Multi-beam signal fast frame synchronization method in underwater acoustic communication

Abstract

The invention discloses a multi-beam signal fast frame synchronization method in underwater acoustic communication, which is based on a GPU and CPU heterogeneous computing platform, and comprises the following steps: carrying out analog-to-digital conversion on a sound wave signal received by a receiving array, then carrying out full-angle beam forming, and outputting M paths of beam signals; performing thread scheduling and data transmission at the CPU end, and transmitting M paths of beam signals to the GPU end; performing FFT (Fast Fourier Transform) parallel operation on the M paths of beam signals in a GPU (Graphics Processing Unit), performing complex dot product operation on the M paths of beam signals, a local up-modulation signal and a local down-modulation signal subjected to time reversal, performing IFFT (Inverse Fast Fourier Transform) parallel operation and module value normalization parallel operation, and outputting a matched filtering result; and carrying out reduction peak value search and judgment in the GPU according to a matched filtering result, if the judgment result is that a communication signal is captured, recording the peak value size and the peak value appearance position, and obtaining a signal initial position and an optimal direction of arrival and transmitting the signal initial position and the optimal direction of arrival back to the CPU end.

Description

Technical Field

[0001] The present invention relates to the intersection of underwater acoustic communication, signal processing and GPU scientific computing, and in particular to a multi-beam signal fast frame synchronization method in underwater acoustic communication. Background Art

[0002] Underwater wireless communication is a key enabling technology for underwater sensor networks and networking. In scenarios where long-distance communication is required, traditional communication carriers such as lasers and electromagnetic waves suffer from significant energy attenuation in water, making them difficult to meet the requirements for reliable long-distance transmission. In contrast, acoustic waves, with their low attenuation in water, have become the only viable and reliable transmission medium for underwater long-distance communication and have been widely used.

[0003] In underwater communication scenarios, MIMO systems typically work in tandem through the transmitting array at the transmitter and the receiving array at the receiver. This not only significantly improves the transmission rate, but also allows the receiving array to estimate the direction of arrival (DoA) of the received signal, thereby achieving accurate perception of the source direction and meeting the needs of complex tasks such as target detection.

[0004] In underwater networking communications, frame synchronization is the primary step in signal processing at the receiving end. It is used to determine the starting position of the signal and provide a basis for data demodulation. As a key link in digital communications, frame synchronization must have the characteristics of simple structure, accurate positioning, and fast synchronization. With the continuous advancement of underwater acoustic communication MIMO technology and "detection-communication integration" technology on unmanned platforms at sea, underwater communication equipment often needs to receive beam data from multiple directions simultaneously during operation. This requirement significantly increases the computational complexity and time overhead of frame synchronization processing, and thus puts higher demands on the real-time processing capabilities of the system. Therefore, the study of efficient frame synchronization algorithms for multi-beam reception scenarios has become an important issue that needs to be solved urgently.

[0005] A graphics processing unit (GPU) is a highly efficient processor designed specifically for graphics rendering and large-scale parallel computing. Compared to a central processing unit (CPU), which emphasizes low latency, a GPU focuses more on high-throughput architectural design, making it particularly effective at processing parallel tasks. Currently, developers often rely on NVIDIA's parallel computing platform and programming model, the Compute Unified Device Architecture (CUDA), for efficient programming and accelerated computing. Technologies based on GPUs and the CUDA architecture are now widely used in the communications field, with typical scenarios including compute-intensive tasks such as signal processing, modulation and demodulation, and channel coding and decoding.

[0006] In summary, the difficulty of processing multiple beams in the receiving terminal of the underwater acoustic communication system lies in how to reduce the computational complexity and the time overhead required for frame synchronization to meet the real-time requirements of the communication system. Summary of the Invention

[0007] The purpose of the present invention is to overcome the defects of the prior art and propose a fast frame synchronization method for multi-beam signals in underwater acoustic communication, which is suitable for fast frame synchronization of underwater acoustic communication receiving terminals.

[0008] In view of this, the present invention proposes a method for fast frame synchronization of multi-beam signals in underwater acoustic communication, based on a GPU and CPU heterogeneous computing platform, the method comprising:

[0009] Step 1: After analog-to-digital conversion, the acoustic wave signal received by the receiving array is subjected to full-angle beamforming and M-channel beam signals are output;

[0010] Step 2: Perform thread scheduling and data transfer on the CPU side, and transmit M-way beam signals to the GPU side;

[0011] Step 3: Perform parallel FFT operations on the M-way beam signal in the GPU, perform complex point multiplication with the local up-modulated signal and the time-reversed local down-modulated signal, then perform parallel IFFT operations and modulus normalization operations, and output the matched filter result.

[0012] Step 4: Perform a reduced peak search and determination based on the matched filtering results in the GPU. If a communication signal is captured, the peak size and peak position are recorded, and the signal starting position and optimal wave direction are obtained and transmitted back to the CPU.

[0013] Preferably, the method further comprises: before step 1, designing a waveform of an underwater acoustic communication frame synchronization signal, wherein the front end is a linear up-frequency modulation signal of a frame synchronization header, and there is a guard interval between the frame synchronization header and the data block;

[0014] The complex envelope s(t) of the linear up-frequency modulation signal is:

[0015]

[0016] Where A is the signal amplitude, β=BT is the bandwidth product, B and T are the bandwidth and pulse width respectively, t represents the time domain, and j represents the imaginary part.

[0017] Preferably, in step 1, the acoustic wave signal r(t) received by a certain array element of the receiving array through the underwater acoustic channel is:

[0018]

[0019] Where k represents the kth path, K is the number of multipaths, τ k is the corresponding multipath delay, f d,k is the corresponding Doppler shift, α k is the complex gain, n(t) is the noise, and t represents the time domain.

[0020] Preferably, the M-way beam signal y outputted in step 1 is:

[0021]

[0022] in, is the signal vector received by the array element at time t, is the steering vector, θ M is the Mth incident angle, N is the number of elements in the receiving array, and the superscript H represents the conjugate transpose.

[0023] Preferably, the step 2 comprises:

[0024] The generated M-way beam signal stream is stored on the CPU side. When the cache is full, the beam data is transferred to the GPU side through the cudaMemcpy instruction, and the kernel function is started after the grid and thread block size are defined.

[0025] Preferably, before step 3, the method further includes: applying for and initializing memory on the GPU side, and generating an FFT handle and an IFFT handle.

[0026] Preferably, the step 3 comprises:

[0027] Perform signal transposition on the M-way beam signals, realize parallel-to-serial conversion, and convert them into complex values;

[0028] Use CUDA stream to asynchronously perform complex floating-point FFT parallel operations, and then perform complex point multiplication operations with the local up-frequency modulation signal and the local down-frequency modulation signal after time reversal. Use CUDA stream to asynchronously perform IFFT parallel operations, and then perform modulus normalization parallel operations. The output of the m-th matched filter result includes two groups, the z-th group of matched filter results y z (m) for:

[0029] y z (m) =F H Λ z Fx z (m) ,m=1,2,...,M

[0030] Where z = 1, 2, x1 is the local up-modulated signal, x2 is the time-reversed local down-modulated signal, the superscript (m) indicates the corresponding m-th beam signal, F represents the discrete Fourier transform matrix, Λ1 and Λ2 represent the diagonal matrices formed by discrete Fourier transform of the local up-modulated signal and the time-reversed local down-modulated signal, respectively, and the superscript H represents the conjugate transpose.

[0031] Preferably, step 4 includes:

[0032] On the GPU side, a reduced peak search is performed based on the matched filtering results of the M-beams, and the peak values ​​W1 and W2 of the two sets of matched filters of the M-beams and the positions where the peaks appear are searched;

[0033] Calculate the correlation value energy ratio Q = 10·log(W1 / W2) and compare it with the preset threshold Threshold. If Threshold>Q, synchronization is determined to be successful, complete the detection, and record the size and location of peak W1. Divide the location of peak W1 by the length of a single beam signal. The quotient corresponds to the beam sequence information where peak W1 is located, and the remainder corresponds to the location of peak W1 in this beam. This method obtains the optimal arrival direction and signal starting position of the arrival direction corresponding to each of the M beams, and transmits the result to the CPU. Otherwise, the signal is not synchronized, and detection continues.

[0034] Compared with the prior art, the advantages of the present invention are:

[0035] This paper, based on the engineering application perspective of practical underwater acoustic communications, addresses the high real-time requirements for communication signal frame synchronization calculations in receivers of underwater unmanned platform communication systems and proposes a multi-beam parallel frame synchronization method based on a heterogeneous GPU and CPU architecture. This method fully leverages the advantages of GPUs in large-scale parallel computing and performs parallel design and optimization for key computational modules in frequency domain frame synchronization, such as FFT transforms, matched filtering, and correlation peak search. This significantly improves overall processing efficiency and proposes a specific implementation scheme and thread scheduling strategy. This significantly enhances the real-time processing capabilities of underwater acoustic communication receivers and provides key technical support for the research and engineering application of efficient multi-beam signal processing algorithms in future underwater unmanned platform communication systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a communication signal frame structure diagram used in the present invention;

[0037] Figure 2 This is a schematic diagram of the implementation process of the fast frame synchronization method based on a heterogeneous computing platform proposed in the present invention;

[0038] Figure 3 This is a schematic diagram of the GPU-based multi-beam frame synchronization parallel algorithm model proposed in the present invention;

[0039] Figure 4 Schematic diagram of GPU-based peak search proposed by the present invention;

[0040] Figure 5 This is a time comparison chart of the GPU-based peak search algorithm proposed by the present invention and the conventional peak search algorithm;

[0041] Figure 6 This is a schematic diagram of the structure of the multi-beam frame synchronization algorithm based on a heterogeneous computing platform proposed in the present invention;

[0042] Figure 7 The time cost and speedup ratio of the fast frame synchronization method proposed in the present invention in the test;

[0043] Figure 8 It is a time-varying channel impulse response diagram in actual application scenarios;

[0044] Figure 9 It is a comparison between the synchronization position output by the fast frame synchronization method proposed in the present invention and the actual signal starting position in the actual application scenario;

[0045] Figure 10 This is a comparison between the direction of arrival output by the fast frame synchronization method proposed in the present invention and the actual signal direction of arrival in actual application scenarios. DETAILED DESCRIPTION

[0046] The present invention designs a fast frame synchronization method for an underwater acoustic communication receiving terminal based on a heterogeneous computing platform of GPU+CPU. The implementation steps of the method include:

[0047] 1. Underwater acoustic communication frame synchronization signal waveform design, such as Figure 1 As shown in FIG, the front end of the communication signal is the LFM signal of the frame synchronization header, and a protection interval is set up between the frame synchronization header and the data block;

[0048] 2. Initialize the GPU computing platform, perform initialization and memory allocation on the GPU side, and generate FFT handles;

[0049] 3. Multi-beam generation of the underwater communication receiving terminal, which generates beam data through conventional beamforming operations on the received array data;

[0050] 4. Build a heterogeneous computing platform, perform thread scheduling and data transmission on the CPU side, and pass the received data to the GPU side. The GPU side uses the CUDA architecture and CUFFT library to implement parallel computing;

[0051] 5. Frequency domain matched filtering of the received signal and the local replica signal (including the up- and down-modulated signals), performing FFT operations and complex multiplication calculations on the GPU. After the calculations are completed, IFFT calculations and modulus normalization operations are performed on the GPU. Specifically:

[0052] Perform signal transposition on the M-way beam signals, realize parallel-to-serial conversion, and convert them into complex values;

[0053] The CUDA stream is used to asynchronously perform complex floating-point FFT parallel operations, and then complex point multiplication operations are performed with the local up-frequency modulated signal and the time-reversed local down-frequency modulated signal respectively. The CUDA stream is used to asynchronously perform IFFT parallel operations, and then the modulus normalization parallel operations are performed. The output of the m-th matched filter result includes two groups.

[0054] 6. Peak search: Based on the two sets of matched filter results obtained in step 6, a reduced peak search is performed to obtain the peak values ​​W1 and W2 of the two sets of matched filters and the positions of the peak values. The peak values ​​are then compared with the preset threshold to determine whether the signal has been captured. Specifically:

[0055] For each beam, calculate the correlation value energy ratio Q = 10·log(W1 / W2) and compare it with the preset threshold Threshold. If the threshold Threshold>Q, the synchronization is determined to be successful, the detection is completed and the peak W1 size and the position of the peak W1 are recorded. The position of the peak W1 is divided by the length of the single beam signal. The quotient corresponds to the beam sequence information where the peak W1 is located, and the remainder corresponds to the position of the peak W1 in this beam. The best arrival direction and signal starting position in the arrival direction corresponding to each of the M beams are obtained, and the result is passed to the CPU. Otherwise, the signal is not synchronized and the detection continues.

[0056] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0057] Example

[0058] An embodiment of the present invention provides a method for fast frame synchronization of multi-beam signals in underwater acoustic communication.

[0059] 1. The model used in this embodiment is as follows Figure 2 As shown in the figure, after the acoustic wave signal reaches the receiving array, it first undergoes analog-to-digital conversion, followed by full-angle beamforming, and outputs M-path beam signals. The signal is preprocessed by the CPU and transmitted to the GPU. After parallel calculation is completed in the GPU, the matched filter correlation value is output and transmitted back to the CPU. The CPU performs energy detection and calculates the optimal beam angle and signal starting position.

[0060] 2. After starting the program of the present invention, GPU memory is requested and initialized through the cudaMalloc and cudaMemset instructions, and an FFT handle is generated through the cufftHandle instruction.

[0061] 3. For the multi-beam data processing scenario applicable to the present invention, the process of generating a multi-beam signal from a received signal is as follows:

[0062] The complex envelope s(t) of the LFM signal transmitted by the transmitter is:

[0063]

[0064] Where β = BT is the bandwidth product, B and T are the bandwidth and pulse width respectively, then the signal r(t) received by a certain array element at the receiving end after passing through the channel is:

[0065]

[0066] Among them, k represents the kth path, K is the number of multipaths, τ k is the corresponding multipath delay, f d,k is the corresponding Doppler shift, α kis the complex gain, n(t) is the noise, and t represents the time domain.

[0067] After receiving the far-field signal with an incident angle of θ, each array element forms a beam and outputs multiple signal beams in different directions. Assuming that the linear array consists of N evenly distributed array elements, the array element spacing is d, and the signal wavelength is, then the steering vector Defined as:

[0068]

[0069] Then the signal vector received by the array element at time t is r(t)=a(θ)s(t)+n(t), where s is the incident signal and n is the Gaussian white noise vector. For the preset mth beam direction, we have:

[0070]

[0071] in is the conjugate transpose of the weight vector, y i For the direction θ i The beam output result of , then the output result vector for M beam directions is:

[0072]

[0073] in, is the signal vector received by the array element at time t, is the steering vector, θ M is the Mth incident angle, N is the number of elements in the receiving array, and the superscript H represents the conjugate transpose.

[0074] 3. The model of the multi-beam frame synchronization fast algorithm proposed in this invention is as follows: Figure 3As shown, the CPU stores the generated beam data stream. When the cache is full, the CPU passes the beam data to the GPU through the cudaMemcpy instruction, and starts the kernel function after defining the grid and thread block size. First, the signal transposition operation is implemented on the GPU side so that the M-way beam data is arranged in the order required by the FFT matrix. Then, based on the CUFFT (CUDA Fast Fourier Transform Library) library, FFT / IFFT can be efficiently performed on the GPU. The M beam signals are regarded as M short-time signals, and the FFT operation is performed on each signal segment at the same time. Each round of calculation only requires one FFT / IFFT operation. One-dimensional FFT is performed row by row, and a handle is generated through cufftPlan1d, and batch is set to M. After the FFT operation is completed, a complex point multiplication operation is performed on the GPU. After completion, a one-dimensional IFFT is performed row by row, and a handle is generated through cufftPlan1d, and batch is set to M. After the processed data is converted to the time domain, the modulus normalization operation of the matched filter is performed on the GPU. The principle of the above process is:

[0075] Signal in the frequency domain:

[0076]

[0077] The spectrum of the received signal and the spectrum of the local signal are conjugated and multiplied point by point, and then N FFT The inverse Fourier transform of the point gives the correlation value:

[0078]

[0079] When M groups of signals need to be frequency-domain matched filtered in parallel, the input signal matrix is The up-conversion frequency is x. The generalized matrix expression of its frequency domain matched filtering is as follows:

[0080]

[0081] In the formula Indicates N FFT Point DFT operation, DFT matrix is defined as:

[0082]

[0083] Its inverse transformation matrix is ​​F -1 =F H , that is, Hermitian transpose, combine the above steps and further expand the matrix multiplication, and the m-th column of the output matrix, that is, the matched filtering result of the m-th beam is:

[0084] y (m) =F H ΛFx(m) ,m=1,2,...,M

[0085] To reduce the impact of environmental noise, during each matched filtering operation, the received signal will also undergo the same matched filtering process as the time-reversed local down-modulated signal, thus obtaining two sets of matched filtering results:

[0086] y z (m) =F H Λ z Fx z (m) ,m=1,2,...,M

[0087] Where z = 1, 2, x1 is the local up-modulated signal, x2 is the time-reversed local down-modulated signal, the superscript (m) indicates the corresponding m-th beam signal, F represents the discrete Fourier transform matrix, Λ1 and Λ2 represent the diagonal matrices formed by discrete Fourier transform of the local up-modulated signal and the time-reversed local down-modulated signal, respectively, and the superscript H represents the conjugate transpose.

[0088] 4. After the frequency domain matched filtering operation is completed on the GPU side, the starting position of the signal needs to be determined according to the position of the correlation peak. The serial search method is generally a sequential scanning method, which traverses the data set point by point and compares the threshold to detect the peak position. In order to further save the computing time of the peak search, the present invention will calculate the peak size and index at the same time on the GPU side, such as Figure 4 As shown, the total time complexity is O(log2MN). Specifically: On the GPU side, a reduced peak search is performed based on the matched filtering results of all M beams. The peak values ​​W1 and W2 of the two sets of matched filters for all M beams and the locations where the peak values ​​appear are searched. The correlation value energy ratio Q = 10·log(W1 / W2) is calculated and compared with the preset threshold Threshold. If the threshold Threshold>Q, synchronization is determined to be successful. The detection is completed and the size and location of peak W1 are recorded. The location of peak W1 is divided by the length of a single beam signal. The quotient corresponds to the beam sequence information where peak W1 is located, and the remainder corresponds to the location of peak W1 in this beam. The optimal arrival direction and signal starting position of each of the M beams are obtained, and the result is passed to the CPU side. Otherwise, the signal is not synchronized and detection continues.

[0089] In order to verify the performance improvement of the proposed algorithm in the peak search task, the computation time required by three different algorithms is compared on the simulation platform to verify the performance advantage of the proposed algorithm. Figure 5The figure shows a comparison of the time required for peak search among the three algorithms under different computational tasks. The figure shows the speed advantage of the GPU-based reduced peak search method proposed in this invention.

[0090] 5. The frame synchronization algorithm proposed in this invention is based on a heterogeneous computing platform of CPU and GPU. Based on the aforementioned receiver frame synchronization operation process, combined with the multi-threaded requirements for data reception, processing and result transmission during the frame synchronization process, a dynamically loaded multi-threaded library is used to implement multi-threaded parallel operation of the program. The program flow framework of the frame synchronization parallel algorithm is as follows: Figure 6 As shown, the dotted lines represent the data and instruction transmission between different threads. Considering that the main goal of the present invention is to improve the operation speed during the frame synchronization process, in order to verify the acceleration performance of the proposed algorithm, the processing time required for the GPU acceleration algorithm and the CPU serial algorithm to complete a frame synchronization is recorded respectively, and the time variation curve of each processing step in the GPU algorithm is plotted. The processing time and acceleration ratio results are shown in Figure 2. Figure 7 As shown in the figure, it can be seen that the present invention has an advantage in computing speed, and meets the frame synchronization real-time requirements of the receiving terminal of the communication system.

[0091] In practical application scenarios where measured data is generated, the time-varying channel impulse response is as follows: Figure 8 As shown;

[0092] In order to verify the accuracy of the present invention, in the actual data verification, the performance of the method of the present invention in the output frame synchronization position is as follows Figure 9 As shown, the performance of the method of the present invention in the output wave direction is as follows Figure 10 As shown in the figure, it can be seen that the accuracy and reliability of the present invention in performing the multi-beam frame synchronization task are verified, and the results verify the effectiveness of the present invention.

[0093] Finally, it should be noted that the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention and are intended to be encompassed by the claims of the present invention.

Claims

1. A method for fast frame synchronization of multi-beam signals in underwater acoustic communication, based on a GPU and CPU heterogeneous computing platform, comprising: Step 1: After analog-to-digital conversion, the acoustic wave signal received by the receiving array is subjected to full-angle beamforming and M-channel beam signals are output; Step 2: Perform thread scheduling and data transfer on the CPU side, and transmit M-way beam signals to the GPU side; Step 3: Perform parallel FFT operations on the M-way beam signal in the GPU, perform complex point multiplication with the local up-modulated signal and the time-reversed local down-modulated signal, then perform parallel IFFT operations and modulus normalization operations, and output the matched filter result. Step 4: Perform a reduced peak search and determination based on the matched filtering results in the GPU. If a communication signal is captured, the peak size and peak position are recorded, and the signal starting position and optimal wave direction are obtained and transmitted back to the CPU.

2. The method for rapid frame synchronization of multi-beam signals in underwater acoustic communication according to claim 1, characterized in that: The method further comprises: before step 1, designing a waveform of an underwater acoustic communication frame synchronization signal, wherein the front end is a linear up-frequency modulation signal of a frame synchronization header, and a guard interval is provided between the frame synchronization header and the data block; The complex envelope s(t) of the linear up-frequency modulation signal is: Where A is the signal amplitude, β=BT is the bandwidth product, B and T are the bandwidth and pulse width respectively, t represents the time domain, and j represents the imaginary part.

3. The method for rapid frame synchronization of multi-beam signals in underwater acoustic communication according to claim 2, characterized in that: In step 1, the acoustic wave signal r(t) received by a certain element of the receiving array through the underwater acoustic channel is: Among them, k represents the kth path, K is the number of multipaths, τ k is the corresponding multipath delay, f d,k is the corresponding Doppler shift, α k is the complex gain, n(t) is the noise, and t represents the time domain.

4. The method for rapid frame synchronization of multi-beam signals in underwater acoustic communication according to claim 1, characterized in that: The M-way beam signal y outputted in step 1 is: in, is the signal vector received by the array element at time t, is the steering vector, θ M is the Mth incident angle, N is the number of elements in the receiving array, and the superscript H represents the conjugate transpose.

5. The method for rapid frame synchronization of multi-beam signals in underwater acoustic communication according to claim 1, characterized in that: The step 2 includes: The generated M-way beam signal stream is stored on the CPU side. When the cache is full, the beam data is transferred to the GPU side through the cudaMemcpy instruction, and the kernel function is started after the grid and thread block size are defined.

6. The method for rapid frame synchronization of multi-beam signals in underwater acoustic communication according to claim 1, characterized in that: The above-mentioned step 3 also includes: applying for and initializing memory on the GPU side, and generating FFT handles and IFFT handles.

7. The method for rapid frame synchronization of multi-beam signals in underwater acoustic communication according to claim 1, characterized in that: The step 3 comprises: Perform signal transposition on the M-way beam signals, realize parallel-to-serial conversion, and convert them into complex values; Use CUDA stream to asynchronously perform complex floating-point FFT parallel operations, and then perform complex point multiplication operations with the local up-frequency modulation signal and the local down-frequency modulation signal after time reversal. Use CUDA stream to asynchronously perform IFFT parallel operations, and then perform modulus normalization parallel operations. The output of the m-th matched filter result includes two groups, the z-th group of matched filter results y z (m) for: y z (m) =F H Λ z Fx z (m) ,m=1,2,...,M Where z = 1, 2, x1 is the local up-modulated signal, x2 is the time-reversed local down-modulated signal, the superscript (m) indicates the corresponding m-th beam signal, F represents the discrete Fourier transform matrix, Λ1 and Λ2 represent the diagonal matrices formed by discrete Fourier transform of the local up-modulated signal and the time-reversed local down-modulated signal, respectively, and the superscript H represents the conjugate transpose.

8. The method for rapid frame synchronization of multi-beam signals in underwater acoustic communication according to claim 1, characterized in that: The step 4 comprises: On the GPU side, a reduced peak search is performed based on the matched filtering results of the M-beams, and the peak values ​​W1 and W2 of the two sets of matched filters of the M-beams and the positions where the peaks appear are searched; Calculate the correlation value energy ratio Q = 10·log(W1 / W2) and compare it with the preset threshold Threshold. If Threshold>Q, synchronization is determined to be successful, complete the detection, and record the size and location of peak W1. Divide the location of peak W1 by the length of a single beam signal. The quotient corresponds to the beam sequence information where peak W1 is located, and the remainder corresponds to the location of peak W1 in this beam. This method obtains the optimal arrival direction and signal starting position of the arrival direction corresponding to each of the M beams, and transmits the result to the CPU. Otherwise, the signal is not synchronized, and detection continues.

Images