A Real-time Beamforming and Interference Cancellation Method for Phased Array Computing Terminals

Abstract

The present invention discloses a method for real-time beamforming and interference cancellation in a phased array computing terminal, including: Step 1: The observation target collects observation signals. For the frequency-domain baseband signals after channelization, a separate narrowband beamformer is calculated for each sub-band channel, thereby realizing real-time beamforming in the frequency domain for all baseband signals simultaneously. Based on the dynamic changes in the propagation environment, the beamforming weights are updated, the beam shape and direction are adjusted, and maximum signal-to-noise ratio beamforming is achieved; Step 2: Spatial filtering is used in the phased array computing terminal device to achieve effective interference cancellation. A spatial filter is constructed by combining the subspace projection method, and the interference signals are identified and eliminated through the calculation and eigenvalue decomposition of the covariance matrix, and finally the filtered signal data is obtained. This method can efficiently process a large number of independent tasks in matrix operations, thereby significantly reducing the calculation time and improving the observation ability of the system and the adaptability of the application scenario.

Description

Technical Field

[0001] The present invention relates to the field of radio astronomy technology, and aims at beam synthesis and suppression of radio frequency interference (RFI) in phased array computing terminal devices, in particular to a method for real-time beam synthesis and interference cancellation of a phased array computing terminal. Background Art

[0002] Radio astronomy is a science that studies radio wave radiation in the universe and conducts research by collecting and analyzing electromagnetic wave signals of natural celestial bodies in the universe. The phased array system is a cutting-edge radio observation device. Beamforming is a key technology for realizing beam pointing and gain adjustment of phased array antennas. Digital beamforming enables the phased array system to form multiple real-time beams, thereby expanding the field of view of the telescope. However, the accuracy of beam synthesis coefficients is affected by RFI. Although there are many effective methods for eliminating radio interference, due to the different geographical locations of observation devices, scientific objectives, and the complexity and time-variability of RFI, there are few general RFI elimination methods applicable to the data of all observation devices.

[0003] Currently, the astronomical phased array system belongs to the international forefront research field. Internationally, the relatively leading ones are the Parkes Telescope and the Australian Square Kilometre Array Pathfinder (ASKAP). Parkes and ASKAP respectively achieve 70×2 and 36×2 real-time beams. In addition, ASKAP has also carried out experimental research on interference cancellation and achieved certain results.

[0004] However, the current astronomical phased array system has limitations in data processing capabilities when facing a large number of array elements, and the complexity of signal components increases. While removing RFI, it is easy to affect non-RFI data, resulting in technical problems such as insufficient accuracy of real-time beam synthesis and inflexible interference cancellation algorithms. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for real-time beam synthesis and interference cancellation of a phased array computing terminal, which significantly expands the scale of data received, processed and output, shortens the processing time, realizes flexible selection of frequency signals that match the characteristics of the observation target, and at the same time can avoid RFI in certain frequency channels, flexibly adjust the beam shape and pointing, and improves the observation ability of the system and the adaptability of the application scenario.

[0006] To achieve the above purpose, the present invention provides a method for real-time beam synthesis and interference cancellation of a phased array computing terminal, and the method includes the following steps:

[0007] Step 1: The observation target collects the observation signal. For the frequency-domain baseband signal after channelization, a separate narrowband beamformer is calculated for each sub-band channel, and then real-time frequency-domain beamforming for all baseband signals is realized. Based on the dynamic changes in the propagation environment, the beamforming weights are updated, the beam shape and direction are adjusted, and maximum signal-to-noise ratio beamforming is achieved.

[0008] Step 2: Spatial filtering is used in the phased array computing terminal device to effectively reduce interference. A spatial filter is constructed by combining the subspace projection method. The interference signal is identified and eliminated through the calculation and eigenvalue decomposition of the covariance matrix, and finally the filtered signal data is obtained.

[0009] Furthermore, the phased array computing terminal device includes a phased array receiving unit, a low-noise amplifier group, an ADC+FPGA module, a GPU cluster, a multi-science target terminal, a pulsar terminal, a fast radio burst terminal, and a spectral line terminal.

[0010] Furthermore, the phased array system where the phased array computing terminal device is located consists of 224 array elements and will form 90×2 real-time beams.

[0011] Furthermore, in the GPU cluster, by combining the advantages of intensive parallel computing of the GPU and the advantages of complex process control of the CPU, the control and calculation of real-time data streams are realized.

[0012] Furthermore, the observation targets include pulsars and fast radio bursts, and the observation signal is a point source signal.

[0013] Furthermore, the point source signal is processed by a high-speed digital receiving unit to complete the channelization operation, and thus the frequency-domain baseband signal is obtained; the bandwidth of each channel will meet the narrowband constraint conditions of beamforming; in the GPU cluster, the system calculates and configures a corresponding narrowband phase-shifting beamformer for each independent sub-band channel to realize real-time frequency-domain beamforming for all baseband signals simultaneously.

[0014] Furthermore, the specific process of using maximum signal-to-noise ratio beamforming in Step 1 is as follows:

[0015] S1.1. Calculate the array covariance matrix during passive observation;

[0016] S1.2. Calculate the array covariance matrix during active observation;

[0017] S1.3. Calculate an optimal set of weights using the array covariance matrix during passive observation and the array covariance matrix during active observation;

[0018] S1.4. Use this weight vector for beamforming; calculate the beam output;

[0019] S1.5. Obtain the beam synthesis output and calculate the signal-to-noise ratio.

[0020] Furthermore, the array covariance matrix during passive observation ;

[0021]

[0022] where and are the noise vector and its conjugate transpose respectively, and E[...] represents the mean value.

[0023] Furthermore, the array covariance matrix during active observation ;

[0024]

[0025] where and are the signal vector and its conjugate transpose respectively.

[0026] Furthermore, use the optimal weight vector and the narrowband signal matrix to calculate the beam output, and its expression is:

[0027]

[0028] where, represents the conjugate transpose of the matrix.

[0029] Beneficial effects:

[0030] 1. The present invention designs a signal processing strategy in a phased array computing terminal device with 224 array elements and forming 90×2 real-time beams, including beam synthesis and interference suppression, and real-time processes high-throughput data under a heterogeneous computing system based on a central processing unit and a graphics processing unit, significantly expanding the scale of received, processed, and output data and shortening the processing time.

[0031] 2. Based on the interference cancellation algorithm of spatial filtering, a spatial filter is constructed by combining the subspace projection method, which has strong adaptability in identifying RFI, and at the same time ensures that the data processing efficiency meets the requirements of real-time processing, and reduces the impact on non-RFI data while removing RFI.

[0032] 3. For the frequency-domain baseband signal after channelization, a separate narrowband beam synthesizer is calculated for each sub-band channel, and then the frequency-domain real-time beam synthesis for all baseband signals is realized. It is possible to flexibly select and observe the frequency signals that match the target characteristics, and at the same time avoid the RFI of certain frequency channels.

[0033] 4. Update the beam synthesis weights based on the dynamic changes in the propagation environment, and flexibly adjust the beam shape and direction. The present invention realizes maximum signal-to-noise ratio beam synthesis, and can simultaneously generate 90×2 (including horizontal and vertical polarizations) beams. These real-time synthesized beams can be respectively directed to different targets or cover a large area, enhancing the system's observation ability and adaptability to application scenarios.

[0034] 5. Finally, the mutual coupling effect between receiving units will also increase the complexity of signal components. These factors pose challenges to the optimization of the algorithm itself and its real-time implementation. In response, the present invention realizes beam synthesis and interference suppression based on the maximum signal-to-noise ratio beam synthesis method and spatial filtering method. On this basis, a hierarchical and overlapping processing framework is designed to optimize the collaborative mechanism of beam synthesis and interference suppression. By reasonably allocating hardware resources and computing time, and dynamically scheduling computing tasks: tasks with dependencies are executed in sequence, and independent tasks are executed overlapped, maximizing the overall processing efficiency, thereby realizing real-time processing of large-scale data, including two effects: one is to quickly process the received signals to ensure the timeliness of data; the other is to dynamically calculate the covariance matrix of the received signals to reflect the changes in the propagation environment in real time, thereby enhancing the system's adaptive ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a schematic diagram of the overall framework structure of the phased array system according to the present invention;

[0036] Figure 2 is a schematic diagram of the element distribution of a planar array;

[0037] Figure 3 is a schematic diagram of beam synthesis principle;

[0038] Figure 4 is a flowchart of beam synthesis;

[0039] Figure 5 Schematic diagram of the spatial filtering data processing flow;

[0040] Figure 6 is a diagram of the hierarchical and overlapping processing framework. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0042] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation to the present invention. In addition, the terms "first", "second", "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0043] In the description of the present invention, it should be noted that unless otherwise clearly specified and defined, the terms "installed", "connected", "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0044] The following will be combined with Figures 1 - 6 to detail the specific embodiments of the present invention. It should be understood that the specific embodiments described herein are only for the purpose of illustrating and explaining the present invention and are not used to limit the present invention.

[0045] For the observation data of the astronomical radio telescope, the phased array system is responsible for receiving signals and processing data in real time. As Figure 1 shown, the phased array system includes a phased array receiving unit, a low-noise amplifier group, an ADC+FPGA module, a GPU cluster, a multi-science target terminal, a pulsar terminal, a fast radio burst terminal, and a spectral line terminal.

[0046] Among them, the phased array receiving unit is responsible for receiving radio signals from celestial bodies, converting them into processable electronic signals, further enhancing the weak radio signals through a low-noise amplifier group. After that, the signal enters the analog-to-digital converter (ADC) + field-programmable gate array (FPGA) module, where the analog signal is converted into a digital signal, and digital processing and channelization of the signal are performed. Next, the graphics processing unit (GPU) cluster is the hardware device for implementing the algorithms and frameworks in the present invention, which includes a large number of GPUs and a small number of central processing units (CPUs). The CPU is used to allocate and schedule the graphics processing unit (GPU) to execute various computing tasks, and beam synthesis and interference cancellation are completed here. During the signal processing of beam synthesis and RFI suppression in the GPU cluster, large-scale matrix and vector operations are involved, including solving the covariance matrix of signals between array elements, matrix eigenvalue decomposition, and high-complexity calculations such as matrix multiplication. The GPU has a highly parallel computing ability and can efficiently process a large number of independent tasks in matrix operations, thus significantly reducing the computing time. Finally, the processed data is output to the multi-science target terminal for specialized subsequent analysis for different targets.

[0047] The phased array computing terminal real-time beam synthesis and interference cancellation method according to the present invention includes the following steps:

[0048] Step 1: The observation target collects observation signals. For the signals after channelization, calculate separate narrowband beam synthesis for each sub-band channel, and then realize beam synthesis for all signals simultaneously. Update the beam synthesis weight based on the dynamic changes in the propagation environment, adjust the beam shape and direction, and realize maximum signal-to-noise ratio beam synthesis.

[0049] Step 2: Use spatial filtering in the phased array computing terminal device to achieve effective interference cancellation. Combine the subspace projection method to achieve spatial filtering, and identify and eliminate interference signals through the calculation and eigenvalue decomposition of the covariance matrix, and finally obtain the filtered signal data;

[0050] Both pulsars and fast radio bursts belong to the main observation targets of the system of the present invention, and they are point source signals. The basic principle of narrowband point source maximum signal-to-noise ratio beam synthesis is as follows:

[0051] First, introduce the calculation method of the maximum signal-to-noise ratio. For a point source, the signal-to-noise ratio achieved by the beam is

[0052]

[0053] Where: and are the beam weight vector and its conjugate transpose respectively, is the signal covariance matrix, is the noise covariance matrix.

[0054] If the source is known or assumed to be a point source, then is a matrix of rank and then we will get

[0055]

[0056] where are the weight values that satisfy the maximum signal-to-noise ratio beamforming, is a constant generated due to the normalization condition in the formula derivation, is the inverse matrix of is the array response steering vector in the direction of the source.

[0057] Figure 2 The following shows the schematic diagram of the element distribution of a planar array. This figure shows an array arranged in a three-dimensional coordinate system, where a rectangular coordinate system is used to identify the positions of the elements and their relative relationships. Among them, each black dot represents an element, and an element is a receiver for receiving signals. and The single arrows on the and axes point to the positive directions on these coordinate axes. and

[0058]

[0059] where is the signal wavelength. represents the abscissa of the element, represents the ordinate of the element, and (taking non-negative integers greater than or equal to zero respectively) jointly represent the position coordinates of the element. For example, and The coordinates of the black dot at the intersection of the and axes are (0,0). The coordinates of the black dot adjacent to it in the positive direction of the x-axis are (1,0), and the coordinates of the black dot adjacent to it in the positive direction of the y-axis are (0,1); the actual distance between the two black dots with coordinates (0,0) and (1,0) is and the actual distance between the two black dots with coordinates (0,0) and (0,1) is . Calculate the values of each element to form

[0060] During the radio observation process, the receiver continuously collects the signal stream that changes over time. By sampling the signal stream at fixed time intervals, uniformly distributed discrete electrical signals are obtained to form signal data. In the digital terminal data stream targeted by the present invention, after the signal passes through the high-speed digital receiving unit, the channelization operation is completed, and the broadband signal is decomposed into multiple narrowband signals with different center frequencies. Each sub-channel corresponds to a narrowband signal, and its bandwidth satisfies the narrowband constraint conditions for beam synthesis. Then, the data of each channel is input into the GPU cluster. In the GPU cluster, the system calculates the beam synthesis separately for each independent sub-channel, so that real-time beam synthesis can be achieved for all baseband signals simultaneously.

[0061] As Figure 3 shown in the beam synthesis principle, the oblique solid line represents the observation source, which represents the astronomical signal from outer space and is the desired reception object of astronomical observation equipment; the oblique dotted line represents the interference source, which represents the signals emitted by artificial satellites, radio and television, airplanes, and mobile communication base stations. The triangle represents the array element, that is, the receiver, which itself has random noise, such as the mutual coupling effect between array elements. These two belong to the signals that the astronomical observation equipment does not want to receive. Each array element receives the signals from the observation source and the interference source, and noise will be mixed in during the transmission process of the hardware system. These three together constitute the received signal of each array element. represents the weight factor corresponding to the signals of different receivers. The core principle of beam synthesis is summarized as the weighted summation of the received signals of each array element ( Figure 3 in represents summation. Due to the properties of matrix multiplication, this summation operation is implicit in the operation of S1.4. below). This "weight" is the weight factor, and there are various construction methods. The present invention adopts the maximum signal-to-noise ratio method for construction. The result of calculating the beam synthesis is . Figure 4 This is the process of beam synthesis. Specifically, for the array element, the steps of using the maximum signal-to-noise ratio beam synthesis in step 1 are as follows:

[0062] S1.1. Calculate the array covariance matrix during passive observation ;

[0063]

[0064] Among them, and are the noise vector and its conjugate transpose respectively, and E[...] is the mean value.

[0065] S1.2. Calculate the array covariance matrix during active observation ;

[0066]

[0067] Among them, and are the signal vector and its conjugate transpose, respectively.

[0068] S1.3. Use and to calculate an optimal set of weights;

[0069] Perform eigenvalue decomposition on the signal covariance matrix and the noise covariance matrix :

[0070]

[0071] where is the eigenvalue matrix, is the eigenvector matrix, and eig represents calculating the eigenvalues and eigenvectors of the matrix.

[0072] The eigenvalues and eigenvectors are in one-to-one correspondence, and the magnitude of the eigenvalue means the importance degree of the component of its corresponding eigenvector in the signal space. Sort the eigenvalues and extract the eigenvector corresponding to the largest eigenvalue as the optimal weight vector , denotes transpose. In practical applications, the beamforming weights will be continuously calculated and updated.

[0073] S1.4. Use this weight vector for beamforming;

[0074] Use the optimal weight vector and the narrowband signal matrix (the matrix formed by combining the signals received by all array elements) to calculate the beam output, and its expression is:

[0075]

[0076] where, denotes the conjugate transpose of the matrix.

[0077] S1.5. Obtain the beamforming output.

[0078] is the result of beamforming, reflecting the intensity of the signal in different directions. This is the formation of a single beam. Performing the above homogeneous operations in parallel on different array element combinations can generate multiple beams simultaneously. Finally, the data of different beams are stored separately as the input for RFI suppression calculation.

[0079] When part of the processing units in the GPU cluster are used for beamforming calculation, another part of the processing units are used for RFI suppression processing.

[0080] Among them, the data processing flow of interference cancellation based on spatial filtering is asFigure 5 As shown. For a large number of interference sources at unknown positions in space, it is necessary to adaptively estimate the spatial characteristics of the interference. The specific process of step 2 is as follows:

[0081] S2.1. Data preprocessing. Store the data of each beam in matrix form respectively. Preprocess this data to ensure the accuracy of the data in a scientific sense. The preprocessing is to uniformly correct the elements in the signal matrix. The signal matrices of each beam after preprocessing are used as the input data for the next calculation to calculate the covariance matrix. It includes S2.1.1 - S2.1.3:

[0082] S2.1.1 Calibration

[0083] Calibrate the data to eliminate the differences in the telescope system over time and frequency. What the radio telescope actually measures is the electrical signal power recorded by the receiver, called the raw data. Convert it into a meaningful physical quantity, temperature, using noise tube calibration.

[0084] S2.1.2 Band - pass removal

[0085] The band - pass and baseline include differences between different beams. For example, in the on - source and off - source observation modes, subtract the median from the input data at fixed intervals to remove the band - pass.

[0086] S2.1.3 Baseline removal

[0087] Cut the data into sub - bands by frequency, fit the baseline using methods such as second - order and third - order polynomials, and select a better one using the chi - square test.

[0088] S2.2. Calculate the covariance matrix.

[0089] For each beam signal matrix after preprocessing, solve the covariance between beam signals and construct the covariance matrix , and its expression is:

[0090]

[0091] Among them, represents the th beam signal, is the expectation, is 's complex conjugate. Each element in represents the dot product of different beam signal combinations. This matrix is the key foundation of spatial filtering technology, containing the correlation characteristics of beam data in the spatial dimension, especially the characteristics of signals and interference.

[0092] S2.3. Eigenvalue decomposition to distinguish the interference subspace and the target signal subspace;

[0093] Since the astronomical signal terms, interference terms, and receiver noise terms in each element of the covariance matrix are combined additively, the covariance matrix can be divided into an astronomical signal matrix, an interference matrix, and a noise matrix, corresponding respectively to , , and . In antenna array applications, the integration time of the covariance matrix should be short enough so that the celestial signal part is negligible relative to the interference and noise. Then we have:

[0094]

[0095] Given the symmetry of the covariance matrix, the singular value decomposition method is used to perform eigenvalue decomposition on it, thereby decomposing it into the product form of eigenvalues and eigenvectors to extract spatial information. With the help of eigen-decomposition technology, the interference subspace and the target signal subspace are identified and distinguished, laying the foundation for subsequent interference suppression.

[0096] S2.4. Interference subspace removal. Use the interference projection matrix to construct a filter to remove the interference subspace.

[0097] After obtaining the covariance matrix and eigen-decomposition, the present invention uses the interference projection matrix to construct a filter. Assume that is the known interference space structure, and the covariance matrix containing interference and noise becomes:

[0098]

[0099] where represents the variance of the noise, and represents an identity matrix.

[0100] The projection matrix on the interference space is defined as:

[0101]

[0102] Here, and represent the transpose matrix and the inverse matrix respectively.

[0103] Using the result of eigenvalue decomposition, the interference subspace is removed through the projection matrix . Specifically, the projection matrix of the interference signal is calculated, and the interference component is removed from the covariance matrix to obtain the filtered covariance matrix without RFI, that is:

[0104]

[0105] This can significantly reduce the impact of interference.

[0106] S2.5. Reconstruct data from the covariance matrix. Extract the diagonal elements from the covariance matrix after removing interference to reconstruct the original data after interference suppression.

[0107] From the covariance matrix after removing interference extract the diagonal elements, which represent the autocorrelation components of the signal, so as to be able to reconstruct the original data after interference suppression. Subsequently, the reconstructed data is stored in a standard format for subsequent in-depth analysis and processing.

[0108] S2.6. Residual calculation and threshold setting. Calculate the residual between the data after spatial filtering and the original input data, set a threshold, retain the data that meets the threshold limit and eliminate the data that does not meet the threshold limit.

[0109] Calculate the residual between the data after spatial filtering and the original input data and use the mean value of the residual and the standard deviation multiplied by the set adjustment factor to set the threshold which is expressed as:

[0110]

[0111] Retain the data that meets the threshold limit and eliminate the data that does not meet the threshold limit, so as to effectively identify and mark the remaining low-intensity interference regions.

[0112] S2.7. Output of spatial filtering results

[0113] Save the final spatial filtering results in the.npz format as the basic file for subsequent data analysis and processing. This file records the data after spatial filtering processing, realizing the suppression of interference signals and the enhancement of target signals.

[0114] The entire process realizes the identification and elimination of interference signals through the calculation and eigenvalue decomposition of the covariance matrix, and finally obtains high-quality signal data after filtering. This method is applicable to signal extraction in large-scale arrays and complex interferences.

[0115] The present invention improves the real-time processing strategy of beamforming and interference cancellation. The present invention hierarchizes and overlaps tasks, that is, logically divides according to the dependency relationship between tasks, sequentially executes hierarchical tasks with dependencies, and overlaps tasks without dependencies, thereby realizing parallelization and improving the processing speed. As Figure 6The figure shows a framework diagram for hierarchical and overlapping processing. Among them, the arrows pointing from the outside to the input layer represent the input data, which is the narrowband signal data of each channel. The arrows pointing from the output layer to the outside represent the output data, which is the beam signals of each channel after interference elimination. The other arrows represent the flow relationship of data input and output within the framework. The specific implementation architecture is as follows:

[0116] Input layer: The input layer is responsible for receiving the channelized narrowband signal data of each channel by the CPU and distributing it to the GPU nodes;

[0117] Processing layer: In the processing layer, the GPU nodes are mainly responsible for executing three types of core tasks. When executed for the first time, Task 1, Task 2, and Task 3 are started in sequence; thereafter, different tasks and homogeneous operations within the same task are all computed in parallel. The processing layer adopts a pipeline architecture and starts processing from the data at a certain moment, specifically including:

[0118] Task 1: Calculate the covariance matrix during active and passive observations and calculate the weight factors for different array element combinations;

[0119] Task 2: Calculate the beamforming result for the narrowband signal data of each channel;

[0120] Task 3: Perform spatial filtering on the input data to suppress interference and simultaneously start the above operations at the next moment. Here, "the next moment" only represents the logical order, and the output corresponding to beamforming is the input for interference cancellation;

[0121] And so on, forming a pipeline in time series. The CPU dynamically schedules the execution of each GPU to perform different tasks.

[0122] Decision-making layer: The CPU adjusts the weight factors in real time according to the processing results and judges whether the maximum number of iterations is reached or other termination conditions are met.

[0123] Output layer: Generate multi-beam signals after interference cancellation and transmit them to the multi-science target terminal.

[0124] The technical advantages of the present invention are as follows:

[0125] The present invention designs a signal processing strategy in a phased array computing terminal device for 224 array elements to form 90×2 (including dual polarization) real-time beams, including beamforming and interference suppression. These two signal processing operations are performed by Figure 1The "GPU cluster" in completes the calculation. For a phased array system, the amount of data processed by its digital backend operation is directly proportional to the number of array elements, the number of beams, and the instantaneous bandwidth of the signal. The phased array system targeted by the present invention has an array element and data scale that have not been achieved by other existing astronomical phased array systems. Therefore, the present invention designs a strategy that can meet the requirements of processing such a large amount of data, and can also be migrated and applied to other similar systems. The advantage is that it can not only quickly process the received signal, but also continuously calculate the covariance matrix of the received signal to reflect the dynamic changes of the propagation environment, enabling the system to perform beam synthesis and interference cancellation calculations in real time in complex and changing scenarios.

[0126] The present invention proposes an efficient real-time signal processing architecture. On the basis of retaining the cores of the beam synthesis based on the maximum signal-to-noise ratio and the interference cancellation algorithm based on spatial filtering, the specific tasks in the two algorithms are hierarchically and overlapped processed to fully realize the parallelization of tasks and improve the data processing speed. In the architecture, the CPU is responsible for receiving and distributing narrowband signal data to the GPU nodes. The GPU nodes rely on the pipeline architecture to parallelly execute three types of core tasks: one is to calculate the covariance matrix and its weight factors during active and passive observations; the second is to perform beam synthesis; the third is to implement spatial filtering to suppress interference and continuously initiate subsequent operations. The decision-making layer dynamically adjusts the weight factors by the CPU, judges the processing iteration conditions, and finally generates multi-beam signal data after interference cancellation. This framework makes full use of the dense parallel computing advantage of the GPU and the complex process control advantage of the CPU: the CPU is mainly responsible for complex logical operations such as process control, such as monitoring the data flow of ports and distributing it to different GPU cards for processing; the GPU is responsible for large-batch matrix operations in the algorithm process, such as solving the covariance matrix of signals between array elements, matrix eigenvalue decomposition, and matrix multiplication. Efficient dynamic scheduling is achieved.

[0127] The broadband signal is divided into multiple sub-band channels, and separate narrowband beam synthesis is calculated for each sub-band channel, thereby realizing real-time beam synthesis in the frequency domain for all baseband signals at the same time. It is possible to flexibly select frequency signals that match the characteristics of the observation target, and at the same time avoid RFI in some frequency channels.

[0128] Update the beam synthesis weights based on the dynamic changes in the propagation environment, and flexibly adjust the beam shape and direction. When astronomical signals from distant cosmic space propagate to ground receivers on Earth, they are affected by various factors, such as interference from passing airplanes and satellites, random noise of the hardware devices themselves, and changes in propagation channel conditions. These factors cause the amplitude and phase characteristics of the received signals to change randomly. By using the covariance matrix, analyze the covariance of different received signals, quantify the statistical characteristics of these signals, and provide information about various components in the propagation environment, so as to identify and eliminate unwanted signal components. By continuously calculating the covariance matrix, enable the system to continuously update the estimation of the propagation environment characteristics based on the newly received data. The dynamic update can capture the real-time changes in the propagation environment. For example, when it is detected that the change in the short-term covariance matrix is significantly higher than that in the long-term covariance matrix, it may be an environmental mutation caused by a sudden interference such as an airplane passing by. Such abnormal signals are considered as interference and these components need to be subtracted from the signal data. The present invention realizes maximum signal-to-noise ratio beam synthesis, and can simultaneously generate 90×2 (including dual polarization) beams. These real-time synthesized beams can be respectively directed to different targets or cover a large area, improving the observation ability of the system and the adaptability of the application scenario.

[0129] Implement an interference cancellation algorithm based on spatial filtering, and combine the subspace projection method to achieve spatial filtering. It has strong self-adaptability when identifying RFI, and at the same time ensures that the data processing efficiency meets the requirements of real-time processing.

[0130] Any process or method described in the flowchart of the present invention or described in other ways herein can be understood as representing a module, segment, or part of code including one or more executable instructions for implementing a specific logical function or process, and can be implemented in any computer-readable medium for an instruction execution system, apparatus, or device. The computer-readable medium can be any medium including storage, communication, propagation, or transmission programs for use by an instruction execution system, apparatus, or device, including read-only memory, magnetic disks, or optical discs, etc.

[0131] In the description of this specification, the descriptions referring to terms such as "embodiment", "example", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. In addition, those skilled in the art can combine or combine different embodiments or examples described in this specification and the features therein without conflict.

[0132] Although the above has shown and described embodiments of the present invention, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can perform update operations such as changes, modifications, substitutions, and variations on the above embodiments within the scope of the present invention.

Claims

1. A real-time beamforming and interference cancellation method for a phased array computing terminal, characterized in that, The method comprises the following steps: Step 1: The observation target collects observation signals and calculates a separate narrowband beamformer for each sub-band channel of the channelized frequency domain baseband signal. This enables real-time frequency domain beamforming of all baseband signals simultaneously. The beamforming weights are updated based on dynamic changes in the propagation environment, and the beam shape and pointing are adjusted to achieve beamforming with maximum signal-to-noise ratio. Step 2: Use spatial filtering in the phased array computing terminal device to achieve effective interference reduction. Combined with the subspace projection method, a spatial filter is constructed. The interference signal is identified and eliminated through covariance matrix calculation and eigendecomposition, and the filtered signal data is finally obtained. Wherein, step 2 includes: S2.

1. Data preprocessing: Store the data for each beam in matrix form and preprocess this data. The signal matrix of each beam after preprocessing serves as the input data for the next step of calculation to calculate the covariance matrix. S2.

2. Calculate the covariance matrix: For each beam signal matrix after preprocessing, solve the covariance between beam signals and construct the covariance matrix , whose expression is: ； Among them, represents the th beam signal, , is the expectation, is 's complex conjugate; S2.

3. Eigenvalue decomposition, the covariance matrix is decomposed into an astronomical signal matrix , an interference matrix and a noise matrix . If the integration time of the covariance matrix is short enough, then we have: ； The singular value decomposition method is used to decompose the eigenvalue, breaking it down into the product of eigenvalue and eigenvector to extract spatial information, identify and distinguish the interference subspace and target signal subspace; S2.

4. Interference subspace removal: Use the interference projection matrix to construct a filter to remove the interference subspace; Assume is a known interference spatial structure, and the covariance matrix containing interference and noise becomes: ； in represents the variance of the noise, represents an identity matrix; The projection matrix in the interference space is defined as: ； and denote the transposed matrix and the inverse matrix respectively; Calculate the projection matrix of the interference signal and remove the interference component from the covariance matrix to obtain the filtered covariance matrix without RFI: ； S2.

5. Reconstructing Data from the Covariance Matrix: Extract the diagonal elements from the interference-removed covariance matrix to reconstruct the original data after interference suppression. S2.

6. Residual calculation and threshold setting: Calculate the residual between the spatially filtered data and the original input data , and use the mean of the residuals and the standard deviation multiplied by a set adjustment factor to set the threshold , which is expressed as: ； The data that meets the threshold limit is retained and the data that does not meet the threshold limit is eliminated, thereby effectively identifying and marking the remaining low-intensity interference areas; S2.

7. Output of spatial filtering results.

2. The real-time beam synthesis and interference cancellation method for a phased array computing terminal according to claim 1, wherein The phased array computing terminal device includes a phased array receiving unit, a low-noise amplifier group, an ADC+FPGA module, a GPU cluster, a multi-scientific target terminal, a pulsar terminal, a fast radio burst terminal and a spectral line terminal.

3. The real-time beam synthesis and interference cancellation method for a phased array computing terminal according to claim 2, wherein The phased array system where the phased array computing terminal device is located consists of 224 array elements, which will form 90×2 real-time beams.

4. The real-time beam synthesis and interference cancellation method for a phased array computing terminal according to claim 2, characterized in that In a GPU cluster, the intensive parallel computing advantages of the GPU and the complex process control advantages of the CPU are combined to achieve real-time data flow control and calculation.

5. The real-time beamforming and interference reduction method for a phased array computing terminal according to claim 1, characterized in that: The observation targets include pulsars and fast radio bursts, and the observation signals are point source signals.

6. The real-time beam synthesis and interference cancellation method for a phased array computing terminal according to claim 5, wherein The point source signal is processed by a high-speed digital receiving unit to complete the channelization operation, thereby obtaining the frequency domain baseband signal; the bandwidth of each channel will meet the narrowband constraint of beamforming; in the GPU cluster, the system calculates and configures the corresponding narrowband phase-shifted beamformer for each independent sub-band channel, realizing real-time frequency domain beamforming for all baseband signals at the same time.

7. The real-time beam synthesis and interference cancellation method for a phased array computing terminal according to claim 6, wherein, The specific process of using maximum signal-to-noise ratio beamforming in step 1 is as follows: S1.

1. Calculate the array covariance matrix for passive observations; S1.

2. Calculate the array covariance matrix for active observations; S1.

3. Calculate an optimal set of weights using the array covariance matrix in passive observation and the array covariance matrix in active observation; S1.

4. Use the weight vector of this set of weights for beamforming; calculate the beam output; S1.

5. Obtain the beamforming output and calculate the signal-to-noise ratio.

8. The real-time beamforming and interference reduction method for a phased array computing terminal according to claim 7, characterized in that: Array covariance matrix during passive observation ; ； where and are the noise vector and its conjugate transpose respectively, and E[...] represents the mean value.

9. The method for real-time beam synthesis and interference cancellation of a phased array computing terminal according to claim 5, characterized in that, Array covariance matrix during active observation ; ； wherein, and are the signal vector and its conjugate transpose respectively.

10. The method for real-time beam synthesis and interference cancellation of a phased array computing terminal according to claim 5, wherein, Using the optimal weight vector and the narrowband signal matrix calculate the beam output, and its expression is: ； Among them, represents the conjugate transpose of a matrix.

Images