Satellite signal receiving method of a ship automatic identification system

Abstract

The application relates to the technical field of satellite communication signal processing, and discloses a satellite signal receiving method of a ship automatic identification system. The method comprises the following steps: collecting and converting satellite radio frequency signals into initial digital sample streams; implementing multi-level spectrum sensing, positioning the carrier frequency point and bandwidth distribution of an effective ship identification signal, and constructing a carrier frequency point fingerprint atlas; based on the atlas guidance, separating out a narrowband sample stream of a target signal and a wideband environmental noise sample stream; according to the instantaneous power fluctuation of the narrowband sample stream, dynamically configuring a depth suppression coefficient of an adaptive notch filter; according to the statistical characteristics of the noise sample stream, combining the coefficient, and calculating a radiation pattern reconstruction vector; applying the vector to the narrowband sample stream for wavefront phase alignment and amplitude shaping, generating a purified baseband symbol stream; and inputting the ship message analysis engine to complete decoding. The application improves the weak signal capturing capacity and anti-interference performance in a complex electromagnetic environment.

Description

Technical Field

[0001] This invention relates to the field of satellite communication signal processing technology, specifically to a method for receiving satellite signals in an automatic identification system for ships. Background Technology

[0002] Existing satellite-based Automatic Identification System (AIS) receiver technologies typically rely on predefined channel lists or energy detection based on fixed thresholds to capture signals. In the complex electromagnetic environment of space, numerous ship signals, adjacent channel interference, and background noise mingle, making it difficult for traditional methods to dynamically and accurately identify and separate each valid ship identification signal. This coarse-grained signal acquisition approach easily leads to the omission of weak signals or the inability to effectively distinguish between signal and noise components, limiting the capacity and sensitivity of the receiving system.

[0003] Current adaptive antenna technology primarily employs interference suppression strategies that rely on static or gradual adjustments based on the statistical characteristics of the interference source or preset rules. However, the strength of ship signals received by satellites fluctuates rapidly and drastically due to factors such as ship attitude, marine environment, and antenna pointing. Fixed suppression strategies, or those detached from signal changes, often face a dilemma in such dynamic scenarios: deep nulls designed to suppress strong interference may simultaneously weaken or even obliterate the already weak useful ship signals; conversely, relaxing suppression to preserve small signals fails to effectively combat interference. Existing technologies lack a mechanism that can respond in real-time to changes in signal quality and finely adjust suppression behavior accordingly. Summary of the Invention

[0004] The purpose of this invention is to provide a satellite signal receiving method for an automatic identification system for ships, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a satellite signal receiving method for an automatic identification system (AIS), the method comprising:

[0006] The satellite radio frequency signals sensed by the acquisition and processing antenna are converted into an initial digital sample stream;

[0007] Multi-level spectrum sensing is performed on the initial digital sample stream to locate the carrier frequency points and bandwidth distribution of the effective ship identification signals and to construct a carrier frequency fingerprint map.

[0008] Based on the carrier frequency fingerprint spectrum, signal component extraction guided by the carrier frequency fingerprint spectrum is performed to separate the narrowband sample stream of the target ship identification signal and the accompanying broadband environmental noise sample stream.

[0009] Based on the instantaneous power fluctuations of the narrowband sample stream, the depth suppression coefficient of the adaptive null filter of the signal processing antenna is dynamically configured.

[0010] Based on the statistical characteristics of the broadband environmental noise sample stream and in conjunction with the depth suppression coefficient, the radiation pattern reconstruction vector of the signal processing antenna is calculated.

[0011] The radiation pattern reconstruction vector is applied to perform wavefront phase alignment and amplitude shaping on the narrowband sample stream to generate a cleaned baseband symbol stream;

[0012] The purified baseband symbol stream is input into the ship message parsing engine to complete the decoding of ship identity and navigation parameters.

[0013] Preferably, multi-level spectrum sensing is performed on the initial digital sample stream to locate the carrier frequency points and bandwidth distribution of valid ship identification signals, and to construct a carrier frequency fingerprint spectrum, including:

[0014] The initial digital sample stream is windowed and overlapped to obtain a continuous sequence of sample frames;

[0015] Perform high-resolution spectral estimation for each sample frame to generate a fine power spectral density curve for each sample frame;

[0016] Accumulated averaging over time is performed on all fine power spectral density curves to form a comprehensive power spectral density profile;

[0017] Identify all spectral peaks exceeding a preset power threshold from the integrated power spectral density profile, and record the center frequency and half-power bandwidth of each peak;

[0018] Based on the center frequency of each spectral peak, the fine power spectral density curve of each sample frame is traced back to extract the intensity fluctuation trajectory of the corresponding spectral peak on the time axis.

[0019] By combining the center frequency, half-power bandwidth, and intensity fluctuation trajectory of each spectral peak, a carrier frequency fingerprint entry characterizing the signal stability and occurrence pattern is generated.

[0020] Aggregate all carrier frequency fingerprint entries to establish a carrier frequency fingerprint map containing multi-dimensional features such as frequency, bandwidth, and time stability.

[0021] Preferably, based on the carrier frequency fingerprint spectrum, signal component extraction guided by the carrier frequency fingerprint spectrum is performed to separate the narrowband sample stream of the target ship identification signal and the accompanying broadband environmental noise sample stream, including:

[0022] The carrier frequency fingerprint spectrum is analyzed to select candidate carrier frequency fingerprint entries that correspond to the standard protocol frequency band of the Automatic Identification System for ships and whose intensity fluctuation trajectory shows periodicity.

[0023] Based on the center frequency and half-power bandwidth of the candidate carrier frequency fingerprint entries, define the corresponding bandpass filter parameters on the initial digital sample stream;

[0024] The initial digital sample stream is filtered by applying the defined bandpass filter parameters to extract multiple potential signal component streams corresponding to candidate carrier frequency fingerprint entries;

[0025] Calculate the cyclostationary characteristic measure of each potential signal component stream, and determine the potential signal component stream with the cyclostationary characteristic measure above the threshold as the target ship identification signal component stream;

[0026] All target ship identification signal component streams are frequency shifted and merged to form a narrowband sample stream;

[0027] Subtracting the components corresponding to the narrowband sample stream from the initial digital sample stream yields the broadband environmental noise sample stream.

[0028] Preferably, the depth suppression coefficient of the adaptive null filter of the signal processing antenna is dynamically configured based on the instantaneous power fluctuation of the narrowband sample stream, including:

[0029] Short-time energy calculations are performed on the narrowband sample stream to obtain the power time series;

[0030] Detect power drop events on the power time series and record the start time, duration, and power decay depth of each power drop event;

[0031] The frequency of power drop events and the average power attenuation depth within a preset time window were statistically analyzed.

[0032] Based on the occurrence frequency and average power attenuation depth, a preset mapping table is consulted to obtain a preliminary suggested value for the depth suppression coefficient.

[0033] Based on historical environmental interference data of the current operating frequency band of the signal processing antenna, the preliminary suggested value of the depth suppression coefficient is weighted and corrected.

[0034] The weighted correction result is quantized into a finite number of discrete levels, and the quantized value of the corresponding level is selected as the final depth suppression coefficient for the adaptive zero-dimple filter.

[0035] Preferably, based on the statistical characteristics of the broadband environmental noise sample stream and in conjunction with the depth suppression coefficient, the radiation pattern reconstruction vector of the signal processing antenna is calculated, including:

[0036] The spatial-temporal covariance matrix of the broadband environmental noise sample stream is estimated to obtain the noise spatial correlation matrix;

[0037] The noise spatial correlation matrix is ​​subjected to eigenvalue decomposition to extract its eigenvectors and eigenvalues;

[0038] Based on the depth suppression coefficient, the larger eigenvalues ​​corresponding to the strong interference direction and their associated eigenvectors are identified from the eigenvalues.

[0039] Based on the larger eigenvalues ​​and their associated eigenvectors, a projection matrix for the interference subspace is constructed.

[0040] An optimized objective function is constructed based on the criterion of suppressing the interference subspace while maintaining the target direction gain;

[0041] Solving the objective function yields a set of optimal complex weight values;

[0042] The complex weight values ​​are arranged and distributed according to the geometric positions of the signal processing antenna array elements to form a radiation pattern reconstruction vector.

[0043] Preferably, the radiation pattern reconstruction vector is applied to the narrowband sample stream to perform wavefront phase alignment and amplitude shaping to generate a purified baseband symbol stream, including:

[0044] The narrowband sample streams received by each array element of the signal processing antenna are used to form a spatial sample vector sequence.

[0045] The radiation pattern reconstruction vector and the spatial sample vector sequence are multiplied by a complex number to achieve beamforming, resulting in a synthesized single-channel sample sequence.

[0046] Carrier synchronization is performed on the synthesized single-channel sample sequence to compensate for the Doppler frequency offset and initial phase deviation caused by satellite motion, resulting in a phase-aligned sample sequence.

[0047] The envelope of the phase-aligned sample sequence is detected, and the sample amplitude is normalized and adjusted according to the envelope fluctuation to stabilize the amplitude within a preset range.

[0048] Matched filtering is applied to the amplitude-normalized sample sequence to improve the signal-to-noise ratio and suppress inter-symbol interference;

[0049] The matched-filtered sample sequence is judged at the optimal sampling point, and discrete symbol estimates are output.

[0050] The discrete symbol estimates are deinterleaved and forward error corrected by channel decoding to recover the original binary bit stream, which is the purified baseband symbol stream.

[0051] Preferably, the cyclostationary characteristic metric of each potential signal component stream is calculated, and potential signal component streams with a cyclostationary characteristic metric higher than a threshold are identified as target ship identification signal component streams, including:

[0052] For each potential signal component stream, calculate its cyclic autocorrelation function;

[0053] Extract characteristic peaks in the cyclic frequency domain from the cyclic autocorrelation function;

[0054] Calculate the average height and concentration index of the characteristic peaks;

[0055] The product of the average height and the concentration index is defined as a measure of the cyclostationary characteristics of the potential signal component stream;

[0056] The cyclostationary property measure of each potential signal component stream is compared with an adaptive threshold obtained by training with historical data;

[0057] Potential signal component streams whose cyclostationary property measure exceeds an adaptive threshold are labeled as candidate target signals;

[0058] Verify whether the cyclic frequency of the candidate target signal matches the known cyclic frequency set of the Automatic Identification System (AIS) signal, and finally determine the matching signal as the target ship identification signal component stream.

[0059] Preferably, the synthesized single-channel sample sequence is carrier synchronized to compensate for the Doppler frequency offset and initial phase deviation caused by satellite motion, resulting in a phase-aligned sample sequence, including:

[0060] A nonlinear transformation is performed on the synthesized single-channel sample sequence to generate harmonic components containing frequency offset information;

[0061] The phase of the harmonic components is tracked by a phase-locked loop structure to estimate the instantaneous Doppler frequency offset.

[0062] By smoothing the instantaneous Doppler frequency offset value, a stable Doppler frequency offset estimate is obtained;

[0063] A stable Doppler frequency offset estimate is used to generate the corresponding digitally controlled oscillator signal;

[0064] The frequency offset compensation is achieved by performing complex mixing between the digitally controlled oscillator signal and the synthesized single-channel sample sequence.

[0065] Phase error detection is performed on the frequency offset compensated sample sequence to obtain the residual instantaneous phase error;

[0066] The instantaneous phase error is filtered by a loop filter, and a phase control signal is generated to drive the phase rotator to rotate the phase of the sample sequence, finally obtaining the phase-aligned sample sequence.

[0067] Preferably, the preliminary suggested value of the depth suppression coefficient is weighted and corrected by combining historical environmental interference data of the current operating frequency band of the signal processing antenna, including:

[0068] Retrieve the time series of environmental interference intensity recorded by the signal processing antenna in the current operating frequency band over a recent period;

[0069] Calculate the mean and variance of the time series of environmental disturbance intensity;

[0070] Based on the mean and variance, calculate the stability factor of environmental disturbances;

[0071] Establish a correspondence between stability factors and weight coefficients; the higher the stability, the larger the weight coefficient.

[0072] The preliminary suggested values ​​for depth suppression coefficients are multiplied and weighted using the calculated weighting coefficients.

[0073] Check whether the weighted depth suppression coefficient value exceeds the upper and lower limits that the hardware can achieve. If it does, truncate it to the limit value.

[0074] The truncated values ​​are used as suggested values ​​for the weighted and corrected depth suppression coefficient for subsequent calculations.

[0075] Preferably, the instantaneous Doppler frequency offset value is smoothed and filtered to obtain a stable Doppler frequency offset estimate, including:

[0076] Set a sliding time window with a fixed number of sampling points;

[0077] The instantaneous Doppler frequency offset sequence is input into the sliding time window;

[0078] Within the sliding time window, median filtering is applied to the instantaneous Doppler frequency offset value to remove impulse jumps;

[0079] The result after median filtering is then subjected to mean filtering to obtain the average frequency offset value within the window.

[0080] Move the sliding time window forward by one sampling point, repeat the median filtering and mean filtering process, and obtain the next average frequency offset value;

[0081] All continuous average frequency offset values ​​are output in chronological order to form a stable sequence of Doppler frequency offset estimates.

[0082] Compared with the prior art, the beneficial effects of the present invention are:

[0083] By dynamically constructing a carrier frequency fingerprint map through multi-level spectrum sensing, this map accurately records the carrier and bandwidth information of each effective component in the signal environment. Using this map as a guide for signal component extraction overcomes the limitations of fixed frequency lists or single energy thresholds. This process enables precise location and separation of non-standard, weak target signals from a noisy background, providing a high-purity narrowband sample stream for subsequent processing and enhancing the system's initial signal acquisition and resolution capabilities under complex spectral conditions.

[0084] Based on the instantaneous power fluctuations of the separated narrowband sample stream, the depth suppression coefficient of the adaptive null filter is adjusted in real time. This mechanism uses the real-time quality changes of the target signal itself as the core decision parameter of the anti-interference strategy. Combining the statistical characteristics of broadband environmental noise, a precise radiation pattern reconstruction vector is calculated, and then the signal is subjected to wavefront phase alignment and amplitude shaping. This closed-loop processing enables the antenna pattern to be dynamically reshaped according to the fluctuations of the signal state, achieving deep suppression in the interference direction while maintaining optimal response in the signal arrival direction, thus solving the dynamic balance problem between signal fidelity and interference suppression in strong interference environments. Attached Figure Description

[0085] Figure 1 This is a schematic diagram illustrating the working principle of the satellite signal receiving method of the Automatic Identification System for Ships described in this invention.

[0086] Figure 2 A flowchart for constructing a carrier frequency fingerprint spectrum;

[0087] Figure 3 A flowchart for signal component extraction guided by carrier frequency fingerprinting;

[0088] Figure 4 The signal power distribution for the 8-channel antenna;

[0089] Figure 5 This is a comparison chart showing the effects of Doppler frequency offset smoothing filtering. Detailed Implementation

[0090] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0091] Please see Figure 1This invention provides a satellite signal receiving method for an Automatic Identification System (AIS). The method includes: a signal processing antenna sensing and receiving a satellite radio frequency signal, which is collected and converted into an initial digital sample stream. Multi-level spectrum sensing is performed on the initial digital sample stream to locate the carrier frequency and bandwidth distribution of the valid ship identification signal, and a carrier frequency fingerprint spectrum is constructed. Based on the constructed carrier frequency fingerprint spectrum, spectrum-guided signal component extraction is performed to separate the narrowband sample stream of the target ship identification signal from the accompanying broadband environmental noise sample stream. The depth suppression coefficient of the adaptive null filter within the signal processing antenna is dynamically configured according to the instantaneous power fluctuation of the narrowband sample stream. Based on the statistical characteristics of the broadband environmental noise sample stream and in conjunction with the aforementioned depth suppression coefficient, a radiation pattern reconstruction vector of the signal processing antenna is calculated. This radiation pattern reconstruction vector is applied to perform wavefront phase alignment and amplitude shaping on the narrowband sample stream to generate a purified baseband symbol stream. The purified baseband symbol stream is input to a ship message parsing engine to complete the decoding of ship identity and navigation parameters.

[0092] In one embodiment of the present invention, see [reference] Figure 2 The initial digital sample stream was sampled at a frequency of 2 MHz and lasted for 10 seconds. In the implementation, the initial digital sample stream was windowed and segmented using a Hanning window with 1024 sampling points, with 512 sampling points overlapping between adjacent segments to obtain a continuous sequence of sample frames. High-resolution spectral estimation based on the periodogram method was performed on each sample frame to generate a fine power spectral density curve for each frame, with a frequency resolution of approximately 1.95 kHz for each curve. In the implementation, a cumulative average was performed over time on all fine power spectral density curves. The power values ​​at corresponding frequency points of more than 300 sample frames were arithmetically averaged to form a comprehensive power spectral density profile. All spectral peaks exceeding a preset power threshold were identified from the comprehensive power spectral density profile. The preset power threshold was set to the average power of the comprehensive power spectral density profile plus 6 dB. The center frequency and half-power bandwidth of each peak exceeding the threshold were recorded. For example, in one instance, a spectral peak with a center frequency of 162.025 MHz was identified, with a half-power bandwidth of 5 kHz.

[0093] In some embodiments, based on the center frequency of each spectral peak, the fine power spectral density curve of each sample frame is traced back to extract the intensity fluctuation trajectory of the corresponding spectral peak on the time axis. Specifically, for a spectral peak with a center frequency of 162.025 MHz, the power value at that frequency point is read from the fine power spectral density curve of each frame, forming a time-power sequence spanning all sample frames. This sequence is the intensity fluctuation trajectory of the spectral peak. Combining the center frequency, half-power bandwidth, and intensity fluctuation trajectory of each spectral peak, a carrier frequency fingerprint entry characterizing the signal stability and occurrence regularity is generated. A carrier frequency fingerprint entry includes the center frequency value, half-power bandwidth value, variance of the intensity fluctuation trajectory, and periodic statistics of the intensity fluctuation trajectory crossing a specific threshold.

[0094] It is understandable that all carrier frequency fingerprint entries are aggregated to establish a carrier frequency fingerprint spectrum containing multi-dimensional features such as frequency, bandwidth, and time stability. In a specific implementation, the carrier frequency fingerprint spectrum is represented as a data structure or database table, with each record corresponding to a carrier frequency fingerprint entry. For multiple spectral peaks identified from the integrated power spectral density profile, such as three peaks with center frequencies of 161.975 MHz, 162.025 MHz, and 162.075 MHz, three independent carrier frequency fingerprint entries are generated and stored in the carrier frequency fingerprint spectrum. In some embodiments, the identification of spectral peaks and the generation of fingerprint entries can be determined and recorded using the following relationship:

[0095]

[0096] in: Indicates at the center frequency The spectral peak power value at that location, This represents the average power of the combined power spectral density profile. This indicates a preset power threshold offset, such as 6 dB. Spectral peaks that meet this condition will have their features extracted and form carrier frequency fingerprint entries.

[0097] Optionally, temporal stability characteristics are calculated by analyzing the intensity fluctuation trajectory. For example, the standard deviation of the intensity fluctuation trajectory within a time window is calculated; a smaller standard deviation indicates greater stability of the carrier signal in the time dimension. The construction of the carrier frequency fingerprint spectrum is a continuously updating process. As new initial digital sample streams are processed, new sample frame sequences are analyzed, the comprehensive power spectral density profile is updated, new or changing spectral peaks are identified, and corresponding carrier frequency fingerprint entries are added or modified, thereby maintaining the timeliness and accuracy of the carrier frequency fingerprint spectrum. It can be understood that the accuracy of the carrier frequency fingerprint spectrum, as a guiding basis for subsequent signal component extraction, directly depends on the rationality of parameter settings in multi-level spectrum sensing, such as the window function type, window length, overlap rate, and the choice of spectral estimation method. In specific implementations, these parameters can be fixedly configured or adaptively adjusted based on the hardware characteristics of the signal processing antenna and prior knowledge of the ship's automatic identification system signal.

[0098] In one embodiment of the present invention, see [reference] Figure 3 The carrier frequency fingerprint spectrum contains three carrier frequency fingerprint entries: entry A has a center frequency of 161.975 MHz and a half-power bandwidth of 4.8 kHz; entry B has a center frequency of 162.025 MHz and a half-power bandwidth of 5.0 kHz; and entry C has a center frequency of 162.075 MHz and a half-power bandwidth of 25 kHz. In practice, the carrier frequency fingerprint spectrum is analyzed to select candidate carrier frequency fingerprint entries that correspond to the standard protocol frequency band of the Automatic Identification System (AIS) and exhibit periodic intensity fluctuations. The standard protocol frequency band of the AIS ranges from 161.975 MHz to 162.025 MHz. The center frequencies of entries A and B fall within this range, and their intensity fluctuation trajectories show periodic power fluctuations at approximately 600-second intervals. Although entry C is also near the protocol frequency band, its intensity fluctuation trajectory is not periodic, exhibiting random fluctuations. Therefore, entries A and B are selected as candidate carrier frequency fingerprint entries.

[0099] In some embodiments, bandpass filter parameters are defined on the initial digital sample stream based on the center frequency and half-power bandwidth of the candidate carrier frequency fingerprint entries. For candidate carrier frequency fingerprint entry A, the center frequency of the bandpass filter is defined as 161.975 MHz, and the passband width is defined as 4.8 kHz. For candidate carrier frequency fingerprint entry B, the center frequency of the bandpass filter is defined as 162.025 MHz, and the passband width is defined as 5.0 kHz. The defined bandpass filter parameters are applied to filter the initial digital sample stream, extracting the potential signal component stream A corresponding to candidate carrier frequency fingerprint entry A, and the potential signal component stream B corresponding to candidate carrier frequency fingerprint entry B. Potential signal component stream C is not extracted because entry C was not selected as a candidate.

[0100] It is understandable that for each potential signal component stream, its cyclic autocorrelation function (CEC) is calculated. The calculation of the CEC involves a two-dimensional operation on the potential signal component stream using time delay and cyclic frequency. Characteristic peaks in the cyclic frequency domain are extracted from the CEC; these characteristic peaks are the CEC amplitudes that appear at a specific cyclic frequency and are significantly higher than the surrounding background. The average height and concentration index of these characteristic peaks are calculated. The average height is the arithmetic mean of the amplitudes of all extracted characteristic peaks. The concentration index quantifies the degree of clustering of characteristic peaks in the cyclic frequency domain. One calculation method is to define a narrow-band window centered on the typical cyclic frequency of the Automatic Identification System (AIS) signal, and calculate the ratio of the sum of squares of the amplitudes of characteristic peaks falling within this window to the sum of squares of all characteristic peak amplitudes; this ratio is the concentration index. The product of the average height and the concentration index is defined as a measure of the cyclic stationarity of the potential signal component stream. The calculation of the cyclic stationarity measure can be formally expressed as:

[0101]

[0102] in: This represents a measure of cyclic stationarity. Indicates the average height of the characteristic peaks. This represents a concentration index. Cyclostationary characteristics metrics are calculated for potential signal component streams A and B, respectively. and .

[0103] Optionally, the adaptive threshold obtained through training with historical data is a dynamic threshold value set based on statistical analysis of the cyclostationary characteristics of a large number of known ship AIS signals and interference signals. The cyclostationary characteristics of each potential signal component are compared with the adaptive threshold. In this implementation, the adaptive threshold is assumed to be... Calculations yielded , Both exceed the adaptive threshold, therefore potential signal component streams A and B are marked as candidate target signals. The cyclic frequencies of the candidate target signals are verified to conform to the known cyclic frequency set of the Automatic Identification System (AIS) signals. The known cyclic frequency set contains typical cyclic frequencies under the AIS GMSK modulation format, such as integer multiples of the symbol rate of 1 / 600 Hz. The cyclic autocorrelation functions of potential signal component streams A and B are examined for the occurrence of characteristic peaks, confirming that their main peaks occur near 1 / 600 Hz and its harmonic frequencies. Those conforming to this standard are ultimately determined as the target ship identification signal component streams.

[0104] In practical implementation, all target ship identification signal component streams are frequency-shifted and merged to form a narrowband sample stream. The center frequencies of target ship identification signal component streams A and B are shifted to zero intermediate frequency (IF) using digital mixing. Then, the two zero IF sample streams are complexly added and merged into a single complex signal sample stream, i.e., the narrowband sample stream. The corresponding components of the narrowband sample stream are subtracted from the initial digital sample stream. Specifically, the narrowband sample stream, after frequency shifting and merging, is restored to its original frequency band through the inverse process of its corresponding bandpass filter bank (i.e., a synthesis filter bank). Then, the corresponding segments of the initial digital sample stream are subtracted point by point, and the remaining part is the broadband environmental noise sample stream. In some embodiments, the power spectrum of the broadband environmental noise sample stream exhibits a flat or gradually changing characteristic over a wide frequency band, contrasting with the peak characteristics of the narrowband sample stream at specific frequency points. It can be understood that the calculation of the cyclostationary characteristic metric and threshold comparison are the core steps in distinguishing the ship automatic identification system signals with specific periodic modulation characteristics from random noise or interference that do not possess such characteristics.

[0105] In one embodiment of the invention, the narrowband sample stream contains signals from the Automatic Identification System (AIS) of two ships, with a sampling rate of 48 kHz. Short-time energy calculations are performed on the narrowband sample stream using a rectangular window with 256 sampling points of non-overlapping dimensions, resulting in a power time series containing approximately 22,500 power estimation points. Power drop events are detected on the power time series, with the detection criterion being that the power values ​​at three consecutive points are lower than 70% of the moving average power value of the previous ten points. The start time, duration, and power attenuation depth of each power drop event are recorded. The power attenuation depth is defined as the ratio of the power at the start of the drop to the power at the trough. A total of 18 power drop events were detected within 120 seconds.

[0106] In some embodiments, the frequency and average power attenuation depth of power sag events within a preset time window are statistically analyzed. The preset time window is set to 30 seconds, with a sliding step of 5 seconds. For example, within a time window from the 25th to the 55th second, 5 power sag events are counted, with an occurrence frequency of 10 times per minute. The arithmetic mean of the power attenuation depth ratio of these 5 events within the window is 8.5 dB. Based on the occurrence frequency and average power attenuation depth, a preset mapping table is consulted. The mapping table takes the occurrence frequency and average power attenuation depth as two-dimensional inputs and outputs a floating-point number ranging from 0 to 1 as a preliminary suggested value for the depth suppression coefficient. Taking an occurrence frequency of 10 times per minute and an average power attenuation depth of 8.5 dB as an example, the preliminary suggested value for the depth suppression coefficient obtained from the table is 0.65.

[0107] It can be understood that the process involves retrieving the environmental interference intensity time series recorded by the signal processing antenna over the most recent period in the current operating frequency band. The current operating frequency band is around 162 MHz. The process retrieves the environmental interference intensity time series over the most recent 300 seconds, with one sampling point per second. This series contains 300 data points, measured in milliwatts (dB). The mean and variance of the environmental interference intensity time series are calculated. Assuming the calculated mean is -110 DB and the variance is 4.0 (dB)², the stability factor of the environmental interference is calculated based on the mean and variance. One way to calculate the stability factor is to normalize the inverse of the variance of the environmental interference intensity time series. The specific calculation can be expressed as:

[0108]

[0109] in: The stability factor representing environmental disturbances. The variance of the time series representing the intensity of environmental disturbances. This is a very small positive integer used to prevent division by zero errors; in this example, it is set to 0.01. It is a normalization constant, which makes the stability factor The range of values ​​falls between 0 and 1, here we set... Substituting the calculated variance value of 4.0 into the equation yields the stability factor. It is approximately 0.124.

[0110] Optionally, a correspondence can be established between stability factors and weight coefficients; higher stability corresponds to larger weight coefficients. One mapping relationship is a linear one: weight coefficients... Use the calculated weighting coefficients. The initial suggested depth suppression coefficient is multiplied and weighted. The initial suggested depth suppression coefficient is 0.65, with weighting coefficients... The initial value is 0.124, and after multiplication and weighting, the result is approximately 0.0806. The weighted depth suppression coefficient is checked to see if it exceeds the upper or lower limits achievable by the hardware, which sets the depth suppression coefficient range to 0 to 1.0. The value 0.0806 falls within this range and does not need to be truncated. The truncated value is used as the suggested weighted depth suppression coefficient for subsequent calculations; in this example, it is 0.0806. In practice, the suggested weighted depth suppression coefficient value of 0.0806 is quantized into a finite number of discrete levels, set to [0, 0.2, 0.4, 0.6, 0.8, 1.0]. The closest level value, 0.2, is selected as the final depth suppression coefficient configured for the adaptive zero-dimple filter.

[0111] In some embodiments, the preset mapping table is obtained by fitting a large amount of experimental data. The mapping table maps high occurrence frequency and deep attenuation depth to higher initial suggested values ​​for the depth suppression coefficient, for example, when the occurrence frequency is greater than 15 times per minute and the average attenuation depth is greater than 10 dB, the suggested value may be 0.9; and maps low occurrence frequency and shallow attenuation depth to lower suggested values, for example, when the occurrence frequency is less than 5 times per minute and the average attenuation depth is less than 5 dB, the suggested value may be 0.2. The stability factor of environmental interference can be calculated in various ways, such as based on the autocorrelation function or power spectral flatness of the environmental interference intensity time series. It is understood that the purpose of weighted correction using historical environmental interference data is to ensure that the configuration of the depth suppression coefficient depends not only on the instantaneous power fluctuation characteristics of the current narrowband sample stream, but also on the long-term statistical characteristics of the current RF environment. When environmental interference fluctuations are severe, the response weight to instantaneous fluctuations is reduced to avoid frequent and drastic jumps in the depth suppression coefficient.

[0112] In one embodiment of the present invention, the broadband ambient noise sample stream currently received by the signal processing antenna is an 8-channel complex data stream, with each channel containing 10,000 samples at a sampling rate of 96 kHz. The spatial-temporal covariance matrix of the broadband ambient noise sample stream is estimated to obtain the noise spatial correlation matrix. The estimation method involves calculating the covariance of 10,000 snapshot samples, resulting in an 8x8 complex matrix, i.e., the noise spatial correlation matrix. Eigenvalue decomposition is then performed on the noise spatial correlation matrix to extract its eigenvectors and eigenvalues. The eigenvalue decomposition yields 8 eigenvalues ​​and corresponding 8 eigenvectors, each of which is an 8-dimensional complex column vector.

[0113] Based on the depth suppression coefficient, the larger eigenvalues ​​and their associated eigenvectors corresponding to the strong interference directions are identified from the eigenvalues. The final depth suppression coefficient configured for the adaptive null filter is 0.2. A threshold related to the depth suppression coefficient is set, for example, 0.2 times the largest eigenvalue is used as the threshold; eigenvalues ​​exceeding this threshold are considered larger eigenvalues. Assuming the eigenvalue decomposition results are arranged in descending order, refer to Table 1 for the eigenvalues ​​and corresponding indices of the noise spatial correlation matrix.

[0114] Table 1: Eigenvalues ​​of the Noise Spatial Correlation Matrix

[0115] Eigenvalue index Eigenvalues ​​(after normalization) 1 10.2 2 5.1 3 1.2 4 0.9 5 0.6 6 0.4 7 0.3 8 0.2

[0116] The largest eigenvalue is 10.2, and 0.2 times that is 2.04. Based on this threshold, the eigenvalues ​​corresponding to eigenvalue indices 1 and 2 (10.2 and 5.1) are identified as larger eigenvalues, and their associated eigenvectors (denoted as...) are... and ) were also extracted.

[0117] In some embodiments, a disturbance subspace projection matrix is ​​constructed based on larger eigenvalues ​​and their associated eigenvectors. Disturbance subspace projection matrix Calculated using the following formula: ,in It is composed of eigenvectors and As an 8x2 matrix composed of column vectors, the superscript Let represent the conjugate transpose. An optimization objective function is constructed based on the criterion of suppressing interference subspaces while preserving the target direction gain. The target direction is assumed to be the array normal direction, and the corresponding array manifold vector is denoted as . The optimization objective function is set to be within the constraints. Minimize output power ,in The unknown is the 8-dimensional complex weight vector. This is the noise spatial correlation matrix. Solving this objective function yields a set of optimal complex weight values. These complex weight values ​​are then distributed according to the geometric positions of the signal processing antenna array elements, forming a radiation pattern reconstruction vector. Radiation pattern reconstruction vector This is the optimal complex weight vector obtained by solving the problem, where each element corresponds to the complex weighting coefficient of an antenna element.

[0118] It can be understood that the narrowband sample stream received by each array element of the signal processing antenna constitutes a spatial sample vector sequence. At any sampling time, the outputs of the eight array elements form an 8-dimensional complex column vector, and the column vectors at all sampling times are arranged in chronological order to form the spatial sample vector sequence. Complex multiplication is performed between the radiation pattern reconstruction vector and the spatial sample vector sequence to achieve beamforming, resulting in a synthesized single-channel sample sequence. Specifically, at each sampling time, the radiation pattern reconstruction vector... The conjugate transpose of the vector is multiplied by the current spatial sample vector (8-dimensional) to output a complex scalar. The scalar outputs at all sampling times are arranged in order to obtain the synthesized single-channel sample sequence.

[0119] In practical implementation, the synthesized single-channel sample sequence undergoes carrier synchronization to compensate for the Doppler frequency offset and initial phase deviation caused by satellite motion, resulting in a phase-aligned sample sequence. The carrier synchronization process includes steps such as coarse frequency synchronization and phase-locked loop tracking, ultimately outputting a complex sample sequence whose frequency and phase are aligned with the transmitting carrier. The envelope of the phase-aligned sample sequence is detected, and the sample amplitude is normalized based on envelope fluctuations to stabilize the amplitude within a preset range. Envelope detection is achieved by calculating the magnitude of the sample sequence, and normalization is performed through an automatic gain control loop that scales the sample amplitude to a fixed root mean square value, such as 1.0. Matched filtering is applied to the amplitude-normalized sample sequence to improve the signal-to-noise ratio and suppress inter-symbol interference. The impulse response of the matched filter matches the GMSK pulse shaping waveform used in the Automatic Identification System (AIS) signal, and the filtering process is implemented through convolution.

[0120] Optionally, the matched-filtered sample sequence is judged at the optimal sampling point, and discrete symbol estimates are output. The optimal sampling point is determined by a timing synchronization loop, sampling is performed near the center point of each symbol period, and the sampled values ​​are mapped to binary symbols according to a certain decision threshold. Deinterleaving and forward error correction of channel decoding are performed on the discrete symbol estimates to recover the original binary bit stream, which is the purified baseband symbol stream. The deinterleaving operation is performed according to the interleaving pattern specified in the Automatic Identification System (AIS) protocol, and the channel decoding uses the Viterbi algorithm or a similar algorithm to decode the convolutional code. In some embodiments, the estimation of the noise spatial correlation matrix can adopt an exponentially weighted recursive update method to adapt to non-stationary interference environments. It can be understood that the core of the calculation of the radiation pattern reconstruction vector lies in constructing the interference subspace using the statistical characteristics of the broadband environmental noise sample stream, and aligning the nulls of the array receiving radiation pattern with the interference direction through constraint optimization, while maintaining the gain of the desired signal direction. This process significantly depends on the accuracy of eigenvalue decomposition and the reasonable setting of the optimization objective function.

[0121] See Figure 4The figure presents the differentiated characteristics of the received power of each antenna channel. Specifically, the figure uses "antenna channel" as the horizontal axis and "received power (dB)" as the vertical axis to quantify the received power values ​​of the eight channels: Channel 1 has a received power of -1.80dB, Channel 2 -1.28dB, Channel 3 -0.80dB, Channel 4 -0.35dB, Channel 5 0.08dB, Channel 6 0.51dB, Channel 7 0.89dB, and Channel 8 1.28dB. From the perspective of signal reception characteristics, the significant differences in the received power of each channel reflect the spatial channel non-uniformity of the multi-channel antenna array in real-world scenarios. This non-uniformity is related to the spatial layout of the antenna array, the channel fading characteristics of each channel, and may also be affected by the spatial distribution of environmental interference. This power distribution data can serve as the basic input for subsequent space-time signal processing (such as interference subspace construction and radiation pattern reconstruction). For example, combining the relative differences in the power of each channel can help optimize the suppression direction and depth of the adaptive null filter, thereby improving the reception gain of the target signal.

[0122] In one embodiment of the present invention, the specific process of Doppler frequency offset compensation and smoothing filtering in carrier synchronization is based on a synthesized single-channel sample sequence. The synthesized single-channel sample sequence has a length of 0.1 seconds, a sampling rate of 48 kHz, and contains 4800 complex sampling points. A nonlinear transformation is performed on the synthesized single-channel sample sequence to generate harmonic components containing frequency offset information. The nonlinear transformation uses a square-law operation, i.e., taking the square of the modulus of each complex sampling point. The phase of the harmonic components is tracked through a phase-locked loop (PLL) structure to estimate the instantaneous Doppler frequency offset value. The phase detector of the PLL compares the phase of the harmonic components with the phase difference of a local numerically controlled oscillator (NCO) signal. After passing through a loop filter, the NCO drives the NCO. The PLL outputs an instantaneous Doppler frequency offset estimate at each sampling point, thus obtaining a sequence containing 4800 instantaneous Doppler frequency offset values, where the unit of the instantaneous Doppler frequency offset value is Hertz (Hz).

[0123] In the specific implementation, a sliding time window with a fixed number of sampling points is set. The length of the sliding time window is set to 101 sampling points. The instantaneous Doppler frequency offset value sequence is input into the sliding time window. Within the sliding time window, median filtering is performed on the instantaneous Doppler frequency offset values ​​to remove impulsive jumps. For example, for the 101 instantaneous Doppler frequency offset values ​​within the sliding time window, they are sorted by size, and the 51st value after sorting is taken as the output of the median filter. The result after median filtering is then subjected to mean filtering to obtain the average frequency offset value within the window. The mean filtering calculates the arithmetic mean of the 10 points before and after the median filter output value (a total of 21 points). The sliding time window is moved forward by one sampling point, and the median filtering and mean filtering processes are repeated to obtain the next average frequency offset value. All continuous average frequency offset values ​​are output in chronological order to form a stable Doppler frequency offset estimation value sequence. Because the sliding window will have missing data at both ends of the sequence, the length of the stable Doppler frequency offset estimate sequence is slightly shorter than the original instantaneous Doppler frequency offset sequence, for example, the length is 4700 points.

[0124] In some embodiments, the estimation process of the instantaneous Doppler frequency offset can be represented by the following relationship:

[0125]

[0126] in: Indicates the first The instantaneous Doppler frequency offset value estimated at each sampling time point Indicates the sampling period. This indicates that the phase-locked loop is in the first... The phase error increment detected at each sampling time.

[0127] It is understandable that phase error detection is performed on the frequency offset-compensated sample sequence to obtain the residual instantaneous phase error. Phase error detection is typically implemented using a phase detector, such as a Costas ring phase detection algorithm, which compares the difference between the phase of the compensated signal and a local reference phase. The instantaneous phase error is filtered by a loop filter, generating a phase control signal to drive a phase rotator, rotating the phase of the sample sequence to obtain the phase-aligned sample sequence. The loop filter is usually a low-pass filter used to smooth high-frequency noise in the instantaneous phase error. The phase rotator, based on the phase control signal, applies a slowly changing phase rotation to the frequency offset-compensated sample sequence to compensate for the initial phase deviation and residual low-frequency phase jitter. In practice, assuming that the initial instantaneous Doppler frequency offset sequence fluctuates rapidly in the range of -500 Hz to +500 Hz, after passing through a sliding window with a length of 101 sampling points, the fluctuation range of the output stable Doppler frequency offset estimate sequence is significantly tightened, for example, it changes smoothly in the lower frequency range of -50 Hz to +50 Hz. This reflects the effect of smoothing filtering in removing pulse jumps and short-term fluctuations.

[0128] Optionally, the length of the sliding window needs to be chosen with a trade-off between tracking speed and smoothness. A longer sliding window can provide a smoother sequence of Doppler frequency offset estimates, but its ability to track rapid changes in Doppler frequency will decrease; a shorter sliding window can track changes faster, but its smoothing and noise reduction effect will be weakened. In some embodiments, the length of the sliding window can be adaptively adjusted based on prior knowledge of the satellite orbit or the signal-to-noise ratio of the signal. The order of median filtering and mean filtering can be interchanged, or other smoothing methods such as weighted moving average can be used. It can be understood that the smoothing filtering process is the key to connecting instantaneous frequency offset estimation and stable frequency offset control. Its purpose is to use statistical characteristics in the time domain to extract a stable estimate that is closer to the true Doppler dynamic trend from the noise-contaminated instantaneous observation, thereby driving the digitally controlled oscillator to achieve more accurate carrier frequency compensation.

[0129] See Figure 5In the Doppler frequency offset compensation process of carrier synchronization, a comparison of the time-domain characteristics of the original instantaneous frequency offset and the smoothed frequency offset is presented. Specifically, the red curve in the figure corresponds to the instantaneous Doppler frequency offset value sequence (sampling rate 48kHz, duration 0.1 seconds) estimated by a phase-locked loop from the synthesized single-channel sample sequence. Its fluctuation range covers -500Hz to +500Hz, reflecting the pulse jumps and rapid fluctuations under noise interference. The green curve is the stable frequency offset estimate sequence obtained by applying a sliding window of 101 sampling points to the original instantaneous frequency offset sequence, first performing median filtering (removing pulse jumps), and then performing mean filtering (smoothing short-term fluctuations). Its fluctuation range is tightened to -50Hz to +50Hz, intuitively demonstrating the smoothing and noise suppression effect of the "median + mean" combined filtering on the frequency offset sequence. This result provides a stable input basis for the subsequent accurate frequency offset compensation of the digitally controlled oscillator.

[0130] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0131] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for receiving satellite signals in an Automatic Identification System (AIS), characterized in that, include: The satellite radio frequency signals sensed by the acquisition and processing antenna are converted into an initial digital sample stream; Multi-level spectrum sensing is performed on the initial digital sample stream to locate the carrier frequency points and bandwidth distribution of the effective ship identification signals and to construct a carrier frequency fingerprint map. Based on the carrier frequency fingerprint spectrum, signal component extraction guided by the carrier frequency fingerprint spectrum is performed to separate the narrowband sample stream of the target ship identification signal and the accompanying broadband environmental noise sample stream. Based on the instantaneous power fluctuations of the narrowband sample stream, the depth suppression coefficient of the adaptive null filter of the signal processing antenna is dynamically configured. Based on the statistical characteristics of the broadband environmental noise sample stream and in conjunction with the depth suppression coefficient, the radiation pattern reconstruction vector of the signal processing antenna is calculated. The radiation pattern reconstruction vector is applied to perform wavefront phase alignment and amplitude shaping on the narrowband sample stream to generate a cleaned baseband symbol stream; The purified baseband symbol stream is input into the ship message parsing engine to complete the decoding of ship identity and navigation parameters.

2. The satellite signal receiving method for a ship automatic identification system according to claim 1, characterized in that, Multi-level spectrum sensing is performed on the initial digital sample stream to locate the carrier frequency points and bandwidth distribution of valid ship identification signals, and to construct a carrier frequency fingerprint spectrum, including: The initial digital sample stream is windowed and overlapped to obtain a continuous sequence of sample frames; Perform high-resolution spectral estimation for each sample frame to generate a fine power spectral density curve for each sample frame; Accumulated averaging over time is performed on all fine power spectral density curves to form a comprehensive power spectral density profile; Identify all spectral peaks exceeding a preset power threshold from the integrated power spectral density profile, and record the center frequency and half-power bandwidth of each peak; Based on the center frequency of each spectral peak, the fine power spectral density curve of each sample frame is traced back to extract the intensity fluctuation trajectory of the corresponding spectral peak on the time axis. By combining the center frequency, half-power bandwidth, and intensity fluctuation trajectory of each spectral peak, a carrier frequency fingerprint entry characterizing the signal stability and occurrence pattern is generated. Aggregate all carrier frequency fingerprint entries to establish a carrier frequency fingerprint map containing multi-dimensional features such as frequency, bandwidth, and time stability.

3. The satellite signal receiving method for a ship automatic identification system according to claim 2, characterized in that, Based on the carrier frequency fingerprint spectrum, signal component extraction guided by the carrier frequency fingerprint spectrum is performed to separate the narrowband sample stream of the target ship identification signal and the accompanying broadband environmental noise sample stream, including: The carrier frequency fingerprint spectrum is analyzed to select candidate carrier frequency fingerprint entries that correspond to the standard protocol frequency band of the Automatic Identification System for ships and whose intensity fluctuation trajectory shows periodicity. Based on the center frequency and half-power bandwidth of the candidate carrier frequency fingerprint entries, define the corresponding bandpass filter parameters on the initial digital sample stream; The initial digital sample stream is filtered by applying the defined bandpass filter parameters to extract multiple potential signal component streams corresponding to candidate carrier frequency fingerprint entries; Calculate the cyclostationary characteristic measure of each potential signal component stream, and determine the potential signal component stream with the cyclostationary characteristic measure above the threshold as the target ship identification signal component stream; All target ship identification signal component streams are frequency shifted and merged to form a narrowband sample stream; Subtracting the components corresponding to the narrowband sample stream from the initial digital sample stream yields the broadband environmental noise sample stream.

4. The satellite signal receiving method for a ship automatic identification system according to claim 1, characterized in that, Based on the instantaneous power fluctuations of the narrowband sample stream, the depth suppression coefficient of the adaptive null filter of the signal processing antenna is dynamically configured, including: Short-time energy calculations are performed on the narrowband sample stream to obtain the power time series; Detect power drop events on the power time series and record the start time, duration, and power decay depth of each power drop event; The frequency of power drop events and the average power attenuation depth within a preset time window were statistically analyzed. Based on the occurrence frequency and average power attenuation depth, a preset mapping table is consulted to obtain a preliminary suggested value for the depth suppression coefficient. Based on historical environmental interference data of the current operating frequency band of the signal processing antenna, the preliminary suggested value of the depth suppression coefficient is weighted and corrected. The weighted correction result is quantized into a finite number of discrete levels, and the quantized value of the corresponding level is selected as the final depth suppression coefficient for the adaptive zero-dimple filter.

5. The satellite signal receiving method for a ship automatic identification system according to claim 4, characterized in that, Based on the statistical characteristics of the broadband environmental noise sample stream and in conjunction with the depth suppression coefficient, the radiation pattern reconstruction vector of the signal processing antenna is calculated, including: The spatial-temporal covariance matrix of the broadband environmental noise sample stream is estimated to obtain the noise spatial correlation matrix; The noise spatial correlation matrix is ​​subjected to eigenvalue decomposition to extract its eigenvectors and eigenvalues; Based on the depth suppression coefficient, the larger eigenvalues ​​corresponding to the strong interference direction and their associated eigenvectors are identified from the eigenvalues. Based on the larger eigenvalues ​​and their associated eigenvectors, a projection matrix for the interference subspace is constructed. An optimized objective function is constructed based on the criterion of suppressing the interference subspace while maintaining the target direction gain; Solving the objective function yields a set of optimal complex weight values; The complex weight values ​​are arranged and distributed according to the geometric positions of the signal processing antenna array elements to form a radiation pattern reconstruction vector.

6. The satellite signal receiving method for a ship automatic identification system according to claim 5, characterized in that, The radiation pattern reconstruction vector is applied to the narrowband sample stream to perform wavefront phase alignment and amplitude shaping, generating a cleaned baseband symbol stream, including: The narrowband sample streams received by each array element of the signal processing antenna are used to form a spatial sample vector sequence. The radiation pattern reconstruction vector and the spatial sample vector sequence are multiplied by a complex number to achieve beamforming, resulting in a synthesized single-channel sample sequence. Carrier synchronization is performed on the synthesized single-channel sample sequence to compensate for the Doppler frequency offset and initial phase deviation caused by satellite motion, resulting in a phase-aligned sample sequence. The envelope of the phase-aligned sample sequence is detected, and the sample amplitude is normalized and adjusted according to the envelope fluctuation to stabilize the amplitude within a preset range. Matched filtering is applied to the amplitude-normalized sample sequence to improve the signal-to-noise ratio and suppress inter-symbol interference; The matched-filtered sample sequence is judged at the optimal sampling point, and discrete symbol estimates are output. The discrete symbol estimates are deinterleaved and forward error correction is performed during channel decoding to recover the original binary bit stream, which is the purified baseband symbol stream.

7. The satellite signal receiving method for a ship automatic identification system according to claim 3, characterized in that, Calculate the cyclostationary characteristic metric for each potential signal component stream, and determine the potential signal component streams with a cyclostationary characteristic metric higher than a threshold as target ship identification signal component streams, including: For each potential signal component stream, calculate its cyclic autocorrelation function; Extract characteristic peaks in the cyclic frequency domain from the cyclic autocorrelation function; Calculate the average height and concentration index of the characteristic peaks; The product of the average height and the concentration index is defined as a measure of the cyclostationary characteristics of the potential signal component stream; The cyclostationary property measure of each potential signal component stream is compared with an adaptive threshold obtained by training with historical data; Potential signal component streams whose cyclostationary property measure exceeds an adaptive threshold are labeled as candidate target signals; Verify whether the cyclic frequency of the candidate target signal matches the known cyclic frequency set of the Automatic Identification System (AIS) signal, and finally determine the matching signal as the target ship identification signal component stream.

8. The satellite signal receiving method for a ship automatic identification system according to claim 6, characterized in that, Carrier synchronization is performed on the synthesized single-channel sample sequence to compensate for the Doppler frequency offset and initial phase deviation caused by satellite motion, resulting in a phase-aligned sample sequence, including: A nonlinear transformation is performed on the synthesized single-channel sample sequence to generate harmonic components containing frequency offset information; The phase of the harmonic components is tracked by a phase-locked loop structure to estimate the instantaneous Doppler frequency offset. By smoothing the instantaneous Doppler frequency offset value, a stable Doppler frequency offset estimate is obtained; A stable Doppler frequency offset estimate is used to generate the corresponding digitally controlled oscillator signal; The frequency offset compensation is achieved by performing complex mixing between the digitally controlled oscillator signal and the synthesized single-channel sample sequence. Phase error detection is performed on the frequency offset compensated sample sequence to obtain the residual instantaneous phase error; The instantaneous phase error is filtered by a loop filter, and a phase control signal is generated to drive the phase rotator to rotate the phase of the sample sequence, finally obtaining the phase-aligned sample sequence.

9. A satellite signal receiving method for a ship automatic identification system according to claim 4, characterized in that, Based on historical environmental interference data of the current operating frequency band of the signal processing antenna, the preliminary suggested value of the depth suppression coefficient is weighted and corrected, including: Retrieve the time series of environmental interference intensity recorded by the signal processing antenna in the current operating frequency band over a recent period; Calculate the mean and variance of the time series of environmental disturbance intensity; Based on the mean and variance, calculate the stability factor of environmental disturbances; Establish a correspondence between stability factors and weight coefficients; the higher the stability, the larger the weight coefficient. The preliminary suggested values ​​for depth suppression coefficients are multiplied and weighted using the calculated weighting coefficients. Check whether the weighted depth suppression coefficient value exceeds the upper and lower limits that the hardware can achieve. If it does, truncate it to the limit value. The truncated values ​​are used as suggested values ​​for the weighted and corrected depth suppression coefficient for subsequent calculations.

10. A satellite signal receiving method for a ship automatic identification system according to claim 8, characterized in that, Smoothing filters are applied to the instantaneous Doppler frequency offset values ​​to obtain stable Doppler frequency offset estimates, including: Set a sliding time window with a fixed number of sampling points; The instantaneous Doppler frequency offset sequence is input into the sliding time window; Within the sliding time window, median filtering is applied to the instantaneous Doppler frequency offset value to remove impulse jumps; The result after median filtering is then subjected to mean filtering to obtain the average frequency offset value within the window. Move the sliding time window forward by one sampling point, repeat the median filtering and mean filtering process, and obtain the next average frequency offset value; All continuous average frequency offset values ​​are output in chronological order to form a stable sequence of Doppler frequency offset estimates.

Images