressive sampling result. At this time, the measurement kernel for the m-th (1 < m ≤ M) compressive sampling can be directly designed and obtained based on the sampling data from the 1st to the (m - 1)-th within the symbol period; the remaining "x design unit" refers to the result when using the corresponding number of adaptive design units when the time required for the design unit to design a single measurement kernel is equal to or slightly less than the corresponding multiple of the compressive sampling time interval.

[0091] From Figure 5 、 Figure 6 、 Figure 7 the following characteristics can be seen:

[0092] (1) The symbol error rate of the system decreases with the increase of signal-to-noise ratio, and increases with the increase of compressed sensing compression ratio;

[0093] (2) When the system of the present invention is used to perform symbol detection on a signal, when the signal-to-noise ratio and the compression ratio of compressed sensing are constant, the symbol error rate obtained is greater than the symbol error rate obtained by the traditional non-compression method and less than the symbol error rate obtained by using a random measurement kernel;

[0094] (3) For a certain compressed sensing compression ratio, the more time it takes to design a single compressed measurement core, the more adaptive design units are needed, and the higher the bit error rate will be at a certain signal-to-noise ratio.

[0095] When the compressed sensing compression is equal to 5, if the design time of a single compressed measurement kernel is small enough and can be less than or equal to the Nyquist sampling interval of the signal, then the obtained symbol error rate can be close to the symbol error rate corresponding to traditional non-compressed sampling.

[0096] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0097] Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. Those skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and purpose of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for adaptive symbol detection of a spread spectrum signal, characterized in that: The following steps are involved: S1. Obtain the spread spectrum signal to be detected and estimate the signal parameters; S2. Generate or update a signal sparse expression dictionary within a single symbol period according to the estimated signal parameters; S3, constructing an adaptive compressed sensing measurement kernel using the signal sparse expression dictionary; S4, after denoising and filtering the spread spectrum signal to be detected, compress it using the adaptive compressed sensing measurement kernel in S3 to obtain compressed sampling data; After each compression sampling, the signal sparse expression dictionary, the compressed sensing measurement matrix and the obtained compressed sampling data are used to design the compressed measurement kernel for future signal compression sampling based on the linear discrimination analysis system; S5. Perform codeword detection based on the maximum a posteriori probability according to the compressed sampling data, the signal sparse expression dictionary and the compressed sensing measurement matrix to obtain a codeword detection result.

2. The method for adaptive symbol detection of spread spectrum signals according to claim 1, characterized in that: In S1, among the spread spectrum signals to be detected, the spread spectrum signal types include: frequency hopping, direct sequence spread spectrum or time hopping; the signal parameter estimation includes at least: signal code element period, carrier frequency range of frequency hopping signal, pseudo-random sequence of direct sequence spread spectrum signal, transmission time of time hopping signal.

3. The method for adaptive symbol detection of spread spectrum signals according to claim 1, characterized in that: In S2, the signal sparse expression dictionary can be expressed by the following formula (1): Ψ=[ψ 0,1 ,ψ 0,2 ,…,ψ 0,K ,ψ 1,1 ,ψ 1,2 ,…,ψ 1,K ,…,ψ R-1,1 ,ψ R-1,2 ,…,ψ R-1,K ] (1) Among them, each sub-block ψ r,k (0 ≤ r < R, 1 ≤ k ≤ K) represents a column, that is, a possible expression of the signal in one symbol period; R represents the number of symbol types.

4. The method for adaptive symbol detection of spread spectrum signals according to claim 1, characterized in that: In S3, when constructing the adaptive compressed sensing measurement kernel, the scatter matrix is ​​first calculated using the following formula (2): Where l represents the lth compressed sensing measurement kernel, The dictionary element ψ is obtained by analyzing the measured compressed sensing data and the measurement matrix r,k The posterior probability of use; ψ r,k,l is r,k The dictionary element fragment consisting of the (l-1)CR+1th element to the l×CRth element in, Ψ l is a fragment consisting of the (l-1)CR+1th row to the l×CRth row of the dictionary Ψ; diag() represents the operation of converting a column vector into a diagonal matrix; () H represents the Hermitian operation of the matrix; p is the intermediate vector.

5. The method for adaptive symbol detection of spread spectrum signals according to claim 3, characterized in that: It also includes designing a compressed measurement core by using multiple design units to cooperate with each other, wherein the number of adaptive design units can be obtained as follows (3): Among them, T adp is the time consumed by the measurement core designed for a compressed sampling; T samp It is the time interval between two compressed samplings of the system.

6. The method for adaptive symbol detection of spread spectrum signals according to claim 1, characterized in that: In S4, the spread spectrum signal first passes through a pre-filter to filter out interference and noise outside the spread spectrum band; the time domain signal obtained by time-domain multiplication of the compressed measurement kernel and the pre-filtered signal passes through an analog-to-digital converter composed of a low-pass filter and a downsampler to obtain the compressed sensing data of the signal; among them, the low-pass filter acts as an integrator, and the downsampler operates at a frequency lower than the Nyquist sampling rate.

7. The method for adaptive symbol detection of spread spectrum signals according to claim 1, characterized in that: In S5, the process of codeword detection includes: The posterior usage probability of each dictionary element is calculated by using the Bayesian formula, the compressed sensing matrix and the compressed measurement data; The dictionary elements to be used are determined by the maximum a posteriori probability criterion; According to the codeword carried by the signal in the codeword period, the codeword carried by the dictionary element is determined.

8. The method for adaptive symbol detection of spread spectrum signals according to claim 1, characterized in that: The posterior usage probability of each dictionary element is calculated using the compressed sensing matrix and compressed measurement data, as shown in formula (4): Where y and Φ are all compressed sampling data and compressed measurement matrix in one symbol period respectively; P(ψ r,k ) is the dictionary element ψ r,k The prior probability used; p(y|Φ,ψ r,k ) is a given compressed sensing matrix and dictionary element ψ r,k , the probability density of the compressed sampling data y.

9. The method for adaptive symbol detection of spread spectrum signals according to claim 1, characterized in that: The dictionary elements used are determined by the maximum a posteriori probability criterion, which is expressed by formula (5): sym=arg r maxP(ψ r,k |y,Φ) (5) Among them, y and Φ are all compressed sampling data and compressed measurement matrix in one code element period respectively.

Images