Low earth orbit satellite capturing method based on perception assistance and RCNN enabling
Through perception assist and RCNN technology, low-orbit satellite signal capture is optimized, which solves the problems of traditional methods in multipath interference, resource overhead and poor adaptability, and achieves efficient and accurate signal capture, suitable for low-cost mobile devices.
Patent Information
- Application Number
- CN202411401245.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-09
- Publication Date
- 2024-12-24
AI Technical Summary
Traditional satellite communication systems face problems such as multipath interference, high computing resource overhead and poor adaptability in low-orbit environments, especially in low-cost mobile receiving equipment and high-speed motion, which is difficult to achieve efficient signal capture.
Using low-orbit satellite capture method based on perception assistance and RCNN, Doppler frequency bias prior information is generated through perception assistance module, and combined with RCNN capture module, including data-driven convolutional units and recurrent neural network units, optimize signal processing and reduce calculations. Complexity and time resource overhead to improve capture accuracy.
It effectively reduces the impact of multipath interference, reduces the demand for computing resources, improves the accuracy and adaptability of signal capture, is suitable for energy-constrained mobile platforms, and improves the performance of low-orbit satellite communications.
Smart Images
Figure CN119182449A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of satellite capture technology, and in particular to a low-orbit satellite capture method based on perception assistance and RCNN empowerment. Background Art
[0002] With the rapid development of communication technology, especially the application of future 6GHz and higher frequency bands (such as terahertz), satellite communication systems play an increasingly important role in the "integrated air, land, and sea" network. Perception-assisted AI-enabled satellite communication is one of the key technologies in this field. It uses intelligent algorithms to improve the performance of communication systems, especially in signal capture, tracking, despreading and demodulation. Due to its low orbital altitude, low-orbit satellites have the advantages of shorter transmission delays and higher data transmission rates, but they also face challenges brought by complex dynamic environments.
[0003] Traditional satellite communication systems usually rely on methods based on parallel processing in the time domain or frequency domain to achieve signal capture. These methods often require sacrificing computing resources or time complexity in exchange for higher accuracy, such as using RAKE receivers to combat multipath interference. However, in some cost-sensitive application scenarios, such as low-cost mobile receiving devices, such high-complexity processing methods are not practical. In addition, due to their high-speed motion characteristics, low-orbit satellites will produce a significant Doppler effect, which makes signal capture more difficult and requires the capture algorithm to be able to quickly and accurately estimate and compensate for this effect.
[0004] Current satellite communication systems also face other problems, including but not limited to:
[0005] ① Multipath interference: In urban environments, buildings and other obstacles can cause serious multipath propagation, which can degrade signal quality and affect capture performance.
[0006] ② Resource overhead: Traditional methods usually require a lot of computing resources to achieve good capture performance, which is unacceptable for energy-constrained mobile platforms.
[0007] ③Poor adaptability: Traditional acquisition methods have difficulty adapting to changing channel conditions, especially in low-orbit satellite communications, where channel characteristics may change rapidly.
[0008] In order to solve the above problems, researchers have begun to explore the application of artificial intelligence (AI) technology in the field of satellite communications in recent years. AI technology can optimize the signal processing process by learning historical data and pattern recognition, thereby reducing the demand for computing resources while maintaining or even improving capture performance. In particular, deep learning models such as convolutional neural networks (CNN) and recurrent neural networks (RNN) have shown great potential in processing nonlinear and complex patterns. Summary of the invention
[0009] In order to overcome the shortcomings in the background technology, the present invention discloses a low-orbit satellite capture method based on perception assistance and RCNN empowerment. Compared with the existing RAKE receiver capture method based on time-frequency parallelism, the method does not use multi-path parallelism to diversity-combine the influence of multipath interference, reduces the calculation and time complexity, and uses AI to improve the capture accuracy. In particular, perception-assisted communication is used to reduce time resource overhead, which not only effectively improves the RCNN-enabled low-orbit satellite capture performance, but also reduces the influence of multipath interference on capture.
[0010] In order to achieve the above-mentioned invention object, the present invention adopts the following technical scheme:
[0011] A low-orbit satellite acquisition method based on perception assistance and RCNN empowerment, specifically:
[0012] S1. The perception auxiliary module obtains the mobile mode of the receiver based on the acquired public waveform information, and generates the prior information of Doppler frequency deviation in combination with the low-orbit satellite constellation information;
[0013] S2. According to the characteristics of the satellite downlink signal, an RCNN capture module is designed, and the RCNN capture module includes a first convolution unit based on data driving, a second convolution unit based on model driving, and multiple groups of RNN units;
[0014] S3. Based on the prior information generated by the perception-assisted communication module and the carrier Doppler interference of the compensated satellite signal, the RCNN capture module calculates the capture information.
[0015] Preferably, the common waveform information includes sparsity, geometric characteristics and signaling information of millimeter wave communications.
[0016] Preferably, the downlink signal characteristics include training sequence length, spread spectrum code length and message signal cycle length.
[0017] Preferably, in step S1, the specific steps of generating the prior information by the perception auxiliary module are:
[0018] S101. The receiver passively receives the common waveform signal y=[y0,y1,…,y L-1] T , and obtain the mobile mode information I of the receiver through the following calculation model, namely:
[0019]
[0020] vect(Y DD )=[Y DD (0,0),Y DD (1,0),…,Y DD (N-1,0),Y DD (0,1),…,Y DD (N-1,M-1)] T ;
[0021]
[0022] Wherein, L represents the number of samples of the received common waveform signal; N represents the number of discrete Fourier transform points; M is obtained by M×N=L; vect -1 is an operation to devectorize a vector array to obtain a matrix; the F N Expressed as an N-dimensional DFT matrix, represents the N-dimensional conjugate DFT matrix, Represented as an M-dimensional DFT matrix; the Y DD is represented as a delay-Doppler domain matrix; the said represents the operation of dematrixing the matrix array to obtain a vector; the said f s represents the sampling rate, and I represents the mobile mode information of the receiver, including speed and distance;
[0023] S102. Calculate the prior information of Doppler frequency deviation based on the mobile mode information and the constellation information of the low-orbit satellite. The calculation formula is as follows:
[0024]
[0025] Where v represents the satellite speed in the constellation information; f c It is represented by the radio frequency; c is represented by the speed of light, in m / s; T space is represented by the spatial sampling interval; ΔS is the distance change between the satellite and the receiver, and n / f s Used to construct the receiver position.
[0026] Preferably, the first convolution unit determines the length L of the spread spectrum code or training sequence in the downlink signal characteristics sent by the satellite. seq , and the message signal period L msg Drive the data to get the receptive field parameters of the first convolution unit, namely:
[0027]
[0028] Among them, K conv1 It is represented by the receptive field size of the first convolution unit’s first layer of convolution; P avg is the average pooling layer parameter between the first and second convolution layers, 2 is because the signal of the input convolution module is the concatenation of the real and imaginary parts; the K conv2 is the receptive field size of the second layer convolution module.
[0029] Preferably, the second convolution unit is based on the spread spectrum code or training sequence in the satellite downlink signal characteristics and message signals The driving model is calculated, and the specific calculation formula is:
[0030]
[0031] The Γ l ,l=0,1,…,|L seq -L msg |-1 is the input of the second convolution unit,
[0032] According to the driving model, |L seq -L msg |, the length of the spreading code or training sequence L seq and the message signal period L msg , and obtain the receptive field parameters of the second convolutional unit, namely:
[0033]
[0034] The K conv3 K is the receptive field size of the first layer of the second convolutional unit; conv4 Represents the receptive field size of the second layer convolutional unit.
[0035] Preferably, the RNN unit is based on the output L of the first convolution unit mod ×1 Dimension and the output L of the second convolutional unit data ×1 Dimension Design the input amount of the fusion input layer, namely:
[0036]
[0037] According to the input r of the fusion input layer in,t , and combined with the satellite signal quality and satellite signal characteristics, the state tensor number t of the RNN unit is designed max , that is:
[0038]
[0039] The C / N0 represents the receiving carrier-to-noise ratio, which is obtained by on-site equipment testing; the lg(·) function represents a logarithmic function with a base of 10;
[0040] According to the state tensor number t max , design the middle layer of the RNN unit, expressed as:
[0041] γ t =W in r t +W hide h t-1 +b;
[0042]
[0043] Among them, the γ t is the output of the hidden layer of the RNN unit at the tth timestamp; the W in Represented as the input weight matrix; W hide Represented as the internal state input weight matrix; h t-1 is the internal state vector obtained at the tth timestamp; b is the bias weight vector of the hidden layer; Represented as the internal state activation function; η t Represented as the RNN output at the tth timestamp; Represented as Sigmoid function; W out is the output weight matrix;
[0044] Using t max The output of each RNN unit is designed to output multiple pairs of RNN units as a whole, specifically:
[0045]
[0046] Preferably, the internal state activation function adopts a bitangent function or a linear correction unit function.
[0047] Preferably, the Sigmoid function is expressed as:
[0048] Preferably, the calculation method of the captured information is:
[0049] S301. The perception auxiliary module receives the mobile mode information of the receiver and the L received by the receiver. recv Sample signal Doppler interference is compensated by the following calculation:
[0050]
[0051] Wherein, z′ is the received signal after compensating Doppler interference;IF Expressed as the intermediate carrier frequency;
[0052] S302. The computational model for capturing information is:
[0053]
[0054] in, Represented as the code phase offset value in the captured information; o d Represents the output of the dth group of RNN units.
[0055] Due to the adoption of the above-mentioned technical solution, the present invention has the following beneficial effects:
[0056] The present invention discloses a low-orbit satellite acquisition method based on perception assistance and RCNN empowerment, which optimizes the signal processing process through AI technology completed by the perception assistance module and the RCNN acquisition module, avoiding the use of high-complexity processing solutions such as multi-channel parallel RAKE receivers, thereby effectively reducing the computing resource requirements and time costs;
[0057] In addition, by combining the prior information provided by the perception auxiliary module with the data-driven first convolution unit, model-driven second convolution unit and RNN unit under the RCNN architecture, it is possible to capture the target signal more accurately and counteract the Doppler effect caused by the high-speed movement of low-orbit satellites.
[0058] In addition, AI technology is used to identify the signal characteristics of different paths and automatically adjust parameters to reduce or even eliminate the negative impact of multipath propagation, ensuring good communication quality even in complex urban environments.
[0059] In addition, for mobile platforms, especially those energy-constrained application scenarios, reducing unnecessary computing burden means saving more power, extending the service life of the device or increasing the operating time of other functions;
[0060] In addition, for scenarios requiring high-quality communications of low-orbit satellites, the AI technology formed by the perception auxiliary module and the RCNN capture module in the present invention can improve the capture performance in actual low-orbit satellite communication scenarios while reducing resource overhead, which brings many implementation plans for future 6 GHz and even terahertz satellite communication capture, and is of great significance. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] Figure 1 It is a schematic diagram of the process of the present invention.
[0062] Figure 2 Schematic diagram of the perception auxiliary module of the present invention.
[0063] Figure 3Schematic diagram of the first convolution unit of the RCNN capture module of the present invention.
[0064] Figure 4 Schematic diagram of the second convolution unit of the RCNN capture module of the present invention.
[0065] Figure 5 Schematic diagram of multiple groups of RNN modules of the RCNN capture module of the present invention.
[0066] Figure 6 Capture flow chart for satellite. DETAILED DESCRIPTION
[0067] The technical solution of the present invention will be described below in conjunction with the drawings in the embodiments of the present invention. In the description, it should be understood that if there are terms such as "upper", "lower", "front", "back", "left", "right" and the like indicating directions or positional relationships, they are only corresponding to the drawings of the present invention, for the convenience of describing the present invention, and do not indicate or imply that the devices or components referred to must have a specific direction:
[0068] Embodiment 1 is:
[0069] Combined with Figure 1 The low-orbit satellite capture method based on perception assistance and RCNN empowerment is specifically as follows:
[0070] S1. The perception auxiliary module obtains the mobile mode of the receiver based on the acquired public waveform information, and generates the prior information of Doppler frequency deviation in combination with the low-orbit satellite constellation information;
[0071] The key to the perception-assisted module is to use existing communication technology and satellite position information to accurately estimate the movement of the receiver relative to the satellite in order to accurately predict and compensate for the Doppler effect. This allows processing to be performed directly on the expected Doppler frequency deviation range without the need for time-consuming search operations, greatly reducing time and resource overhead; in particular, the common waveform information includes the sparsity, geometric characteristics, and signaling information of millimeter wave communications.
[0072] S2. According to the characteristics of the satellite downlink signal, an RCNN capture module is designed, and the RCNN capture module includes a first convolution unit based on data driving, a second convolution unit based on model driving, and multiple groups of RNN units;
[0073] The first convolution unit refers to optimizing network parameters through a large amount of training data, while the second convolution unit refers to using prior knowledge or theoretical models to guide network design. For example, the length of the convolution kernel or other parameters can be designed based on the characteristics of satellite communication signals, such as the length of the training sequence, the length of the spread spectrum code, and the length of the message signal cycle. Since these designs are based on the understanding and analysis of signal characteristics, the resulting network structure is more in line with the needs of practical applications; and multiple groups of RNN units use recurrent neural networks to process time series data, especially for timing changes in satellite signals, to enhance the system's ability to adapt to dynamic environments;
[0074] S3. The RCNN capture module calculates the capture information based on the prior information generated by the perception-assisted communication module and the carrier Doppler interference of the compensated satellite signal;
[0075] The Doppler frequency deviation prior information generated by the perception auxiliary module is used to compensate the carrier Doppler interference of the received satellite signal. The compensated signal is sent to the RCNN capture module, which finally outputs accurate capture information, such as the code phase offset value.
[0076] In summary, the present invention achieves efficient and accurate low-orbit satellite signal capture by combining perception-assisted technology and deep learning models, especially convolutional neural networks and recurrent neural networks. This method can not only effectively combat multipath interference, but also significantly reduce the demand for computing resources, while improving the system's adaptability.
[0077] Embodiment 2:
[0078] Combined with Figure 2 In step S1, the specific steps of generating prior information by the perception auxiliary module are:
[0079] S101. The receiver passively receives the common waveform signal y=[y0,y1,…,y L-1 ] T , and obtain the mobile mode information I of the receiver through the following calculation model, namely:
[0080]
[0081] vect(Y DD )=[Y DD (0,0),Y DD (1,0),…,Y DD (N-1,0),Y DD (0,1),…,Y DD (N-1,M-1)] T ;
[0082]
[0083] Wherein, L represents the number of samples of the received common waveform signal; N represents the number of discrete Fourier transform points; M is obtained by M×N=L; vect -1 is an operation to devectorize a vector array to obtain a matrix; the F N Expressed as an N-dimensional DFT matrix, represents the N-dimensional conjugate DFT matrix, Represented as an M-dimensional DFT matrix; the Y DD is represented as a delay-Doppler domain matrix; the said represents the operation of dematrixing the matrix array to obtain a vector; the said f s represents the sampling rate, and I represents the mobile mode information of the receiver, including speed and distance;
[0084] S102. Calculate the prior information of Doppler frequency deviation based on the mobile mode information and the constellation information of the low-orbit satellite. The calculation formula is as follows:
[0085]
[0086] Where v represents the satellite speed in the constellation information; f c It is represented by the radio frequency, which is obtained by field equipment testing; c is represented by the speed of light, in m / s; T space It is represented as the spatial sampling interval; ΔS is the distance change between the satellite and the receiver, which is obtained by satellite positioning or other sensor networks.
[0087] Embodiment 3:
[0088] Combined with Figure 3 The first convolution unit is based on the length L of the spread spectrum code or training sequence in the downlink signal characteristics sent by the satellite seq , and the message signal period L msg Drive the data to get the receptive field parameters of the first convolution unit, that is:
[0089]
[0090] Among them, K conv1 It is represented by the receptive field size of the first convolution unit’s first layer of convolution; P avg is the average pooling layer parameter between the first and second convolution layers, 2 is because the signal of the input convolution module is the concatenation of the real and imaginary parts; the K conv2 is the receptive field size of the second layer convolution module.
[0091] Embodiment 4:
[0092] Combined with Figure 4 The second convolution unit, according to the spread spectrum code or training sequence in the satellite downlink signal characteristics and message signals The driving model is calculated, and the specific calculation formula is:
[0093]
[0094] The Γ l ,l=0,1,…,|L seq -L msg |-1 is the input of the second convolution unit,
[0095] According to the driving model, |L seq -L msg |, the length of the spreading code or training sequence L seq and the message signal period L msg , and obtain the receptive field parameters of the second convolutional unit, namely:
[0096]
[0097] The K conv3 K is the receptive field size of the first layer of the second convolutional unit; conv4 Represents the receptive field size of the second layer convolutional unit.
[0098] Embodiment 5:
[0099] Combined with Figure 5 , the RNN unit is based on the output L of the first convolution unit mod ×1 Dimension and the output L of the second convolutional unit data ×1 Dimension Design the input amount of the fusion input layer, namely:
[0100]
[0101] According to the input r of the fusion input layer in,t , and combined with the satellite signal quality and satellite signal characteristics, the state tensor number t of the RNN unit is designed max , that is:
[0102]
[0103] The C / N0 represents the receiving carrier-to-noise ratio, which is obtained by on-site equipment testing; the lg(·) function represents a logarithmic function with a base of 10;
[0104] According to the state tensor number t max , design the middle layer of the RNN unit, expressed as:
[0105] γ t =W in r t +W hide h t-1 +b;
[0106]
[0107] Among them, the γ t is the output of the hidden layer of the RNN unit at the tth timestamp; the W in Represented as the input weight matrix; W hide Represented as the internal state input weight matrix; h t-1 is the internal state vector obtained at the tth timestamp; b is the bias weight vector of the hidden layer; Represented as an internal state activation function, such as the bitangent function, linear rectified unit function and other general continuous activation functions; η t Represented as the RNN output at the tth timestamp; Represented as Sigmoid function, expressed as W out is the output weight matrix;
[0108] Using t max The output of each RNN unit is integrated into the output of multiple pairs of RNN units, specifically:
[0109]
[0110] In particular, the internal state activation function adopts a bitangent function or a linear rectified unit function; in addition, the Sigmoid function is expressed as:
[0111] Embodiment 6:
[0112] As attached Figure 6 As shown, the calculation method of the captured information is:
[0113] S301. The perception auxiliary module receives the mobile mode information of the receiver and the L received by the receiver. recv Sample signal Doppler interference is compensated by the following calculation:
[0114]
[0115] Wherein, z′ is the received signal after compensating Doppler interference; IF Expressed as the intermediate carrier frequency;
[0116] S302. The computational model for capturing information is:
[0117]
[0118] in, Represented as the code phase offset value in the captured information; o d Represents the output of the dth group of RNN units.
[0119] The parts of the present invention that are not described in detail are prior art. It is obvious to those skilled in the art that the present invention is not limited to the details of the above-mentioned exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention; therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is limited by the appended claims rather than the above description. Therefore, it is intended to include all changes within the meaning and scope of the equivalent elements of the claims in the present invention, and any figure marks in the claims should not be regarded as limiting the claims involved.
Claims
1. A low-orbit satellite acquisition method based on perception assistance and RCNN empowerment, characterized by: Specifically: S1. The perception auxiliary module obtains the mobile mode of the receiver based on the acquired public waveform information, and generates the prior information of Doppler frequency deviation in combination with the low-orbit satellite constellation information; S2. According to the characteristics of the satellite downlink signal, an RCNN capture module is designed, and the RCNN capture module includes a first convolution unit based on data driving, a second convolution unit based on model driving, and multiple groups of RNN units; S3. Based on the prior information generated by the perception-assisted communication module and the carrier Doppler interference of the compensated satellite signal, the RCNN capture module calculates the capture information.
2. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 1, characterized in that: The common waveform information includes the sparsity, geometric characteristics and signaling information of millimeter wave communications.
3. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 1, characterized in that: The downlink signal characteristics include training sequence length, spreading code length and message signal cycle length.
4. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 1, characterized in that: In step S1, the specific steps of generating prior information by the perception auxiliary module are: S101. The receiver passively receives the common waveform signal y=[y0,y1,…,y L-1 ] T , and obtain the mobile mode information I of the receiver through the following calculation model, namely: vect(Y DD )=[Y DD (0,0),Y DD (1,0),…,Y DD (N-1,0),Y DD (0,1),…,Y DD (N-1,M-1)] T ; Wherein, L represents the number of samples of the received common waveform signal; N represents the number of discrete Fourier transform points; M is obtained by M×N=L; vect -1 is an operation to devectorize a vector array to obtain a matrix; the F N Expressed as an N-dimensional DFT matrix, represents the N-dimensional conjugate DFT matrix, Represented as an M-dimensional DFT matrix; the Y DD is represented as a delay-Doppler domain matrix; the said represents the operation of dematrixing the matrix array to obtain a vector; the said f s represents the sampling rate, and I represents the mobile mode information of the receiver, including speed and distance; S102. Calculate the prior information of Doppler frequency deviation based on the mobile mode information and the constellation information of the low-orbit satellite. The calculation formula is as follows: in, is the prior information of Doppler frequency deviation; is the prior information of Doppler change rate; v represents the satellite speed in the constellation information; f c It is represented by the radio frequency; c is represented by the speed of light, in m / s; T space It is represented as the spatial sampling interval; ΔS is the distance change between the satellite and the receiver.
5. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 1, characterized in that: The first convolution unit is configured to generate a length L of a spreading code or a training sequence according to the characteristics of the downlink signal sent by the satellite. seq , and the message signal period L msg Drive the data to get the receptive field parameters of the first convolution unit, that is: Among them, K conv1 It is represented by the receptive field size of the first convolution unit’s first layer of convolution; P avg is the average pooling layer parameter between the first and second convolution layers, 2 is because the signal of the input convolution module is the concatenation of the real and imaginary parts; the K conv2 is the receptive field size of the second layer convolution module.
6. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 5, characterized in that: The second convolution unit, according to the spreading code or training sequence in the satellite downlink signal characteristics and message signals The driving model is calculated, and the specific calculation formula is: The Γ l ,l=0,1,…,|L seq -L msg |-1 is the input of the second convolution unit, According to the driving model, |L seq -L msg |, the length of the spreading code or training sequence L seq and the message signal period L msg , and obtain the receptive field parameters of the second convolutional unit, namely: The K conv3 K is the receptive field size of the first layer of the second convolutional unit; conv4 Represents the receptive field size of the second layer convolutional unit.
7. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as described in any one of claim 6, characterized in that: The RNN unit is based on the output L of the first convolution unit. mod ×1 Dimension and the output L of the second convolutional unit data ×1 Dimension Design the input amount of the fusion input layer, namely: According to the input r of the fusion input layer in,t , and combined with the satellite signal quality and satellite signal characteristics, the state tensor number t of the RNN unit is designed max , that is: The C / N0 represents the receiving carrier-to-noise ratio, which is obtained by on-site equipment testing; the lg(·) function represents a logarithmic function with a base of 10; According to the state tensor number t max , design the middle layer of the RNN unit, expressed as: γ t =W in r t +W hide h t-1 +b; Among them, the γ t is the output of the hidden layer of the RNN unit at the tth timestamp; the W in Represented as the input weight matrix; W hide Represented as the internal state input weight matrix; h t-1 is the internal state vector obtained at the tth timestamp; b is the bias weight vector of the hidden layer; Represented as the internal state activation function; η t Represented as the RNN output at the tth timestamp; Represented as Sigmoid function; W out is the output weight matrix; Using t max The output of each RNN unit is designed to output multiple pairs of RNN units as a whole, specifically:
8. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 7, characterized in that: The internal state activation function adopts a bitangent function or a linear correction unit function.
9. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 7, characterized in that: The Sigmoid function is expressed as: σ2 (x) = 1 / 1 + e -x .
10. The low-orbit satellite acquisition method based on perception assistance and RCNN empowerment as claimed in claim 1, characterized in that: The calculation method of the captured information is: S301. The perception auxiliary module receives the mobile mode information of the receiver and the L received by the receiver. recv Sample signal Doppler interference is compensated by the following calculation: Wherein, z′ is the received signal after compensating Doppler interference; IF Expressed as the intermediate carrier frequency; S302. The computational model for capturing information is: in, Represented as the code phase offset value in the captured information; o d Represents the output of the dth group of RNN units.