A method for identifying an M-APSK signal format based on a Gaussian mixture model

Abstract

The present application belongs to the technical field of electronic reconnaissance, and relates to a M-APSK signal format identification method based on a mixed Gaussian model. The present application provides a method for simultaneously identifying the modulation type and constellation distribution of an APSK signal. In combination with the characteristics that different LDPC identifiers correspond to different constellation distributions and the radius of inner and outer constellation points, the method can effectively identify the modulation type, LDPC identifier and constellation distribution of an APSK signal under the premise that a reconnaissance user does not know the LDPC identifier, and provide signal demodulation parameters for a subsequent demodulator. The present application has the beneficial effect that according to the characteristics that different LDPC identifiers correspond to different constellation distributions in the DVB-S2 communication standard, the present application designs an inter-class modulation identification method for M-APSK signals. The method can effectively identify the LDPC identifier and modulation order, and the constellation distribution of an APSK signal. The algorithm of the present application has small computation amount, simple structure, high identification accuracy, and can provide effective assistance for a communication reconnaissance device to demodulate an APSK signal.

Description

Technical Field

[0001] This invention belongs to the field of electronic reconnaissance technology and relates to an M-APSK signal format recognition method based on a Gaussian mixture model. Background Technology

[0002] DVB-S2 is a digital satellite broadcasting and communication standard that uses different coding and modulation schemes to transmit digital data, including video, audio, and data streams. LDPC (Locally Detected and Corrected Programming) is used for error detection and correction to improve the reliability of data transmission. This identifier is used to identify and describe the parameters and configuration of a specific LDPC code used in a DVB-S2 system, including the number of parity bits, code rate, and error correction capability. In the DVB-S2 standard, the selection of the LDPC code identifier is configured based on the communication link, typically involving key factors such as transmission code rate, error correction capability, receiving device support, channel conditions, and application requirements.

[0003] The signal format includes the APSK modulation order and the corresponding LDPC code identifier. These parameters determine the constellation distribution. In electronic reconnaissance of communications, the reconnaissance user has no prior information. They can only determine the signal format by identifying and analyzing the received signal. Based on the analysis results, they decide which constellation diagram the demodulation equipment should select to demodulate the detected signal. Accurately identifying the APSK signal format is crucial for improving the signal demodulation capability of the reconnaissance equipment. Since the constellation points of the APSK signal are arranged at uniform intervals on a unit circle, and the constellation points on each circle form a Gaussian mixture model, this method was invented. Summary of the Invention

[0004] This invention provides a method for simultaneously identifying the modulation type and constellation distribution of APSK signals. By combining the characteristic that different signal formats correspond to different constellation distributions, this method is suitable even when the reconnaissance user is unaware of the APSK signal format. It can effectively identify the APSK signal modulation order and LDPC identifier, providing signal demodulation parameters for subsequent demodulation.

[0005] The technical solution of this invention is as follows: First, the signal undergoes energy normalization preprocessing. Then, matched filtering and timing synchronization are performed to obtain a sampling point corresponding to each symbol. This sampling point is used for modulation identification. The modulation identification method is as follows: First, histogram statistics are performed on each sampling point. The statistical results are then smoothed and filtered to facilitate spectral peak search. The number of spectral peaks corresponds to the number of concentric circles in the APSK signal, and also represents the modulation order of the APSK signal. The horizontal coordinates of the spectral peaks represent the radius of the constellation diagram. By analyzing the proportional relationship between the horizontal coordinates of the spectral peaks, the LDPC identifier can be determined, thus revealing the constellation diagram distribution and defining the signal format.

[0006] A method for M-APSK signal format recognition based on Gaussian mixture model.

[0007] S1: Signal preprocessing

[0008] The received signal is defined as:

[0009]

[0010] Where A is the amplitude factor and f0 is the residual carrier frequency offset. It is the phase shift, g(t) is the root-raised cosine filter, T is the symbol period, w(n) is additive complex white Gaussian noise, and C is the phase shift. i APSK symbol mapping constellation points:

[0011]

[0012] Where R k Let n be the radius of the k-th concentric circle. k Let k be the number of signal points on the k concentric circles. Let i be the initial phase on the concentric circles. k The value ranges from 1 to n k -1. If the amplitude of the original signal is too small, signal preprocessing can be used to increase the amplitude of the signal, thereby normalizing the signal energy.

[0013] Normalizing the energy of the received signal r(n) to make A equal to 1 yields:

[0014]

[0015] S2. Perform matched filtering on the normalized signal to obtain:

[0016]

[0017] in The symbol represents convolution;

[0018] Performing timed synchronization on x(t) at t = kT + τ yields a sampling point corresponding to each symbol:

[0019]

[0020] Matched filtering and timing synchronization aim to extract the signal. However, without carrier and phase synchronization, the resulting sampling points are not the optimal sampling points for the signal. Constellation points cluster around a circle of a certain radius. These constellation points conform to their respective Gaussian distributions due to noise, forming a Gaussian mixture model. The symbol sequence is represented as x(k).

[0021] S3: Perform the following steps to determine the signal format;

[0022] 1) Signal modulus taking: Take the modulus of each sampling point x(k) to obtain the modulus value R. n :

[0023]

[0024] Where x I x Q These represent the real and imaginary parts of the symbol, respectively.

[0025] 2) Histogram Statistics: Perform histogram statistics on the magnitude values. The `nbin` parameter specifies the number of bars in the histogram, changing the granularity of the statistics. This will produce a histogram containing information about the distribution of magnitude values, where the horizontal axis corresponds to R. n That is, the radius of the concentric circles.

[0026] 3) Histogram Smoothing: The histogram is smoothed to highlight peaks in the amplitude distribution. Smoothing helps reduce the impact of noise on peaks, making them easier to identify and locate. The filter coefficients for a window length of N are:

[0027]

[0028] a = 1

[0029] 4) Peak Count Determination: Determine the number of peaks by observing the smoothed histogram. The `findpeaks` function can be used to find peaks. The peak detection threshold `MinPeakHeight` and the maximum number of search peaks `Npeaks` need to be set based on the number of symbols. Each peak corresponds to one cycle in the APSK constellation. For example, if there are two distinct peaks, it indicates that the signal uses 16APSK modulation. The determination method is: two peaks correspond to 16APSK modulation, three peaks correspond to 32APSK modulation, and so on. Radius Determination: Find the horizontal coordinate value corresponding to each peak; these values ​​correspond to different radii `R` in the constellation. n These radius values ​​are used for subsequent signal demodulation.

[0030] 5) LDPC identifier determination: based on radius R n The calculated ratio γ n as follows:

[0031] γ1=R2 / R1

[0032] γ2=R3 / R1

[0033] γ3=R4 / R1

[0034] Based on the DVB standard, determine the identifiers and constellation point distribution for LDPC (Low-Density Parity-Check) encoding. Provide the signal format to assist in subsequent demodulation and decoding.

[0035] The beneficial effects of this invention are as follows: Based on the characteristics of different constellation diagram distributions and LDPC identifiers corresponding to different signal formats in the DVB-S2 communication standard, this invention designs an inter-class modulation identification method for M-APSK signals. This method can effectively identify the LDPC identifier, modulation order, and constellation diagram distribution of APSK signals. The algorithm of this invention has low computational complexity, simple structure, and high identification accuracy, and can provide effective assistance for demodulating APSK signals in communication reconnaissance equipment. Attached Figure Description

[0036] Figure 1 This is a flowchart of the technical solution;

[0037] Figure 2 The constellation diagram and algorithm calculation results;

[0038] Figure 3 To identify the relationship between accuracy and SNR;

[0039] Figure 4 This is a flowchart of the modulation recognition algorithm. Detailed Implementation

[0040] The technical solution of the present invention will be further described below with reference to the accompanying drawings and simulations.

[0041] The proposed method was verified through simulation:

[0042] According to the DVB-S2 and DVB-S2X communication standards, the following nine signals are generated as shown in Table 1:

[0043] Table 1. Correspondence between APSK signal format modulation style and signal parameters

[0044]

[0045] The signal parameters were set as follows: code rate of 50kHz, sampling rate of 600kHz, and carrier frequency offset of 500Hz. A root-raised cosine filter was used as the shaping filter with a roll-off factor of 0.35 and a filter delay of 5. The histogram statistics (nbin) were set to 200, meaning the statistical radius was [0,2]. The peak search threshold (MinPeakHeight) was set to 100, and the maximum number of peaks searched (Npeaks) was 4. Modulation recognition was performed with symbol lengths of 1000 and 10000 symbols, simulating modulation recognition performance at signal-to-noise ratios (SNRs) ranging from 5dB to 20dB, with 100 Monte Carlo simulations conducted.

[0046] The simulation results are shown in the figure:

[0047] Figure 2 The left subplot shows the constellation diagram of the APSK signal after matched filtering and timing synchronization. It reveals the characteristic of different numbers of loops after timing synchronization for APSK signals of different orders: 16 APSK corresponds to 2 loops, 32 APSK to 3 loops, and 64 APSK to 4 loops. The right subplot shows the spectral peaks after statistical calculation. Each peak exhibits normal distribution characteristics and is related to the modulation order. The horizontal axis directly corresponds to the radius of each loop in the constellation diagram. By calculating the relationship between radii and the proportional relationship of spectral peaks, inter-class modulation identification can be performed for these nine APSK signals. The specific judgment process of the modulation identification algorithm is described below. Figure 4 Provided.

[0048] Figure 3 The relationship between modulation recognition accuracy and SNR with different numbers of symbols was simulated. The figure also corroborates the effectiveness of the proposed method. As can be seen from the figure, modulation recognition accuracy is related to the number of symbols; at the same signal-to-noise ratio, the more symbols, the higher the recognition accuracy.

Claims

1. A method for recognizing M-APSK signal format based on a Gaussian mixture model, characterized in that, Includes the following steps: S1. Define the received signal as: ， in It is the amplitude factor. It is a residual carrier frequency offset. It's a phase shift. It is a root-raised cosine filter, where T is the symbol period. It is additive white Gaussian noise. It is the APSK symbol mapping constellation point: ， in For the first The radius of each APSK constellation circle Count the constellation points on the circle. For the first The initial phase on each circle Values ​​range from 1 to ; For the received signal Normalizing the energy to make A equal to 1 yields: ， S2. Perform matched filtering on the normalized signal to obtain: ， in , , The symbol represents convolution; right Do Timed synchronization yields a sampling point corresponding to each symbol: ； S3, for each sampling point Modulo operation yields the modulo value. : ， in These represent the real and imaginary parts of the symbol, respectively. Histogram analysis is performed on the magnitude values, followed by smoothing filtering of the histogram. The filter coefficients for a window length of N are: ， ， The number of peaks is determined using the smoothed histogram. The function finds peaks by setting the peak detection threshold MinPeakHeight and the maximum number of search peaks Npeaks based on the number of symbols. Each peak corresponds to the number of turns in an APSK constellation diagram and also represents the modulation order of the APSK signal. The determination method is: 2 peaks correspond to 16 APSK modulation, 3 peaks correspond to 32 APSK modulation, and so on. Find the x-coordinate value corresponding to each peak; these values ​​correspond to different radii in the constellation chart. According to the radius Calculate the ratio as follows: ， ， ， According to DVB standards and ratios Determine the identifier and constellation point distribution of the LDPC encoding, and provide the signal format.

Images