GNSS signal interference suppression system and method based on multi-algorithm time delay alignment

Abstract

The application relates to a GNSS signal interference suppression system and method based on multi-algorithm time delay alignment, and belongs to the technical field of satellite navigation communication. The system comprises a data acquisition and preprocessing module, an intelligent decision control module, a multi-algorithm execution module, an unbiased anti-interference processing module and a result feedback output module. The data acquisition and preprocessing module is responsible for original signal receiving, interference type detection and identification and feature extraction; the intelligent decision control module outputs an optimal anti-interference algorithm; the multi-algorithm execution module processes GNSS signal interference suppression through three independent processing paths of FDPB, TDPB and an adaptive filter in parallel; the unbiased anti-interference processing module completes time delay modeling analysis, reference determination and compensation operations, finally outputs the GNSS signal after interference suppression and unbiased anti-interference processing to the result feedback output module for post-processing. The application can significantly improve the receiving accuracy and stability of navigation signals.

Description

Technical Field

[0001] This application relates to the field of satellite navigation and communication technology, and in particular to a GNSS signal interference suppression system and method based on multi-algorithm time delay alignment. Background Technology

[0002] To cope with complex interference environments, Global Navigation Satellite Systems (GNSS) have developed a variety of anti-jamming algorithms, such as Frequency Domain Pulse Nullification (FDPB), Time Domain Pulse Nullification (TDPB), and Adaptive Filter (AF) based on LMS (Least Mean Square).

[0003] However, existing technologies generally suffer from the following bottlenecks: First, they lack intelligent decision-making mechanisms and typically employ fixed or manual switching strategies to cope with different interferences. It is difficult to dynamically adapt the optimal processing path according to the channel state, leading to performance degradation under time-varying interference. Second, various algorithms introduce significantly different processing delays due to structural differences. For example, FDPB requires FFT / IFFT (Fast Fourier Transform / Inverse Fast Fourier Transform) operations, AF is affected by the iteration step size and produces time-varying delays, while TDPB has almost zero delay. When the system switches between different algorithms, the time base of the output signal of each path is inconsistent, causing the main peak of the correlation function to drift, resulting in pseudorange abrupt changes, which seriously undermines the stability of the high-precision positioning system.

[0004] Although existing research has attempted to improve anti-interference capabilities through single-algorithm optimization, such as using deep learning to identify interference types and then consistently applying a certain type of filter, the time base misalignment problem caused by multipath processing has not yet been resolved. While some solutions propose delay estimation methods, they only model specific algorithms and cannot achieve unified delay alignment across algorithms. Summary of the Invention

[0005] Therefore, it is necessary to provide a GNSS signal interference suppression system and method based on multi-algorithm delay alignment to address the above-mentioned technical problems. This system can simultaneously achieve "dynamic selection of the optimal anti-interference algorithm" and "multi-path delay consistency guarantee", fundamentally suppressing pseudorange jumps and improving the overall ranging accuracy and availability of the GNSS system.

[0006] A GNSS signal interference suppression system based on multi-algorithm delay alignment, the system comprising: The data acquisition and preprocessing module is used to acquire GNSS signals under complex interference environments and perform interference type detection, identification, and feature extraction. The intelligent decision control module is used to input the interference type detection and identification results and the extracted feature vectors into a pre-trained lightweight neural network model, and output the current optimal anti-interference algorithm; among them, the anti-interference algorithm types include FDPB, TDPB and adaptive filters; The multi-algorithm execution module contains three independent processing paths, each matched with an anti-interference algorithm, which is used to call the path corresponding to the current best anti-interference algorithm to perform GNSS signal interference suppression; The unbiased anti-interference processing module is used to compensate for the fixed delay caused by interference type detection and identification; it is also used to model and analyze the processing delay of each path, identify the path with the maximum delay as a reference, calculate the relative delay difference of the remaining paths, and after aligning the delay of each path based on the delay difference, it uniformly compensates for the dynamic delay caused by interference suppression; and the delay compensation adopts a joint compensation mechanism of integer buffer and fractional delay filtering to align the output signal of each path to the same time reference. The result feedback output module is used to post-process the GNSS signal after interference suppression and unbiased anti-interference processing, including reception, acquisition, tracking, ranging error feedback, and visualization of ranging error and recognition accuracy.

[0007] Furthermore, the interference type detection and identification results are single-tone interference, pulse interference, frequency sweep interference, or narrowband interference; the feature vector includes carrier-to-noise ratio, Doppler frequency shift, historical ranging error sequence, and interference intensity.

[0008] Furthermore, the interference type detection and identification results and feature vectors are normalized and input into a pre-trained lightweight neural network model. Based on the pre-constructed interference type-optimal algorithm mapping table, the model decides the probability distribution of different anti-interference algorithms and selects the algorithm with the highest probability as the current optimal anti-interference algorithm. The interference type-optimal algorithm mapping table is constructed through Monte Carlo simulation.

[0009] Furthermore, the unbiased anti-interference processing module compensates for the fixed delay caused by interference type detection and identification, including: All samples from the GNSS standard test set are sequentially input into the network model for interference type detection and identification, and the single inference delay consumed by each sample from input to output is recorded. ,in, The sample number; All collected Cleaning is performed, including removing outliers using the 3σ principle; The single inference latency dataset constructed after cleaning Calculate the first statistical mean and will As a system constant, it compensates for the fixed delay caused by interference type detection and identification; among which, This represents the number of samples in the GNSS standard test set.

[0010] Furthermore, the unbiased anti-interference processing module models and analyzes the processing latency of each path, including: The path includes the FDPB path, the TDPB path, and the adaptive filter path. The FDPB path transforms the received signal using FFT to the frequency domain, detects frequency bands with abnormal energy in the frequency domain and sets them to zero, and finally restores the processed frequency domain signal to the time domain using IFFT, thus completing interference suppression. The TDPB path directly sets the time domain sampling points in the received signal that exceed the threshold to zero, thus completing interference suppression. The adaptive filter path uses the LMS algorithm to achieve interference suppression by iteratively updating the filter weights. Analyze the processing latency of the FDPB path Represented as: ; in, The FDPB buffer latency is equal to the duration of one data frame. For FFT points, Sampling rate; The total computation time required to perform FFT and IFFT The total computational cost of performing FFT and IFFT The number of floating-point numbers that can be processed per second; To set the frequency band to zero delay and ; The processing delay of the TDPB path is analyzed as a single sampling period, expressed as follows: It can be approximated as zero delay; The analysis of the processing delay of the adaptive filter path includes: performing correlation processing on the signal output by the adaptive filter against interference and the local signal to obtain the correlation function. Represented as: ; in, For an ideal symmetric correlation function, For the time delay variable of related operations; The distortion component introduced by weight updates is introduced by the weight iterative update mechanism of the LMS algorithm; The distortion coefficient is the step size factor of the LMS algorithm. Positive correlation; based on correlation function The distortion characteristics are analyzed, and it is found that the processing delay of the adaptive filter path is a manifestation of the one-beat delay caused by the weight iteration update in the correlation domain. Its magnitude is affected by both the LMS algorithm parameters and the scene conditions.

[0011] Furthermore, the unbiased anti-interference processing module identifies the path with the maximum delay as a reference, calculates the relative delay difference of the remaining paths, and performs delay alignment of each path based on the delay difference, including: Based on the processing delay analysis of the adaptive filter path, a local signal preprocessing mechanism is introduced to eliminate the correlation function distortion caused by the adaptive filter. This mechanism involves preprocessing the local signal to ensure that the coefficient of the principal component of the correlation function in the adaptive filter's anti-interference output is 1, while the coefficients of the correlation function components at other different delay times are 0. This eliminates other delay components besides the principal component of the correlation function, while leaving the remaining principal component unchanged. Furthermore, it eliminates the correlation function superposition effect caused by iteration during the adaptive filter's anti-interference process, thus reducing the processing delay of the adaptive filter path. It can be considered as 0; where the preprocessing of the local signal is implemented in the tracking phase, it will not introduce additional processing delay; Secondly, the processing delay based on the adaptive filter path FDPB path processing latency Processing latency of TDPB path Analysis clarifies the processing latency of the FDPB path. For maximum delay That is, the FDPB path is the path with the maximum delay, and will As a reference benchmark for delay alignment; Calculate the processing delay difference of each adaptive filter path compared to the reference standard. The processing latency difference of the TDPB path compared to the reference baseline , respectively represented as: ; ; Finally, time delay alignment is performed on the output signals from the FDPB path, TDPB path, and adaptive filter path based on the time delay difference; among them, a time delay alignment compensation value is applied to the FDPB path. Apply time delay alignment compensation value to the adaptive filter path. Apply delay alignment compensation value to the TDPB path. .

[0012] Furthermore, after aligning the delays of each path based on the delay difference, the unbiased anti-interference processing module then uniformly compensates for the dynamic delay caused by interference suppression, including: Since the processing delay of the FDPB path is used as the reference benchmark for delay alignment, a statistical FDPB runtime delay strategy is adopted to compensate for the dynamic delay caused by interference suppression. The compensation method is as follows: An interference signal dataset containing single-tone interference, narrowband interference, impulse interference, and frequency sweep interference is constructed and sequentially input into the FDPB algorithm. Ignoring anti-interference performance, the computation delay of each sample after passing through the FDPB algorithm is recorded. ,in, The sample number; All collected Cleaning is performed, including removing outliers using the 3σ principle; The latency dataset was calculated using the FDPB algorithm after cleaning. Calculate the second statistical mean and will As a system constant, it is used to compensate for dynamic delays caused by interference suppression; among which, This represents the number of samples in the interference signal dataset.

[0013] Furthermore, the unbiased anti-interference processing module employs a joint compensation mechanism of integer buffering and fractional delay filtering during delay compensation to align the output signals of each path to the same time base, including: For the delay portion that is an integer multiple of the sampling period, a FIFO buffer queue is used for compensation; for the residual fractional delay, a third-order Farrow fractional delay filter is designed for precise compensation; after compensation, the output time of each path's output signal is aligned with the initial time of the received signal. .

[0014] A GNSS signal interference suppression method based on multi-algorithm time delay alignment, the method being implemented using the aforementioned GNSS signal interference suppression system based on multi-algorithm time delay alignment, includes the following steps: The data acquisition and preprocessing module acquires GNSS signals under complex interference environments and performs interference type detection, identification, and feature extraction. The intelligent decision control module inputs the interference type detection and identification results and the extracted feature vectors into a pre-trained lightweight neural network model, and outputs the current optimal anti-interference algorithm. Among them, the anti-interference algorithm types include FDPB, TDPB and adaptive filters. The system calls a multi-algorithm execution module containing three independent processing paths, each path matching an anti-interference algorithm. The multi-algorithm execution module then calls the path corresponding to the current optimal anti-interference algorithm to perform GNSS signal interference suppression. The fixed delay caused by interference type detection and identification is compensated by an unbiased anti-interference processing module; the processing delay of each path is modeled and analyzed, the path with the maximum delay is identified as a reference benchmark, the relative delay difference of the remaining paths is calculated, and after the delay of each path is aligned based on the delay difference, the dynamic delay caused by interference suppression is compensated uniformly; and a joint compensation mechanism of integer buffer and fractional delay filtering is adopted during delay compensation to align the output signals of each path to the same time benchmark. The result feedback output module performs post-processing on the GNSS signal after interference suppression and unbiased anti-interference processing, including reception, acquisition, tracking, ranging error feedback, and visualization of ranging error and recognition accuracy.

[0015] The aforementioned GNSS signal interference suppression system and method based on multi-algorithm delay alignment follows a workflow sequence of "intelligent decision-making - parallel processing - delay correction - fusion output." The data acquisition and preprocessing module is responsible for receiving the raw signal, detecting and identifying interference types, and extracting features. The intelligent decision control module outputs the optimal anti-interference algorithm, and the multi-algorithm execution module processes GNSS signal interference in parallel through three independent processing paths: FDPB, TDPB, and adaptive filters. This overcomes the limitation of traditional interference suppression methods where a single algorithm can only adapt to specific interference patterns, significantly improving interference suppression performance in complex electromagnetic environments. Adaptability: Effectively adapts to time-varying interference scenarios; through the unbiased anti-interference processing module, time delay modeling analysis, benchmark determination, and compensation operations are completed. This effectively compensates for the fixed time delay caused by interference type detection and identification, as well as the dynamic processing time delay of different anti-interference algorithms. This ensures that all processing paths have a consistent time benchmark, avoiding pseudorange abrupt changes caused by switching anti-interference algorithms. Finally, the GNSS signal after interference suppression and unbiased anti-interference processing is input to the result feedback output module for post-processing, ensuring stable and reliable output. This effectively reduces pseudorange measurement deviation in satellite signal processing and significantly improves the accuracy, stability, and robustness of navigation signal reception. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the structure of a GNSS signal interference suppression system based on multi-algorithm delay alignment; Figure 2 Timing diagram for GNSS signal slice processing; Figure 3 This is a diagram of the adaptive filter structure. Figure 4 The delay graph is an integer multiple of the FIFO buffer queue. Figure 5 Fractional delay plot of a third-order Farrow structure fractional delay filter; Figure 6 The diagram shows the code phase result after interference suppression. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0018] In one embodiment, such as Figure 1As shown, a GNSS signal interference suppression system based on multi-algorithm time delay alignment is provided. This system consists of a data acquisition and preprocessing module, an intelligent decision control module, a multi-algorithm execution module, an unbiased anti-interference processing module, and a result feedback output module. The data acquisition and preprocessing module acquires GNSS signals under complex interference environments and performs interference type detection, identification, and feature extraction. The intelligent decision control module inputs the interference type detection and identification results and extracted feature vectors into a pre-trained lightweight neural network model, and outputs the current optimal anti-interference algorithm. The anti-interference algorithm types include FDPB, TDPB, and adaptive filters. The multi-algorithm execution module contains three independent processing paths, each matched with an anti-interference algorithm, used to call the path corresponding to the current optimal anti-interference algorithm to execute GNSS signal interference suppression. The unbiased anti-interference processing module compensates for interference... The system detects and identifies fixed delays caused by interference types; it also models and analyzes the processing delays of each path, identifies the path with the maximum delay as a reference, calculates the relative delay difference of the remaining paths, and after aligning the delays of each path based on the delay difference, it uniformly compensates for the dynamic delays caused by interference suppression; and the delay compensation adopts a joint compensation mechanism of integer buffering and fractional delay filtering to align the output signals of each path to the same time reference; the result feedback output module is used to post-process the GNSS signals after interference suppression and unbiased anti-interference processing, including feedback of reception, acquisition, tracking, ranging errors, and visualization of ranging errors and recognition accuracy.

[0019] The aforementioned GNSS signal interference suppression system based on multi-algorithm delay alignment adopts a modular design. The modules transmit data streams and timing signals through standardized interfaces to ensure clear logic and functional decoupling. The overall architecture follows a process sequence of "intelligent decision-making - parallel processing - time delay correction - fusion output". The data acquisition and preprocessing module is responsible for receiving the raw signal, detecting and identifying the interference type, and extracting features. The intelligent decision control module outputs the optimal anti-interference algorithm, and combined with the multi-algorithm execution module, it processes GNSS signal interference suppression in parallel through three independent processing paths: FDPB, TDPB, and adaptive filter. This breaks through the limitation of traditional interference suppression methods where a single algorithm can only adapt to specific interference patterns, greatly improving the adaptability of interference suppression in complex electromagnetic environments and effectively adapting to time-varying interference scenarios. The unbiased anti-interference processing module completes time delay modeling analysis, benchmark determination, and compensation operations, which can effectively compensate for the fixed time delay caused by interference type detection and identification and the dynamic processing time delay of different anti-interference algorithms. This ensures that all processing paths have a consistent time benchmark and avoids pseudorange abrupt changes caused by switching anti-interference algorithms. Finally, the GNSS signal after interference suppression and unbiased anti-interference processing is input to the result feedback output module for post-processing, ensuring stable and reliable output. This effectively reduces pseudorange measurement deviation in satellite signal processing and significantly improves the accuracy, stability, and robustness of navigation signal reception.

[0020] Furthermore, the interference type detection and identification results are classified as single-tone interference, impulse interference, frequency sweep interference (which can be divided into fast frequency sweep and slow frequency sweep), or narrowband interference; the feature vector includes carrier-to-noise ratio, Doppler frequency shift, historical ranging error sequence, and interference intensity, etc. Among them, interference type detection and identification can be implemented based on deep learning network models, such as deep convolutional neural networks, residual networks (ResNet) and their variants, etc.

[0021] Furthermore, the interference type detection and identification results and feature vectors are normalized and then input into a pre-trained lightweight neural network model. Based on the pre-built interference type-optimal algorithm mapping table, the model decides the probability distribution of different anti-interference algorithms and selects the algorithm with the highest probability as the current optimal anti-interference algorithm. The interference type-optimal algorithm mapping table is constructed through a large number of Monte Carlo simulations.

[0022] Specifically, the lightweight neural network model preferably uses a simplified version of XGBoost or ResNet-18. The ResNet-18 configuration is as follows: input convolutional layer (7×7 kernel), max pooling layer, four sets of residual blocks (two basic residual blocks per set), global average pooling layer, and fully connected layer. During training, an offline simulation-generated dataset is used, containing four types of interference: single tone, impulse, frequency sweep, and narrowband, as well as interference-free data, with 10,000 samples per class, for a total training set size of 50,000. The model output is the probability distribution of the three action classes: , , The algorithm corresponding to the highest probability is selected as the current optimal choice. After the model is trained, the parameters are fixed and deployed on edge devices to achieve millisecond-level inference response.

[0023] analyze Figure 2 As shown in the GNSS signal slice processing timing diagram, the main processed signals are the interference-free signals in each slice timing diagram. Figure 2 As can be seen, to achieve time base alignment, it is necessary to perform time delay compensation on the interference detection and recognition and suppression algorithm parts respectively.

[0024] Based on this, the unbiased anti-interference processing module compensates for the fixed delay caused by interference type detection and identification, including: Considering that the fixed time delay is determined by the neural network parameters and model structure of the interference type detection and recognition, all samples in the GNSS standard test set are sequentially input into the network model for interference type detection and recognition, and the single inference time delay consumed by each sample from input to output is recorded. ,in, The sample number; All collected Cleaning is performed, including using the 3σ principle to remove obvious outliers caused by external factors such as system scheduling and cache misses; The single inference latency dataset constructed after cleaning Calculate the first statistical mean and will As a system constant, it compensates for the fixed delay caused by interference type detection and identification; among which, This represents the number of samples in the GNSS standard test set.

[0025] The unbiased anti-interference processing module models and analyzes the processing delays of the FDPB path, TDPB path, and adaptive filter path, including: (1) FDPB path: Used to transform the received signal to the frequency domain using FFT, then detect frequency bands with abnormal energy in the frequency domain and set them to zero, and finally restore the processed frequency domain signal to the time domain using IFFT, thus completing interference suppression. Since GNSS spread spectrum signals are flat and dispersed, while signals with small bandwidth such as pulse signals and narrowband interference appear as concentrated energy peaks in the frequency domain, FDPB can accurately eliminate interference signals by utilizing this characteristic. The specific calculation process of FDPB is as follows: First, the signal is transformed to the frequency domain using the Discrete Fourier Transform (DFT): ; It is a signal The spectrum after conversion to the frequency domain D It is the length of the DFT operation. k For frequency domain indexing, n This is a time-domain index. At this point, if the interference is sparse in the frequency domain, it will exhibit a pulse-like distribution. Interference can be mitigated by setting a threshold and removing spectral lines with amplitudes exceeding the threshold. However, setting the threshold has always been a challenge in engineering implementation. This method requires prior information such as receiver noise power and is difficult to adapt to changes in receiver noise background. Therefore, this project adopts a more robust method, which normalizes and suppresses the amplitude of strong interference pulses to the same level as the signal, as shown in the following equation: ; This method eliminates the amplitude advantage of the interference pulse, and the signal is recovered in subsequent correlation processing using the spreading gain of the GNSS signal. Finally, the signal is restored to the time domain using an inverse Fourier transform. .

[0026] Based on the above-mentioned basic principles of FDPB, the processing latency of the FDPB path is analyzed. Represented as: ; in, The FDPB buffer latency is equal to the duration of one data frame. For FFT points, Sampling rate; The total computation time required to perform FFT and IFFT The total computational cost of performing FFT and IFFT is as follows: The computational complexity of FFT is: ; Floating-point numbers that can be processed per second (FLOPs); To set the frequency band to zero delay and .

[0027] (2) TDPB path: used to directly zero out time-domain sampling points in the received signal that exceed the threshold, thus completing interference suppression. It can effectively suppress high-power, short-time interference. This process can be described as follows: .

[0028] Since processing can be performed without caching complete frames, the processing latency of the TDPB path is analyzed as a single sampling period, expressed as... It can be approximated as zero latency.

[0029] (3) Adaptive filter path: used to achieve interference suppression by iteratively updating the filter weights using the LMS algorithm. The adaptive filter structure is as follows: Figure 3As shown, the filter output at the current moment It can be represented as: ; In the formula, yes The input vector of the time-matter filter. Is the filter in The weight vector at time.

[0030] The weight update iteration formula for the LMS algorithm is as follows: ; superscript Indicates conjugate. Step size factor; The output after interference suppression can be expressed as: ; The superscript H indicates conjugate transpose; Substituting the weight update iteration formula, we get: .

[0031] Because the weight iteration process introduces a dynamic delay, its equivalent group delay changes with the convergence state. Furthermore, the analysis of the processing delay of the adaptive filter path includes: performing correlation processing on the signal from the adaptive filter's anti-interference output and the local signal to obtain the correlation function. Represented as: ; in, For an ideal symmetric correlation function, For the time delay variable of related operations; The distortion component introduced by weight update is introduced by the weight iterative update mechanism of the LMS algorithm. This distortion destroys the correlation peak symmetry, resulting in ranging errors that can reach the level of several nanoseconds. The distortion coefficient is the step size factor of the LMS algorithm. Positive correlation; based on correlation function The distortion characteristics are analyzed, and it is found that the processing delay of the adaptive filter path is a manifestation of the one-beat delay caused by the weight iteration update in the correlation domain. Its magnitude is affected by both the LMS algorithm parameters and the scene conditions.

[0032] The unbiased anti-interference processing module identifies the path with the maximum delay as a reference, calculates the relative delay difference of the remaining paths, and performs delay alignment of each path based on the delay difference, including: Based on the processing delay analysis of the adaptive filter path, a local signal preprocessing mechanism is introduced to eliminate the correlation function distortion caused by the adaptive filter. This mechanism involves preprocessing the local signal to ensure that the coefficient of the principal component of the correlation function in the adaptive filter's anti-interference output is 1, while the coefficients of the correlation function components at other different delay times are 0. This eliminates other delay components besides the principal component of the correlation function, while leaving the remaining principal component unchanged. Furthermore, it eliminates the correlation function superposition effect caused by iteration during the adaptive filter's anti-interference process, thus reducing the processing delay of the adaptive filter path. This can be considered as 0; the preprocessing of the local signal is implemented during the tracking phase and does not introduce additional processing delay. The specific implementation steps are as follows: Local signal output through preprocessing filter Expressed as: ; in, For the preprocessed filter coefficient vector, Generate signals locally. The order of the local filter. For preprocessing filter coefficients, This represents the order of the adaptive filter.

[0033] Next, the navigation signal after the adaptive filter is correlated with the preprocessed local signal, and the correlation function is expressed as: ; Due to the conjugate symmetry of the adaptive filter, the correlation function is substituted into... middle: ; in, These are the weighting coefficients of the navigation signal after passing through an adaptive filter. For input signals; Further calculations of the above-mentioned functions yield the following: ; Further considering the case where the order of the filter preprocessing the local pseudocode is equal to the order of the adaptive filter, i.e., 2... L +1=2 M +1, further simplification yields: ; When the order of the preprocessing filter for the local pseudocode is less than the order of the adaptive filter, i.e., 2... L +1<2 M +1, further simplification yields: ; Since the increased distortion of the correlation peak is due to the superposition effect of the correlation functions at different times after the signal passes through the adaptive filter, the above equation uses preprocessing of the local signal to make the principal components of the correlation function more uniform. The coefficient before is 1, and the rest are different delay times. The coefficient before is 0, that is: ; right The solution is given in matrix form as follows: ; The above formula ensures that the relevant functions are accurate. Principal components It does not change and can eliminate the superposition effect of correlation functions caused by iteration in the anti-interference process, thus ensuring robust ranging of correlation peaks in the acquisition and tracking stage.

[0034] Secondly, the processing delay based on the adaptive filter path FDPB path processing latency Processing latency of TDPB path Analysis clarifies the processing latency of the FDPB path. For maximum delay That is, the FDPB path is the path with the maximum delay, and will As a reference benchmark for time delay alignment.

[0035] Calculate the processing delay difference of each adaptive filter path compared to the reference standard. The processing latency difference of the TDPB path compared to the reference baseline , respectively represented as: ; .

[0036] Finally, time delay alignment is performed on the output signals from the FDPB path, TDPB path, and adaptive filter path based on the time delay difference; among them, a time delay alignment compensation value is applied to the FDPB path. Apply time delay alignment compensation value to the adaptive filter path. Apply delay alignment compensation value to the TDPB path. .

[0037] The unbiased anti-interference processing module aligns the delays of each path based on the delay difference, and then uniformly compensates for the dynamic delay caused by interference suppression, including: Since the processing delay of the FDPB path is used as the reference benchmark for delay alignment, a statistical FDPB runtime delay strategy is adopted to compensate for the dynamic delay caused by interference suppression. The compensation method is as follows: An interference signal dataset containing single-tone interference, narrowband interference, impulse interference, and frequency sweep interference is constructed and sequentially input into the FDPB algorithm. Ignoring anti-interference performance, the computation delay of each sample after passing through the FDPB algorithm is recorded. ,in, The sample number; All collected Cleaning is performed, including removing outliers using the 3σ principle; The latency dataset was calculated using the FDPB algorithm after cleaning. Calculate the second statistical mean and will As a system constant, it is used to compensate for dynamic delays caused by interference suppression; among which, This represents the number of samples in the interference signal dataset.

[0038] The unbiased anti-interference processing module employs a joint compensation mechanism of integer buffering and fractional delay filtering for time delay compensation, aligning the output signals of each path to the same time base. This includes: using a FIFO (First-In-First-Out) buffer queue for compensation of delays that are integer multiples of the sampling period; and designing a third-order Farrow structure fractional delay filter for accurate compensation of residual fractional delays. After compensation, the output time of each path's output signal is aligned with the initial time of the received signal. .

[0039] In essence, a FIFO is a serial data buffer queue. In digital signal processing systems, it is used as a purely hardware / logic-implementable, zero-distortion integer sample-point delay unit. In this application, it is implemented using a shift register, with the specific structure as follows... Figure 4 As shown. A third-order Farrow structure fractional delay filter is shown below. Figure 5 As shown, each sub-filter These are all FIR (Finite Length Impulse Response) filters. This structure supports sub-nanosecond delay adjustment, ensuring strict synchronization of all path output signals and guaranteeing main peak alignment.

[0040] In one embodiment, a GNSS signal interference suppression method based on multi-algorithm time delay alignment is provided. This method is implemented using the aforementioned GNSS signal interference suppression system based on multi-algorithm time delay alignment, and includes the following steps: The data acquisition and preprocessing module acquires GNSS signals under complex interference environments and performs interference type detection, identification, and feature extraction. The intelligent decision control module inputs the interference type detection and identification results and the extracted feature vectors into a pre-trained lightweight neural network model, and outputs the current optimal anti-interference algorithm. Among them, the anti-interference algorithm types include FDPB, TDPB and adaptive filters. The system calls a multi-algorithm execution module containing three independent processing paths, each path matching an anti-interference algorithm. The multi-algorithm execution module then calls the path corresponding to the current optimal anti-interference algorithm to perform GNSS signal interference suppression. The fixed delay caused by interference type detection and identification is compensated by an unbiased anti-interference processing module; the processing delay of each path is modeled and analyzed, the path with the maximum delay is identified as a reference benchmark, the relative delay difference of the remaining paths is calculated, and after the delay of each path is aligned based on the delay difference, the dynamic delay caused by interference suppression is compensated uniformly; and a joint compensation mechanism of integer buffer and fractional delay filtering is adopted during delay compensation to align the output signals of each path to the same time benchmark. The result feedback output module performs post-processing on the GNSS signal after interference suppression and unbiased anti-interference processing, including reception, acquisition, tracking, ranging error feedback, and visualization of ranging error and recognition accuracy.

[0041] Furthermore, taking the B3I signal as an example, a MATLAB simulation platform was built to verify the effectiveness of the proposed method. During verification, a simulation environment was constructed containing four types of interference: single-tone, pulse, frequency sweep, and narrowband, with the interference-to-signal ratio randomly generated within the range of 40–70 dB. The ranging performance of the proposed method was compared with that of traditional switching methods, and the ranging deviation was calculated to obtain the code phase result after interference suppression, as shown below. Figure 6 As shown. Figure 6 It can be seen that when switching interference types, the method proposed in this application can always maintain the pseudorange measurement deviation within 0.3 ns. In contrast, traditional switching methods always exhibit a varying ranging error when interference changes. This application achieves dynamic matching of the optimal ranging algorithm through intelligent decision control and completes unified compensation for multi-path delays by combining multi-algorithm delay alignment. This significantly improves the adaptability and ranging accuracy of GNSS systems in complex electromagnetic environments. This technology meets the stringent requirements of stability and continuity for high-precision positioning systems and shows promising application prospects in autonomous positioning of low-Earth orbit satellite receivers.

[0042] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0043] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application.

Claims

1. A GNSS signal interference suppression system based on multi-algorithm time delay alignment, characterized in that, The system includes: The data acquisition and preprocessing module is used to acquire GNSS signals under complex interference environments and perform interference type detection, identification, and feature extraction. The intelligent decision control module is used to input the interference type detection and identification results and the extracted feature vectors into a pre-trained lightweight neural network model, and output the current optimal anti-interference algorithm; among them, the anti-interference algorithm types include FDPB, TDPB and adaptive filters; The multi-algorithm execution module contains three independent processing paths, each matched with an anti-interference algorithm, which is used to call the path corresponding to the current best anti-interference algorithm to perform GNSS signal interference suppression; The unbiased anti-interference processing module is used to compensate for the fixed delay caused by interference type detection and identification; it is also used to model and analyze the processing delay of each path, identify the path with the maximum delay as a reference, calculate the relative delay difference of the remaining paths, and after aligning the delay of each path based on the delay difference, it uniformly compensates for the dynamic delay caused by interference suppression; and the delay compensation adopts a joint compensation mechanism of integer buffer and fractional delay filtering to align the output signal of each path to the same time reference. The result feedback output module is used to post-process the GNSS signal after interference suppression and unbiased anti-interference processing, including reception, acquisition, tracking, ranging error feedback, and visualization of ranging error and recognition accuracy.

2. The GNSS signal interference suppression system based on multi-algorithm time delay alignment according to claim 1, characterized in that, The interference type detection and identification results are single-tone interference, pulse interference, frequency sweep interference, or narrowband interference; the feature vector includes carrier-to-noise ratio, Doppler frequency shift, historical ranging error sequence, and interference intensity.

3. The GNSS signal interference suppression system based on multi-algorithm time delay alignment according to claim 2, characterized in that, The interference type detection and identification results and the feature vector are normalized and then input into a pre-trained lightweight neural network model. The model outputs the probability distribution of different anti-interference algorithms based on a pre-constructed interference type-optimal algorithm mapping table, and selects the algorithm with the highest probability as the current optimal anti-interference algorithm. The interference type-optimal algorithm mapping table is constructed through Monte Carlo simulation.

4. The GNSS signal interference suppression system based on multi-algorithm time delay alignment according to claim 1, characterized in that, The unbiased anti-interference processing module compensates for the fixed delay caused by interference type detection and identification, including: All samples from the GNSS standard test set are sequentially input into the network model for interference type detection and identification, and the single inference delay consumed by each sample from input to output is recorded. ,in, The sample number; All collected Cleaning is performed, including removing outliers using the 3σ principle; The single inference latency dataset constructed after cleaning Calculate the first statistical mean and will As a system constant, it compensates for the fixed delay caused by interference type detection and identification; among which, This represents the number of samples in the GNSS standard test set.

5. The GNSS signal interference suppression system based on multi-algorithm time delay alignment according to claim 4, characterized in that, The unbiased anti-interference processing module models and analyzes the processing delay of each path, including: The path includes an FDPB path, a TDPB path, and an adaptive filter path. The FDPB path transforms the received signal using FFT to the frequency domain, detects frequency bands with abnormal energy in the frequency domain and sets them to zero, and finally restores the processed frequency domain signal to the time domain using IFFT, thus completing interference suppression. The TDPB path directly sets the time-domain sampling points in the received signal that exceed a threshold to zero, thus completing interference suppression. The adaptive filter path uses the LMS algorithm to achieve interference suppression by iteratively updating the filter weights. Analyze the processing latency of the FDPB path Represented as: ； in, The FDPB buffer latency is equal to the duration of one data frame. For FFT points, Sampling rate; The total computation time required to perform FFT and IFFT The total computational cost of performing FFT and IFFT The number of floating-point numbers that can be processed per second; To set the frequency band to zero delay and ; The processing delay of the TDPB path is analyzed as a single sampling period, expressed as follows: It can be approximated as zero delay; The analysis of the processing delay of the adaptive filter path includes: performing correlation processing on the signal output by the adaptive filter against interference and the local signal to obtain the correlation function. Represented as: ； in, For an ideal symmetric correlation function, For the time delay variable of related operations; The distortion component introduced by weight updates is introduced by the weight iterative update mechanism of the LMS algorithm; The distortion coefficient is the step size factor of the LMS algorithm. Positive correlation; based on correlation function The distortion characteristics are analyzed, and it is clear that the processing delay of the adaptive filter path is a manifestation of the one-beat delay caused by the weight iteration update in the correlation domain. Its magnitude is affected by both the LMS algorithm parameters and the scene conditions.

6. The GNSS signal interference suppression system based on multi-algorithm time delay alignment according to claim 5, characterized in that, The unbiased anti-interference processing module identifies the path with the maximum delay as a reference, calculates the relative delay difference of the remaining paths, and performs delay alignment of each path based on the delay difference, including: Based on the processing delay analysis of the adaptive filter path, a local signal preprocessing mechanism is introduced to eliminate the correlation function distortion caused by the adaptive filter. This mechanism involves preprocessing the local signal to ensure that the coefficient of the principal component of the correlation function in the adaptive filter's anti-interference output is 1, while the coefficients of the correlation function components at other different delay times are 0. This eliminates other delay components besides the principal component of the correlation function, while leaving the remaining principal component unchanged. Furthermore, it eliminates the correlation function superposition effect caused by iteration during the adaptive filter's anti-interference process, thus reducing the processing delay of the adaptive filter path. It can be considered as 0; where the preprocessing of the local signal is implemented in the tracking phase, it will not introduce additional processing delay; Secondly, the processing delay based on the adaptive filter path FDPB path processing latency Processing latency of TDPB path Analysis clarifies the processing latency of the FDPB path. For maximum delay That is, the FDPB path is the path with the maximum delay, and will As a reference benchmark for delay alignment; Calculate the processing delay difference of each adaptive filter path compared to the reference standard. The processing latency difference of the TDPB path compared to the reference baseline , respectively represented as: ； ； Finally, time delay alignment is performed on the output signals from the FDPB path, TDPB path, and adaptive filter path based on the time delay difference; among them, a time delay alignment compensation value is applied to the FDPB path. Apply time delay alignment compensation value to the adaptive filter path. Apply delay alignment compensation value to the TDPB path. .

7. The GNSS signal interference suppression system based on multi-algorithm time delay alignment according to claim 6, characterized in that, The unbiased anti-interference processing module, after aligning the delays of each path based on the delay difference, then uniformly compensates for the dynamic delay caused by interference suppression, including: Since the processing delay of the FDPB path is used as the reference benchmark for delay alignment, a statistical FDPB runtime delay strategy is adopted to compensate for the dynamic delay caused by interference suppression. The compensation method is as follows: An interference signal dataset containing single-tone interference, narrowband interference, impulse interference, and frequency sweep interference is constructed and sequentially input into the FDPB algorithm. Ignoring anti-interference performance, the computation delay of each sample after passing through the FDPB algorithm is recorded. ,in, The sample number; All collected Cleaning is performed, including removing outliers using the 3σ principle; The latency dataset was calculated using the FDPB algorithm after cleaning. Calculate the second statistical mean and will As a system constant, it is used to compensate for dynamic delays caused by interference suppression; among which, This represents the number of samples in the interference signal dataset.

8. The GNSS signal interference suppression system based on multi-algorithm time delay alignment according to claim 7, characterized in that, The unbiased anti-interference processing module employs a joint compensation mechanism of integer buffering and fractional delay filtering during time delay compensation to align the output signals of each path to the same time base, including: For the delay portion that is an integer multiple of the sampling period, a FIFO buffer queue is used for compensation; for the residual fractional delay, a third-order Farrow fractional delay filter is designed for precise compensation; after compensation, the output time of each path's output signal is aligned with the initial time of the received signal. .

9. A GNSS signal interference suppression method based on multi-algorithm time delay alignment, characterized in that, The method is implemented using the GNSS signal interference suppression system based on multi-algorithm time delay alignment as described in any one of claims 1-7, and includes the following steps: The data acquisition and preprocessing module acquires GNSS signals under complex interference environments and performs interference type detection, identification, and feature extraction. The intelligent decision control module inputs the interference type detection and identification results and the extracted feature vectors into a pre-trained lightweight neural network model, and outputs the current optimal anti-interference algorithm. Among them, the anti-interference algorithm types include FDPB, TDPB and adaptive filters. The system calls a multi-algorithm execution module containing three independent processing paths, each path matching an anti-interference algorithm. The multi-algorithm execution module then calls the path corresponding to the current optimal anti-interference algorithm to perform GNSS signal interference suppression. The fixed delay caused by interference type detection and identification is compensated by an unbiased anti-interference processing module; the processing delay of each path is modeled and analyzed, the path with the maximum delay is identified as a reference benchmark, the relative delay difference of the remaining paths is calculated, and after the delay of each path is aligned based on the delay difference, the dynamic delay caused by interference suppression is compensated uniformly; and a joint compensation mechanism of integer buffer and fractional delay filtering is adopted during delay compensation to align the output signals of each path to the same time benchmark. The result feedback output module performs post-processing on the GNSS signal after interference suppression and unbiased anti-interference processing, including reception, acquisition, tracking, ranging error feedback, and visualization of ranging error and recognition accuracy.

Images