Weak signal-to-noise ratio space-based signal AI training sample set generation method and system

Abstract

The invention discloses a weak signal-to-noise ratio space-based signal AI training sample set generation method and system. The method comprises the steps that a method acquisition system acquires a Starlink beacon signal; performing transit satellite analysis and coarse Doppler stripping based on the Starlink beacon signal; performing correction based on the coarse Doppler stripping result to obtain a Doppler measurement value; and corresponding original data to the Doppler measurement value to obtain a training sample.

Description

Technical Field

[0001] The present application relates to data acquisition technologies, in particular to a method and system for generating an AI training sample set for space-based signals with weak signal-to-noise ratio. Background Art

[0002] The detection of space-based weak signal-to-noise ratio signals is a fundamental and core technology in fields such as communication, radar, wireless communication, and image processing. Research on it can not only provide support for the improvement of existing technologies but also promote the development of technologies to a higher level.

[0003] During the communication process, the transmission quality of signals has an important impact on the reliability of communication. By detecting weak signal-to-noise ratio signals, problems existing in the transmission can be timely discovered, and corresponding measures can be taken for repair and optimization, thereby improving the security and confidentiality of information.

[0004] In deep space exploration, due to the long distance, the signal will be greatly attenuated during transmission, and the signal reaching the Earth is very weak. Research on weak signal-to-noise ratio signal detection technology can help recover these weak radio signals, ensure the reliability of data transmission and navigation positioning, and is of great significance for exploring the universe and studying astrophysics.

[0005] The power of weak signals is extremely low (such as SNR < -50dB), and it is easily masked by noise, making it difficult for traditional detection methods (such as Fourier transform) to extract effective information. While AI technologies (such as deep learning) show potential in signal detection and recognition, the training model depends on a training sample set. Therefore, constructing an AI training sample set suitable for weak signal-to-noise ratio scenarios has become a key technical bottleneck in AI weak signal detection. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art. For this purpose, the present invention provides a method and system for generating an AI training sample set for space-based signals with weak signal-to-noise ratio, which can efficiently construct training samples.

[0007] This solution mainly studies and establishes a method for generating an AI training sample set with weak signal-to-noise ratio based on Starlink beacon signals, supporting the construction of the training sample set and the iterative optimization of the weak signal detection model.

[0008] On the one hand, an embodiment of the present application provides a method for generating an AI training sample set for space-based signals with weak signal-to-noise ratio, including:

[0009] An acquisition system acquires Starlink beacon signals;

[0010] Based on the Starlink beacon signals, transit satellite analysis and rough Doppler stripping are performed;

[0011] The Doppler measurement value is obtained after correction based on the result of the coarse Doppler stripping;

[0012] The original data is corresponded with the Doppler measurement value to obtain training samples.

[0013] In some embodiments, the Starlink beacon signal is collected by using a broadband signal acquisition system, specifically:

[0014] The Starlink beacon signal is collected by using a broadband signal acquisition system, wherein the receiving frequency of the signal is 14.25 GHz, then it is down-converted to 1.5 GHz, and then the received signal is subjected to A / D conversion with 14-bit coding and a sampling bandwidth of 250 MHz and stored in a disk, and the corresponding time of the collected data segment is recorded.

[0015] In some embodiments, the transit satellite analysis is performed based on the Starlink beacon signal and the coarse Doppler stripping is carried out;

[0016] Specifically, it includes:

[0017] The digital signal recorded in the disk is subjected to down-conversion sampling and anti-aliasing filtering to obtain the original signal with a sampling bandwidth of 25 MHz;

[0018] The transit satellite analysis is performed based on the original data and the coarse Doppler stripping is carried out.

[0019] In some embodiments, the transit satellite analysis specifically includes:

[0020] Using the starlink catalog data, analyzing and calculating the satellite transit situation, and selecting the transit satellites corresponding to the data recording period.

[0021] In some embodiments, the coarse Doppler stripping specifically includes:

[0022] For the transit satellites, using the starlink catalog data and the probability position coordinates, calculating the coarse-resolution Doppler frequency, and performing coarse Doppler stripping on the original data.

[0023] In some embodiments, the Doppler measurement value is obtained after correction based on the result of the coarse Doppler stripping, specifically including:

[0024] The result of the coarse Doppler stripping is expressed as:

[0025]

[0026] △f d is the Doppler correction amount

[0027] Perform N-point fast Fourier transform on y u (t) with a duration of T to obtain Yu (f), and then detect and obtain the Doppler correction value through the binary hypothesis testing method, and use the Doppler correction value to compensate the coarse Doppler to obtain the Doppler measurement value.

[0028] In some embodiments, the performing of the coarse Doppler stripping is specifically described as:

[0029]

[0030] Where:

[0031] λ is the wavelength;

[0032] is the unit position vector;

[0033] is the position vector, and x, y, z are the three coordinate components;

[0034] is the velocity vector, v x , v y , v z are the three coordinate components;

[0035] On the other hand, an embodiment of the present application provides a weak signal-to-noise ratio space-based signal AI training sample set generation system, including:

[0036] An acquisition unit, configured to acquire Starlink beacon signals by using a broadband signal acquisition system;

[0037] A coarse stripping unit, configured to perform transit satellite analysis and coarse Doppler stripping based on the Starlink beacon signals;

[0038] A correction unit, configured to perform correction based on the result of the coarse Doppler stripping to obtain a Doppler measurement value;

[0039] A labeling unit, configured to correspond the original data with the Doppler measurement value to obtain training samples.

[0040] On the other hand, an embodiment of the present application provides a weak signal-to-noise ratio space-based signal AI training sample set generation system, including:

[0041] A memory, configured to store programs;

[0042] A processor, configured to load the program to execute the weak signal-to-noise ratio space-based signal AI training sample set generation method.

[0043] On the other hand, an embodiment of the present application provides a computer-readable storage medium, which stores a program, and when the program is loaded by a processor, the weak signal-to-noise ratio space-based signal AI training sample set generation method is implemented.

[0044] By stripping the coarse Doppler in this application, the signal dynamic range is greatly reduced, and the bottleneck problem of long-term incoherent accumulation of signals is solved, significantly improving the weak signal detection and tracking capabilities and accurately obtaining Doppler measurement values. Using a software receiver, signal samples with real signal noise characteristics covering the entire signal-to-noise ratio range are generated. For the first time, the simplest AI signal training samples with original measurement data as input and Doppler measurement values as output are established, laying a foundation for conducting signal acquisition and tracking research based on artificial intelligence. Brief Description of the Drawings

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments.

[0046] Figure 1 is a flowchart of a method provided by an embodiment of the present application;

[0047] Figure 2 is a system block diagram provided by an embodiment of the present application. Detailed Embodiments

[0048] To make the objectives, technical solutions, and advantages of the present application clearer, the following will, with reference to the drawings in the embodiments of the present application, clearly and completely describe the technical solutions of the present application through the embodiments. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0049] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0051] It can be understood that the purpose of this step is to generate samples for training the AI model. And this solution has the following characteristics for the model to be trained. Its input method is the original data, and its output result is the Doppler measurement value.

[0052] Referring to Figure 1 , an embodiment of the present application provides a method for generating an AI training sample set for space-based signals with a weak signal-to-noise ratio, including:

[0053] S1. Use a broadband signal acquisition system to collect Starlink beacon signals. Among them, the Starlink beacon signal refers to the specific signal emitted by low-orbit satellites of Starlink during communication. Although these signals are not designed specifically for navigation, they can be utilized by opportunistic signal navigation technology to achieve positioning and timing services by extracting their physical characteristics (such as Doppler frequency shift, carrier phase, etc.).

[0054] The receiving frequency of the signal is 14.25 GHz, then it is down-converted to 1.5 GHz, and then an A / D conversion is performed on the received signal with 14-bit encoding and a 250 MHz sampling bandwidth and stored on disk, and the corresponding time of the collected data segment is recorded.

[0055] s(t) = A sin(2πf ku t + φ);

[0056] s1(t) = A sin(2π(f ku - f0)t + φ);

[0057] s1(t) = A sin(2πf l t + φ);

[0058] s1(n) = A sin(2πf l n△T + φ), n = 1, 2, 3.....

[0059] Where:

[0060] s(t) is the original signal;

[0061] s1(t) is the signal after stripping the coarse Doppler;

[0062] A is the amplitude;

[0063] f ku is the Starlink carrier signal frequency;

[0064] φ is the initial phase of the Starlink signal;

[0065] f0 is the coarse Doppler frequency;

[0066] f l is the frequency of the signal after Doppler stripping

[0067] △T is the sampling interval;

[0068] S2. Conduct transit satellite analysis and perform coarse Doppler stripping based on the Starlink beacon signal.

[0069] Given that the Starlink beacon signal bandwidth is 1 MHz, according to the Nyquist theorem, the digital signal recorded on the disk is down-converted, sampled, and anti-aliasing filtered to obtain the original signal with a sampling bandwidth of 2.5 MHz;

[0070] s1(k) = A sin(2πf l k△T + φ), k = Mn

[0071] M is the downsampling rate.

[0072] Based on the original data, analyze the passing satellites and perform coarse Doppler stripping.

[0073] The analysis of the passing satellites specifically includes:

[0074] Using the Starlink catalog data, analyze and calculate the satellite passing situation, and select the passing satellites corresponding to the data recording period.

[0075] The coarse Doppler stripping specifically includes:

[0076] For passing satellites, the Doppler change rate of the Starlink signal is relatively large, up to 3.7 kHz / s at most. Direct detection will result in a low coherent integration gain and an inability to detect the signal. Therefore, use the Starlink catalog data and the probabilistic position coordinates to calculate the coarse-resolution Doppler frequency, and perform coarse Doppler stripping on the original data to reduce the signal dynamic range so that it can be processed by long-term accumulation.

[0077]

[0078] Where:

[0079] λ is the wavelength;

[0080] is the unit position vector;

[0081] is the position vector, and x, y, z are the three coordinate components;

[0082] is the velocity vector, v x , v y , v z are the three coordinate components;

[0083] S3. Based on the result of the coarse Doppler stripping, the Doppler measurement value is obtained after correction.

[0084] Among them, the result of the coarse Doppler stripping is expressed as:

[0085]

[0086] △fd is the Doppler correction amount

[0087] Perform an N-point fast Fourier transform on y for a duration of T u (t) to obtain Y u (f), and then detect and obtain the Doppler correction value through the binary hypothesis testing method, and use the Doppler correction value to compensate for the coarse Doppler to obtain the Doppler measurement value.

[0088] S4. Correlate the original data with the Doppler measurement value to obtain a training sample.

[0089] In this step, the original data is correlated with the measured Doppler one by one to form a training sample.

[0090] It can be understood that the training sample is determined by the input data and the labeled output result. Through the training of a large number of training samples, the model can learn the potential relationship between the input and the output.

[0091] It can be understood that through the embodiment of the present application, by adopting the coarse Doppler stripping method, the problem of weak signal processing is solved, and the Doppler measurement value is accurately obtained. By using a software receiver, signal samples with real signal noise characteristics covering the entire signal-to-noise ratio range are generated, and for the first time, the simplest AI signal training sample with the original measurement data as the input and the Doppler measurement value as the output is established, laying a foundation for carrying out signal acquisition and tracking research based on artificial intelligence.

[0092] Refer to Figure 2 , the embodiment of the present application provides a weak signal-to-noise ratio space-based signal AI training sample set generation system, including:

[0093] An acquisition unit, configured to acquire Starlink beacon signals by using a broadband signal acquisition system;

[0094] A coarse stripping unit, configured to perform transit satellite analysis and coarse Doppler stripping based on the Starlink beacon signals;

[0095] A correction unit, configured to perform correction based on the result of the coarse Doppler stripping to obtain a Doppler measurement value;

[0096] A labeling unit, configured to correlate the original data with the Doppler measurement value to obtain a training sample.

[0097] Generally speaking, the working process of this system is as follows:

[0098] Data acquisition and preprocessing

[0099] 1) Data acquisition

[0100] The Starlink beacon signal is collected by a broadband signal acquisition system. The receiving frequency is 14.25 GHz, and it is down-converted to 1.5 GHz. The received signal is subjected to A / D conversion with 14-bit encoding and a sampling bandwidth of 250 MHz and stored on disk, and the corresponding time of the collected data segment is recorded.

[0101] s(t) = A sin(2πf ku t + φ)

[0102] s1(t) = A sin(2π(f ku -f0)t + φ)

[0103] s1(t) = A sin(2πf l t + φ)

[0104] s1(n) = A sin(2πf l n△T + φ), n = 1, 2, 3.....

[0105] Where:

[0106] s(t) is the original signal;

[0107] s1(t) is the signal after stripping the coarse Doppler;

[0108] A is the amplitude;

[0109] f ku is the Starlink carrier signal frequency;

[0110] φ is the initial phase of the Starlink signal;

[0111] f0 is the coarse Doppler frequency;

[0112] f l is the frequency of the signal after Doppler stripping

[0113] △T is the sampling interval;

[0114] 2) Downsampling and anti-aliasing filtering

[0115] The digital signal stored on disk is subjected to down-conversion sampling and anti-aliasing filtering to obtain the original signal with a sampling bandwidth of 25 MHz.

[0116] s1(k) = A sin(2πf l k△T + φ), k = Mn

[0117] M is the downsampling rate

[0118] 3.2 Satellite transit analysis and coarse Doppler stripping

[0119] 1) Satellite transit analysis

[0120] Using Starlink catalog data, analyze and calculate satellite transits, and select the satellites that transit during the corresponding data recording period.

[0121] 2) Coarse Doppler stripping

[0122] For the transiting satellites, use Starlink catalog data and probabilistic position coordinates to calculate the coarse-resolution Doppler frequency, and perform coarse Doppler stripping on the original data.

[0123]

[0124] Where:

[0125] λ is the wavelength;

[0126] is the unit position vector;

[0127] is the position vector, and x, y, z are the three coordinate components;

[0128] is the velocity vector, v x , v y , v z are the three coordinate components;

[0129] 3.3 Signal processing and sample annotation

[0130] 1) Signal processing

[0131] Strip the coarse Doppler signal:

[0132]

[0133] △f d is the Doppler correction amount

[0134] Perform an N-point Fast Fourier Transform (FFT) on y u (t) for a duration of T to obtain y u (f). Then, detect and obtain the Doppler correction value through a binary hypothesis testing method, and compensate the coarse Doppler with the Doppler correction value to obtain the Doppler measurement value.

[0135] 2) Sample annotation

[0136] Correspond the original data with the measured Doppler one by one to form training samples.

[0137] The embodiment of the present application provides a weak signal-to-noise ratio space-based signal AI training sample set generation system, including:

[0138] A memory for storing programs;

[0139] A processor for loading the program to execute the method for generating the AI training sample set of space-based signals with weak signal-to-noise ratio.

[0140] An embodiment of the present application provides a computer-readable storage medium storing a program, which, when loaded by a processor, implements the method for generating the AI training sample set of space-based signals with weak signal-to-noise ratio.

[0141] Through the embodiment of the present application, by adopting the coarse Doppler stripping technology, the problem of weak signal processing is solved, and the Doppler measurement value is accurately obtained. By using a software receiver, signal samples with real signal noise characteristics covering the full signal-to-noise ratio range are generated, and the simplest AI signal training samples with the original measurement data as the input and the Doppler measurement value as the output are established for the first time, laying a foundation for carrying out research on signal acquisition and tracking based on artificial intelligence.

[0142] Note that the above is only a preferred embodiment of the present application and the applied technical principle. Those skilled in the art will understand that the present application is not limited to the specific embodiments described herein, and various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present application. Therefore, although the present application has been described in detail through the above embodiments, the present application is not limited to the above embodiments only. Without departing from the concept of the present application, more other equivalent embodiments can be included, and the scope of the present application is determined by the scope of the appended claims.

Claims

1. A method for generating an AI training sample set of space-based signals with weak signal-to-noise ratio, characterized in that, Including: Collecting Starlink beacon signals; Conducting transit satellite analysis and performing coarse Doppler stripping based on the Starlink beacon signals; Obtaining Doppler measurement values after correction based on the results of the coarse Doppler stripping; Corresponding the original data with the Doppler measurement values to obtain training samples.

2. The method for generating the AI training sample set of the space-based signal with weak signal-to-noise ratio according to claim 1, wherein The collecting of Starlink beacon signals by using a broadband signal acquisition system is specifically as follows: Collecting Starlink beacon signals by using a broadband signal acquisition system, where the receiving frequency of the signal is 14.25 GHz, then down-converting it to 1.5 GHz, and then performing A / D conversion on the received signal with 14-bit coding and a 250 MHz sampling bandwidth and storing it in a disk record, and recording the corresponding time of the collected data segment.

3. The method for generating the AI training sample set of the space-based signal with weak signal-to-noise ratio according to claim 2, wherein Conducting transit satellite analysis and performing coarse Doppler stripping based on the Starlink beacon signals; Specifically including: Performing down-conversion sampling and anti-aliasing filtering on the digital signal recorded in the disk to obtain an original signal with a sampling bandwidth of 25 MHz; Conducting transit satellite analysis and performing coarse Doppler stripping based on the original data.

4. The method for generating an AI training sample set of a space-based signal with weak signal-to-noise ratio according to claim 3, characterized in that The transit satellite analysis specifically includes: Using starlink catalog data to analyze and calculate satellite transit conditions, and selecting the transit satellites corresponding to the data recording period.

5. The method for generating the AI training sample set of the space-based signal with weak signal-to-noise ratio according to claim 3, wherein The coarse Doppler stripping specifically includes: For transit satellites, using starlink catalog data and probability position coordinates to calculate the coarse-resolution Doppler frequency, and performing coarse Doppler stripping on the original data.

6. The method for generating the AI training sample set of the space-based signal with weak signal-to-noise ratio according to claim 1, wherein Obtaining Doppler measurement values after correction based on the results of the coarse Doppler stripping specifically includes: The results of the coarse Doppler stripping are expressed as: △f d is the Doppler correction amount y for a duration of T u Perform an N-point fast Fourier transform on y(t) to obtain Y(f). u Then, detect and obtain the Doppler correction value through a binary hypothesis testing method, and use the Doppler correction value to compensate for the coarse Doppler to obtain the Doppler measurement value.

7. The method for generating the AI training sample set of the space-based signal with weak signal-to-noise ratio according to claim 1, wherein The specific description of performing the coarse Doppler stripping is: Where: λ is the wavelength; is the unit position vector; is the position vector, and x, y, and z are the three coordinate components; is the velocity vector, v x , v y , v z are the three coordinate components.

8. An AI training sample set generation system for space-based signals with weak signal-to-noise ratio, characterized in that, Including: A collecting unit for collecting Starlink beacon signals by using a broadband signal acquisition system; A coarse stripping unit for conducting transit satellite analysis and performing coarse Doppler stripping based on the Starlink beacon signals; A correction unit for obtaining Doppler measurement values after correction based on the results of the coarse Doppler stripping; A marking unit for corresponding the original data with the Doppler measurement values to obtain training samples.

9. An AI training sample set generation system for space-based signals with weak signal-to-noise ratio, characterized in that, Including: A memory for storing programs; A processor for loading the program to execute the method for generating an AI training sample set of a space-based signal with weak signal-to-noise ratio as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a program, and when the program is loaded by the processor, it implements the method for generating an AI training sample set of a space-based signal with weak signal-to-noise ratio as described in any one of claims 1-7.

Images