Anti-interference microwave communication system and optimization method thereof

Abstract

The invention relates to the technical field of communication, in particular to an anti-interference microwave communication system and an optimization method thereof. Comprising a cognitive reconfigurable metasurface aperture, a broadband multi-channel receiver, a cognitive engine and a communication transceiver. The cognitive engine serves as an intelligent decision-making core, and through real-time sensing, spectrum cognition, threat prediction and multi-target optimization of an electromagnetic environment, a control instruction is generated to adaptively reconstruct the metasurface aperture. The system can actively adapt to a dynamic electromagnetic environment through a cognitive closed loop, accurately suppress multi-dimensional interference and continuously optimize communication efficiency.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to an anti-interference microwave communication system and an optimization method thereof. Background Art

[0002] Microwave communication technology, due to its advantages such as high transmission capacity and strong resistance to multipath fading, is widely used in fields such as 5G, satellite communications, and electronic countermeasures. However, with the explosive growth of wireless applications, the electromagnetic environment is becoming increasingly complex, spectrum resources are becoming increasingly congested, and various natural and man-made interferences coexist, making anti-interference capability a key bottleneck in determining system reliability and survivability.

[0003] Existing anti-interference solutions are mostly based on specific physical mechanisms or fixed hardware structures, such as leveraging the ultra-narrowband filtering properties of Rydberg atoms to suppress out-of-band interference. While these solutions are somewhat effective, their anti-interference capabilities are fixed during the system design phase, lacking the ability to perceive and adaptively control the real, dynamic electromagnetic environment in real time. When faced with intelligent interference with time-, frequency-, and space-variant characteristics, these static defense mechanisms expose inherent limitations in adaptability, making it difficult to transition from passive mitigation to active adaptation.

[0004] Therefore, there is an urgent need for a new microwave communication system that can perceive the environment in real time, make intelligent decisions, and dynamically reconstruct in order to cope with complex dynamic interference and improve communication efficiency and robustness. Summary of the Invention

[0005] The technical problem to be solved by the present invention is that when existing microwave communication systems deal with complex intelligent interference with time-varying, frequency-varying and space-varying characteristics, their anti-interference mechanisms are mostly static physical mechanisms or fixed hardware structures, and they lack the closed-loop adaptive capabilities that combine real-time perception of the electromagnetic environment, cognitive decision-making and dynamic allocation of communication resources, resulting in serious lack of adaptability and survivability in dynamic confrontation environments.

[0006] To achieve the above-mentioned purpose of the invention, the present invention provides an interference-resistant microwave communication system and its optimization method, which aims to enable the system to actively adapt to the dynamically changing electromagnetic environment by constructing a cognitive closed loop that integrates real-time environmental perception, intelligent evaluation and prediction, and physical aperture reconstruction capabilities, thereby achieving precise suppression of multi-dimensional interference and continuous optimization of communication efficiency.

[0007] The present invention also provides an anti-interference microwave communication system, which includes: a cognitive reconfigurable metasurface aperture, a broadband multi-channel receiver subsystem, a cognitive engine subsystem, and a communication transceiver subsystem.

[0008] The cognitive reconfigurable metasurface aperture physically constitutes the antenna of the system and is used to interact with electromagnetic waves. The cognitive reconfigurable metasurface aperture includes metasurface units arranged in a two-dimensional array and having tunable electromagnetic response characteristics, a feeding network for feeding or receiving electromagnetic waves to the metasurface units, and a metasurface control matrix. Specifically, the metasurface unit array is arranged on a dielectric substrate, and a metal ground plate is provided on the other side of the dielectric substrate. Each metasurface unit is composed of a metal patch of a specific geometric shape and an active tuning device embedded therein. The active tuning device is a varactor diode, whose capacitance value is determined by the DC bias voltage applied to its two ends. The metasurface control matrix is ​​composed of a high-density digital-to-analog converter array and a programmable logic controller, and each digital-to-analog converter output end of the metasurface control matrix is ​​electrically connected to the bias voltage input end of the varactor diode of a metasurface unit. The metasurface control matrix is ​​connected to the cognitive engine subsystem via a high-speed serial peripheral interface bus, and is used to receive the metasurface control words generated by the cognitive engine subsystem and convert them into a precise analog bias voltage sequence that drives the varactor diodes in each metasurface unit, thereby achieving independent and high-speed control of the reflection phase of each metasurface unit, and then reconstructing the electromagnetic response characteristics of the entire aperture, including the main lobe direction, null position, sidelobe level and polarization mode of the radiation pattern.

[0009] The broadband multi-channel receiver subsystem is electrically connected to the feed network of the cognitive reconfigurable metasurface aperture and is used to capture and digitize the broadband electromagnetic environment signals received by the aperture indiscriminately and with high fidelity. The broadband multi-channel receiver subsystem includes a low-noise amplifier (LNA) covering the operating frequency band, a power distribution network, a multi-channel downconversion module, and a high-speed analog-to-digital converter array. The LNA performs initial amplification of received weak signals, with a gain of no less than 25dB and a noise figure no greater than 1.2dB. The power distribution network evenly distributes the amplified signal to multiple processing channels. The multi-channel downconversion module, consisting of a set of local oscillators and mixers, downconverts RF signals from different frequency bands to a unified intermediate frequency (IF). The high-speed ADC array synchronously samples and quantizes the IF signals of each channel, with a sampling rate of no less than 40GS / s and a quantization bit count of no less than 14 bits, thereby generating a broadband digital IF signal stream with a high dynamic range that fully represents the current electromagnetic environment.

[0010] The cognitive engine subsystem is the intelligent decision-making core, responsible for in-depth processing and analysis of the digital signal stream output by the broadband multi-channel receiver subsystem and generating control instructions for reconfiguring the physical properties of the cognitive reconfigurable metasurface aperture. The cognitive engine subsystem is structurally divided into a real-time spectrum recognition module, a situation assessment and prediction module, and a resource optimization and allocation module.

[0011] Furthermore, the real-time spectrum awareness module is implemented on a field-programmable gate array (FPGA) hardware platform to meet the parallel, pipelined processing requirements of massive data streams. This module receives the digital intermediate frequency (IF) signal stream from the high-speed analog-to-digital converter array and, through its integrated digital signal processing core, performs the following operations: First, it performs high-resolution spectral analysis of the wideband signal using a polyphase filter bank (PFB) or overlap-add fast Fourier transform (FFT) algorithm to generate a real-time environmental spectrum map. It then detects independent signal entities on the spectrum map using an adaptive threshold decision algorithm based on energy detection. Then, for each detected signal entity, it invokes a library of parameter estimation algorithms to extract its key physical parameters. These include digital beamforming or multiple signal classification algorithms for angle of arrival (AoA) estimation, a centroid algorithm for center frequency and bandwidth estimation, an integration algorithm for power estimation, and a high-order cumulant-based classification algorithm for modulation type and polarization state identification. Finally, the real-time spectrum awareness module encapsulates the full set of parameters for each signal entity into a structured environmental state vector and outputs it to the situation assessment and prediction module. The data structure of the environmental state vector is deterministically defined as: signal identifier, timestamp, center frequency, signal bandwidth, signal power, arrival angle azimuth, arrival angle elevation, polarization state code, and signal type.

[0012] Furthermore, the situation assessment and prediction module is implemented based on a graphics processor or a multi-core digital signal processor hardware platform, and is used to perform cognitive tasks that are more complex than real-time processing. The situation assessment and prediction module periodically receives the environmental state vector stream generated by the real-time spectrum recognition module and performs the following operations: First, the module maintains a multi-dimensional, time-evolving electromagnetic environment model in the internal memory and uses the received new environmental state vector to update this model; first, based on the preset threat assessment rule base, the signal entities in the environmental model are divided into target communication signals, non-cooperative signals, interference signals and background noise, and the threat level of the interference signal is calculated according to its power, frequency proximity and directional threat degree to generate a threat assessment matrix; then, for the signal entity identified as a mobile interference source, the module starts the Kalman filter to track its arrival angle and predict its trajectory. The state vector of the Kalman filter includes azimuth, pitch angle and its first-order time derivative. Through the state transfer equation and measurement update equation, the module outputs the predicted value of the position of the interference source in the next decision cycle; in addition, for intelligent interference sources, the module uses the long short-term memory (LSTM) recurrent neural network to analyze its historical behavior pattern (such as frequency hopping sequence, frequency sweep characteristics) to identify its interference strategy. The final output of the situation assessment and prediction module is a threat assessment matrix containing all current threat information and a predicted environmental state for the future environmental state.

[0013] Furthermore, the resource optimization and allocation module is implemented on a central processing unit (CPU) and serves as the final executor of decisions. This module receives the threat assessment matrix and predicted environmental state from the situation assessment and prediction module, as well as the current communication link quality requirements (e.g., target data rate, maximum tolerable bit error rate) from the communication transceiver subsystem as constraints. The core function of this module is to solve a multi-objective optimization problem to generate the optimal metasurface aperture configuration. The objective function of the optimization problem is defined as maximizing the signal-to-interference-and-noise ratio (SINR) of the target communication link while minimizing the gain of the directional pattern in the predicted arrival angles of all high-threat interference sources to below a preset deep nulling threshold. To solve this non-convex optimization problem, the resource optimization and allocation module employs a genetic algorithm. In this algorithm, each chromosome represents a complete set of phase control values ​​covering all metasurface elements; the fitness function is directly determined by the predicted SNR calculated based on the chromosome configuration; and a global search is performed in the solution space using selection, crossover, and mutation genetic operators. When the algorithm converges or reaches a preset number of iterations, the optimal chromosome obtained is decoded into a set of precise digital phase values, encapsulated into the metasurface control word (MCW), and sent to the metasurface control matrix of the cognitive reconfigurable metasurface aperture via a high-speed serial peripheral interface bus.

[0014] On the one hand, the communication transceiver subsystem is connected to the feeding network of the cognitive reconfigurable metasurface aperture, and is used to receive the target communication signal after spatial filtering and purification through the main lobe beam, or transmit the signal in the target direction after the aperture is optimized and reconstructed; on the other hand, it is connected to the resource optimization and configuration module of the cognitive engine subsystem, and is used to feed back the decoded communication link performance indicators (such as real-time bit error rate and signal-to-noise ratio) to the optimization module as a basis for evaluating the effectiveness of the current aperture configuration, and for dynamically adjusting the objective function weight in the next round of optimization process.

[0015] According to another aspect of the present invention, a method for optimizing an anti-interference microwave communication system is provided. The method is executed by the above-mentioned system and is characterized in that it includes the following steps: Step 1: Environmental detection and broadband digitization. The broadband multi-channel receiver subsystem is activated, and the electromagnetic environment signals covering the entire operating frequency band in the space where the system is located are continuously captured through the cognitive reconfigurable metasurface aperture. The high-speed analog-to-digital converter array converts them into a high-fidelity broadband digital intermediate frequency signal stream.

[0016] Step 2: Real-time spectrum recognition and state vector generation: The broadband digital IF signal stream is fed into the real-time spectrum recognition module of the recognition engine subsystem. This module performs parallel, pipelined, high-resolution spectrum analysis on the signal stream, detecting individual signal entities in the spectrum. It accurately extracts the parameters of each signal entity, encapsulates them into a structured environmental state vector, and outputs them. These parameters include frequency, power, angle of arrival, polarization state, and signal type.

[0017] Step 3: Situation Assessment and Dynamic Threat Prediction. The generated series of environmental state vectors are fed into the Situation Assessment and Prediction module. This module updates its internal time-series electromagnetic environment model, identifies and ranks threats for all signal entities, and generates a Threat Assessment Matrix (TAM). Simultaneously, a Kalman filter is used to track the trajectory of identified mobile interference sources and predict their arrival angle for the next time period. For intelligent interference sources, a recurrent neural network is used to analyze their behavior patterns and generate a predicted environmental state (PES).

[0018] Step 4: Multi-objective optimization and aperture configuration strategy generation. The threat assessment matrix and predicted environmental state, along with the performance indicators required by the upper-layer communication task, are fed into the resource optimization and configuration module. This module constructs a multi-objective optimization model centered on maximizing the target signal-to-interference-and-noise ratio (SINR) and constrained by suppressing all known and predicted interference. A genetic algorithm is then used to solve the model until an optimal set of phase control values ​​is found for each metasurface element in the cognitive reconfigurable metasurface aperture. This set of phase control values ​​is encoded as a metasurface control word.

[0019] Step 5: Adaptive Reconstruction of Aperture Physical Properties: The resource optimization and configuration module transmits the generated metasurface control word to the metasurface control matrix of the cognitive reconfigurable metasurface aperture via a high-speed serial peripheral interface bus. The metasurface control matrix decodes it into a set of analog bias voltages and precisely applies them to each element of the metasurface array, thereby physically changing the electromagnetic scattering characteristics of the entire aperture, precisely aligning the main lobe of its radiation pattern in the direction of the target signal and forming deep nulls in all known and predicted interference directions.

[0020] Step 6: Communication execution and closed-loop feedback. After the cognitive reconfigurable metasurface aperture completes the reconfiguration, the communication transceiver subsystem performs the communication task through the optimized spatial channel. At the same time, the communication transceiver subsystem continuously monitors the actual performance parameters of the communication link, such as the bit error rate and the signal-to-noise ratio, and feeds these parameters back to the resource optimization and configuration module of the cognitive engine subsystem in real time. The feedback information is used to verify the effectiveness of the current aperture configuration and as a basis for dynamically adjusting the target function weights or constraint conditions in the subsequent optimization period. The system then returns to step 1 and continuously cycles through the above steps, forming a complete, uninterrupted adaptive closed loop of cognition-decision-execution-feedback, thereby ensuring that the communication system always maintains optimal anti-jamming performance and communication efficiency in a dynamically changing complex electromagnetic environment.

[0021] Compared with the prior art, the beneficial effects of the present application are: The present application introduces a cognitive reconfigurable metasurface aperture as the intelligent physical front end of the system, combines a wideband multi-channel receiver, a cognitive engine, and a closed-loop feedback mechanism to build a complete "perception-cognition-decision-execution-feedback" adaptive loop. The core beneficial effect is to achieve a fundamental leap from "static anti-jamming" to "dynamic cognitive anti-jamming". Specifically, the present application can perform wideband, panoramic, real-time perception of complex electromagnetic environments, accurately identify and extract multiple types of signal parameters including fixed, mobile, and intelligent interference; then, through the built-in situation assessment and prediction module (combining Kalman filtering and LSTM neural network), the behavior of the interference source is tracked and predicted; finally, the metasurface aperture is driven by a multi-objective optimization algorithm (such as genetic algorithm) to perform dynamic reconfiguration in the spatial, frequency, and polarization domains, forming a high-gain beam in the target direction while generating deep nulls (measured to be below -40 dBi) in the known and predicted interference directions. The measured data show that when dealing with complex dynamic interference, the present application can significantly improve the received signal-to-interference-and-noise ratio from -3.5 dB of the traditional system to +18.2 dB, and the bit error rate from an uncommunicable state to a high-quality , and the response time for completing a complete cognitive optimization closed loop is less than 50 milliseconds, with excellent real-time countermeasures capability; In addition, the system uses a heterogeneous computing architecture to reasonably allocate the computing resources of FPGA, GPU, and CPU, combining high-speed real-time processing and complex algorithm decision-making capability, with high system integration and strong engineering realizability. In summary, the present application greatly improves the reliability and survivability of microwave communication in complex electromagnetic environments. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is the overall structure block diagram of the anti-jamming microwave communication system in the present application; Figure 2 ​Schematic diagram of the structure of the cognitive reconfigurable metasurface aperture in the present invention; Figure 3 yes Figure 2 A magnified view of the structure at the supersurface unit A; Figure 4 This is a block diagram of the internal modules of the cognitive engine subsystem of the present invention; Figure 5 It is a flow chart of the anti-interference microwave communication system optimization method of the present invention; Figure 6 Schematic diagram of interference in the radiation direction of the antenna of the system before optimization and reconstruction in the present invention; Figure 7 Schematic diagram of the system in the present invention forming anti-interference in the antenna radiation direction after optimization and reconstruction; Figure 8 Schematic diagram of a specific application in an embodiment of the present invention; Figure 9 This is a flow chart of the situation assessment and prediction module of the present invention performing trajectory tracking and prediction on a mobile interference source; Figure 10 This is a structural block diagram of the reconfigurable metasurface aperture of the present invention.

[0023] The accompanying drawings are numbered as follows: 100, cognitive reconfigurable metasurface aperture; 110, metasurface unit; 111, metal patch; 112, active tuning device; 120, feed network; 130, metasurface control matrix; 140, dielectric substrate; 150, metal ground plate; 200, broadband multi-channel receiver subsystem; 300, cognitive engine subsystem; 310, real-time spectrum recognition module; 320, situation assessment and prediction module; 330, resource optimization and configuration module; 400, communication transceiver subsystem. DETAILED DESCRIPTION

[0024] To make the purpose, technical solutions and advantages of the present invention more clear, the following will describe the specific embodiments of the present invention in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0025] See also Figure 1, which demonstrates the overall system structure of an anti-interference microwave communication system and its optimization method provided by the present invention. The system aims to achieve adaptive countermeasures against complex dynamic electromagnetic environments through a highly integrated cognitive closed loop. In a specific embodiment, the system physically includes a cognitive reconfigurable metasurface aperture 100, a broadband multi-channel receiver subsystem 200, a cognitive engine subsystem 300 as the core of intelligent decision-making, and a communication transceiver subsystem 400 responsible for executing communication tasks. These four subsystems are tightly coupled with the control interface via an internal high-speed data bus, forming a complete functional chain from environmental perception to physical execution.

[0026] Specifically, the cognitive reconfigurable metasurface aperture 100 is the physical interface for the system to exchange energy with the external electromagnetic space. Its function is not limited to the traditional antenna's transceiver function, but also has the ability to dynamically reconfigure its electromagnetic response characteristics according to instructions. Figure 2 The core of the cognitive reconfigurable metasurface aperture 100 is a two-dimensional array consisting of a large number of independently controllable metasurface units 110 arranged in a dense pattern. This array is precisely fabricated on a dielectric substrate 140, the back of which is covered with a complete metal ground plane 150, which acts as a reflective surface and shielding layer.

[0027] Furthermore, the structure of the metasurface unit 110 is carefully designed to achieve a wide range of electromagnetic property tuning. Figure 3 As shown, each metasurface unit 110 is mainly composed of a metal patch 111 with a specific geometric configuration and an active tuning device 112 embedded or integrated in the patch structure. In a preferred embodiment, the geometric shape of the metal patch 111 is designed to be a symmetrical "I" shape, which can provide good resonance characteristics and polarization purity within the target operating frequency band. The active tuning device 112 is the key to achieving electrical tunability. In this embodiment, the active tuning device 112 uses a high-performance gallium arsenide hyperabrupt junction varactor diode. The varactor diode is precisely welded at the break in the center of the "I"-shaped patch, and its junction capacitance value shows a nonlinear but deterministic correspondence with the reverse bias DC voltage applied to its two ends. By precisely controlling the bias voltage, the capacitance value of the varactor diode can be changed, thereby changing the reflection phase of the incident electromagnetic wave of the metasurface unit 110 in which it is located.

[0028] In order to perform high-speed, parallel, and independent control of tens of thousands of metasurface units 110 in the array, the system is equipped with the metasurface control matrix 130. The metasurface control matrix 130 is composed of a high-density, multi-channel digital-to-analog converter array and a programmable logic controller in hardware. The output channel of each multi-channel digital-to-analog converter array of the metasurface control matrix 130 is electrically connected to the bias voltage input terminal of the active tuning device 112 in the only metasurface unit 110 in the array through fine wiring. The programmable logic controller is responsible for receiving a digital instruction stream called a metasurface control word from the cognitive engine subsystem 300 via a high-speed serial peripheral interface bus. After receiving the control word, the controller parses it and distributes it to the multi-channel digital-to-analog converter array, which then converts these digital codes into the precise analog bias voltage sequence required to drive each active tuning device 112 in real time. In this way, the system can independently program the reflection phases of all units across the entire aperture at a response speed on the order of microseconds, thereby macroscopically reconstructing the radiation pattern of the cognitive reconfigurable metasurface aperture 100, including but not limited to precise pointing of the main lobe beam, forming a deep null in any specified direction, finely suppressing the sidelobe level, and changing the polarization of the antenna. In addition, the cognitive reconfigurable metasurface aperture 100 also includes a feeding network 120, which is a spatial feed (such as a horn antenna) or a network feeding structure for feeding the electromagnetic waves to be transmitted to the metasurface array, or collecting the received electromagnetic wave energy from the array and coupling it to the subsequent RF link.

[0029] The feed network 120 is electrically connected to the broadband multi-channel receiver subsystem 200. The core task of the broadband multi-channel receiver subsystem 200 is to capture and digitize the electromagnetic environment signals within the space where the cognitive reconfigurable metasurface aperture 100 is located, covering the entire operating frequency band of the system, indiscriminately and with high fidelity, providing the raw data foundation for subsequent cognitive analysis. In a specific embodiment, the broadband multi-channel receiver subsystem 200 integrates a low-noise amplifier covering the target operating frequency band, a power distribution network, a multi-channel downconversion module, and an analog-to-digital converter array. The signal processing flow is as follows: The broadband RF signal collected by the feed network 120 enters the low-noise amplifier for initial amplification to overcome the noise effects of the system front end and improve detection sensitivity. The low-noise amplifier has a gain of no less than 25 decibels and a noise figure of no more than 1.2 decibels to ensure effective capture of weak signals. The amplified signal then enters the power distribution network and is evenly distributed to multiple parallel processing channels. In each channel, the multi-channel downconversion module utilizes a set of local oscillators and mixers to downconvert signals from different frequency bands or the entire broadband to a unified intermediate frequency (IF) suitable for digital processing. Finally, the IF signals from each channel are fed into an analog-to-digital converter array for synchronous sampling and quantization. To fully characterize the electromagnetic environment, the performance of the analog-to-digital converter array is crucial. In a preferred embodiment, the sampling rate is set to no less than 40 GS / s, and the quantization bit count is set to no less than 14 bits, ensuring that the system can capture signals with a wide bandwidth and high dynamic range, thereby generating a broadband digital IF signal stream that can fully and accurately characterize the current electromagnetic environment and transmit it to the cognitive engine subsystem 300 for processing.

[0030] See also Figure 4 In order to achieve functional decoupling and efficient processing, the cognitive engine subsystem 300 is structurally divided into three collaborative modules: a real-time spectrum cognition module 310, a situation assessment and prediction module 320, and a resource optimization and configuration module 330. The cognitive engine subsystem 300 is the intelligent decision-making center of the present invention, which is used to receive massive digital signal streams from the broadband multi-channel receiver subsystem 200. Through a series of complex signal processing and artificial intelligence algorithms, it performs environmental understanding, situation assessment, and threat prediction, and ultimately generates optimal control instructions for reconstructing the physical properties of the cognitive reconfigurable metasurface aperture 100.

[0031] Furthermore, the real-time spectrum awareness module 310 is the first step in the entire awareness process, placing extremely stringent requirements on processing speed. Therefore, the real-time spectrum awareness module 310 is typically implemented on a field-programmable gate array (FPGA) hardware platform to fully leverage the FPGA's inherent parallel computing and pipeline processing capabilities. The real-time spectrum awareness module 310 directly receives the digital intermediate frequency (IF) signal stream from a high-speed analog-to-digital converter array and performs a series of operations using its integrated, highly optimized digital signal processing core. First, a polyphase filter bank or an overlap-add fast Fourier transform (FFT) algorithm is used to perform high-resolution channelization and spectral analysis on the wideband signal, generating an environmental spectrum map that reflects spectrum occupancy in real time. The module then executes an adaptive threshold decision algorithm based on energy detection. This algorithm dynamically adjusts the decision threshold based on the changing background noise level, accurately detecting all existing independent signal entities on the spectrum map. Once a signal entity is detected, the module immediately invokes a built-in parameter estimation algorithm library to extract multi-dimensional physical parameters for each signal entity. This algorithm library is modular and configurable based on mission requirements. Typical components include: a digital beamforming algorithm or a super-resolution multiple signal classification algorithm for accurately estimating signal arrival angle; a centroid algorithm for estimating signal center frequency and bandwidth; an integration algorithm for calculating signal power; and a classification algorithm based on high-order cumulants or cyclic spectrum features for identifying signal modulation type and polarization state. Ultimately, the real-time spectrum awareness module 310 encapsulates the full set of physical parameters extracted for each identified signal entity into a structured, machine-readable data packet, known as the environment state vector. In a specific embodiment, the data structure of the environment state vector is deterministically defined to include the following fields: signal unique identifier, signal acquisition timestamp, center frequency, signal bandwidth, signal average power, arrival angle azimuth, arrival angle elevation, polarization state code, and a preliminary signal type classification label. This module continuously generates a stream of environment state vectors and outputs them to the upper-layer situation assessment and prediction module 320.

[0032] The situation assessment and prediction module 320 receives the environmental state vector data stream generated by the real-time spectrum awareness module 310. It periodically receives the environmental state vector stream and performs the following core functions: First, the module maintains a multidimensional, time-evolving electromagnetic environment model in its internal large-capacity memory. Each newly received environmental state vector is used to update this model, thereby constructing a dynamic, historically traceable battlefield electromagnetic situation map. Second, based on a preset, user-definable threat assessment rule base, the situation assessment and prediction module 320 identifies and classifies all signal entities in the environmental model, distinguishing them into target communication signals, non-cooperative civilian signals, interference signals with potential or definite threats, and background noise. Subsequently, based on multiple factors such as the interference signal's power, frequency proximity to the target's communication frequency, and spatial threat level, the module comprehensively calculates the threat level of all interference sources and generates a global threat assessment matrix. The situation assessment and prediction module 320 then activates a separate Kalman filter for each signal entity identified as a mobile interference source, performing real-time tracking and trajectory prediction of its angle of arrival (AoA). In a specific embodiment, the Kalman filter's state vector is defined as six dimensions, including the interference source's azimuth and elevation angles, as well as the first-order time derivatives of these two angles (i.e., angular velocity). By continuously iterating the state transition equations and measurement update equations, the Kalman filter not only smooths out noise but also outputs an accurate prediction of the interference source's position within the next decision cycle.

[0033] The resource optimization and allocation module 330 is the highest decision-making layer of the cognitive engine subsystem 300, typically running on a high-performance central processing unit (CPU). It serves as the ultimate decision maker and executor. The resource optimization and allocation module 330 receives the threat assessment matrix and predicted environmental state from the situation assessment and prediction module 320 as environmental inputs. It also receives the current communication link quality requirements (e.g., target minimum data rate, maximum tolerable bit error rate) from the communication transceiver subsystem 400 as task constraints.

[0034] The core function of this module is to solve a multi-objective optimization problem to generate the optimal metasurface aperture configuration. The objective function of this optimization problem is defined as maximizing the signal-to-interference-and-noise ratio (SINR) of the target communication link while minimizing the pattern gain in the current and predicted arrival angles of all high-threat interference sources to below a preset deep nulling threshold. The communication mission requirements (such as target data rate and maximum tolerable bit error rate) serve as constraints in the optimization process.

[0035] To solve this non-convex optimization problem, the resource optimization and configuration module 330 employs a genetic algorithm (GA). The detailed implementation of the genetic algorithm is as follows: Chromosome encoding: Each chromosome represents a complete set of phase control value sequences covering all the metasurface units 110; Preferably, the phase control value of each metasurface unit 110 is encoded as a fixed bit width (e.g., 8 bits or 16 bits) digital integer, which corresponds to a quantized phase value in the range of 0 to 2π (or -π to π). For example, if 8-bit encoding is used, the phase of each unit can be quantized to 256 discrete values; Therefore, a chromosome is a sequence formed by concatenating N such digital phase values, where N is the total number of metasurface units 110.

[0036] Initial population generation: At the beginning of the algorithm, an initial population containing P chromosomes is randomly generated, where P is the population size (e.g., P can be set to 50 to 200); The phase control values in each chromosome are randomly initialized within their allowed quantization range.

[0037] An adaptive function is used to evaluate the pros and cons of each chromosome. In the present application, the fitness function is directly determined by the predicted signal-to-interference-and-noise ratio (SINR) value calculated based on the phase configuration represented by the chromosome. The specific calculation process is as follows: For a given chromosome (i.e., a set of phase control values), first calculate the radiation pattern of the cognitive reconfigurable metasurface aperture 100 formed by the phase configuration through an electromagnetic simulation model or a pre-established lookup table.

[0038] In combination with the predicted environmental state and threat assessment matrix provided by the situation assessment and prediction module 320, obtain the predicted angle of arrival (AoA) and power of the target communication signal, as well as the predicted angles of arrival and powers of all known and predicted interference sources.

[0039] According to the radiation pattern, extract the gain in the direction of the target signal AoA (G_t) and the gain in the direction of all interference source predicted angles of arrival (G_i).

[0040] The calculation formula of the predicted SINR is: SINR = (P_t * G_t) / (Σ(P_i * G_i) + P_n) where P_t is the target signal power, P_i is the i-th interference signal power, and P_n is the system noise power.

[0041] In order to better reflect multi-objective optimization, the fitness function can further introduce a penalty term. For example, if the gain in a certain interference direction fails to reach the deep nulling threshold, the fitness value will be penalized.

[0042] Genetic Operators: Selection: Preferably, a tournament selection method is used. K chromosomes (e.g., K = 3) are randomly selected from the current population, and the chromosome with the highest fitness is chosen as the parent. This process is repeated until a sufficient number of parents are selected. Alternatively, an elite retention strategy can be combined to directly copy the chromosomes with the highest fitness in the current population to the next generation to prevent the optimal solution from being lost during iteration.

[0043] Crossover: Preferably, a two-point crossover is used. A crossover operation is performed on the selected parent chromosomes with a preset crossover probability (e.g., 0.7 to 0.9). Two crossover points are randomly selected on each chromosome, and the gene segments between these two points are exchanged, generating two new daughter chromosomes.

[0044] Mutation: Perform mutation operations on newly generated offspring chromosomes with a preset mutation probability (e.g., 0.01 to 0.05). For each chromosome, one or more gene positions (i.e., the phase value of a hypersurface unit) are randomly selected and randomly changed within the allowed quantization range, or a small random perturbation (e.g., Gaussian perturbation) is added to introduce new features to increase population diversity and avoid falling into local optimality.

[0045] Termination conditions: The algorithm continues to iterate and generate new populations until any of the following conditions is met: Reach a preset maximum number of iterations (e.g., 100 to 500 generations).

[0046] The fitness value of the chromosome with the highest fitness in the population does not increase significantly within several consecutive generations (for example, 10 to 20 generations), indicating that the algorithm has converged.

[0047] Achieve the preset target SINR value or null depth requirement.

[0048] When the algorithm converges or reaches a preset number of iterations, the optimal chromosomes obtained from the population are decoded into a set of precise digital phase values. This set of values ​​is ultimately packaged into the aforementioned metasurface control word (MCW) and sent via a high-speed serial peripheral interface bus to the metasurface control matrix 130 of the cognitive reconfigurable metasurface aperture 100. The control matrix 130 converts the digital phase values ​​into the precise analog bias voltage sequence required to drive the active tuning devices 112 within each metasurface unit 110, thereby driving the adaptive reconstruction of the physical aperture.

[0049] After the cognitive reconfigurable metasurface aperture 100 completes the optimization and reconstruction, the communication transceiver subsystem 400 can use the communication channel purified by spatial filtering to efficiently receive the target communication signal through the main lobe beam of the directional pattern, or accurately transmit the signal in the target direction. On the other hand, it maintains communication with the resource optimization and configuration module 330 of the cognitive engine subsystem 300. In the process of performing communication tasks, the communication transceiver subsystem 400 will continuously monitor the actual performance indicators of the communication link, such as real-time bit error rate and signal-to-noise ratio, and feed back these quantified performance parameters to the resource optimization and configuration module 330 in real time. This feedback information is crucial. It is not only used to verify the actual effectiveness of the current aperture configuration online, but also serves as the basis for dynamically adjusting the objective function weights or constraints in subsequent optimization cycles, thereby truly realizing closed-loop adaptive optimization of system performance.

[0050] Based on the above system architecture, the present invention also provides an optimization method for an anti-interference microwave communication system. Figure 5 ,This method is executed cyclically by the above system, including a complete ,closed-loop process from perception to execution to feedback.

[0051] Step 1: Environmental Detection and Broadband Digitization. In this step, the system activates the broadband multi-channel receiver subsystem 200. Through the cognitive reconfigurable metasurface aperture 100 (which can initially be set to omnidirectional reception or wide-area scanning mode), it continuously captures electromagnetic environmental signals covering the entire operating frequency band within the system's space. The captured analog signals are then fed into a high-speed analog-to-digital converter array and converted into a high-fidelity broadband digital intermediate frequency signal stream.

[0052] Step 2: Real-time Spectrum Cognition and State Vector Generation. The broadband digital IF signal stream generated in the previous step is fed into the real-time spectrum cognitive module 310 within the cognitive engine subsystem 300. Leveraging its FPGA-based parallel pipeline processing architecture, the real-time spectrum cognitive module 310 performs high-resolution spectrum analysis on the signal stream, detecting each individual signal entity in the spectrum and accurately extracting a complete set of physical parameters for each entity, including frequency, power, angle of arrival, polarization state, and signal type. Ultimately, this set of parameters is encapsulated into a structured environmental state vector and output in real time.

[0053] Step 3: Situation Assessment and Dynamic Threat Prediction. The series of environmental state vectors generated in Step 2 are fed into the Situation Assessment and Prediction module 320. This module uses this new data to update its internal time-series electromagnetic environment model. It then identifies and prioritizes threats for all signal entities, generating a global threat assessment matrix. Simultaneously, for identified mobile interference sources, the module activates a Kalman filter to track their trajectory and predict their angle of arrival for the next time period. For intelligent interference sources, a recurrent neural network is used to analyze their historical behavior patterns. The final output of this step is a predicted environmental state that includes current and future threat information.

[0054] Step 4: Multi-Objective Optimization and Aperture Configuration Strategy Generation. The threat assessment matrix and predicted environmental state generated in the previous step, along with performance metrics defined by the upper-layer communication task (e.g., target bit error rate), are fed into the resource optimization and configuration module 330. Using these inputs, the module constructs a multi-objective optimization model centered on maximizing the target signal-to-interference-and-noise ratio (SINR) and constrained by suppressing all known and predicted interference. The module then initiates a genetic algorithm to efficiently solve the model until it finds a set of optimal phase control values ​​that can be applied to each metasurface unit 110 in the cognitive reconfigurable metasurface aperture 100. This set of optimal phase values ​​is encoded as a metasurface control word.

[0055] Step 5: Adaptive Reconfiguration of Aperture Physical Properties. The resource optimization and configuration module 330 transmits the metasurface control word generated in the previous step via a high-speed bus to the metasurface control matrix 130 of the cognitive reconfigurable metasurface aperture 100. The metasurface control matrix 130 immediately decodes it into a set of precise analog bias voltages and synchronously applies them to each metasurface unit 110 in the metasurface array. This process physically and instantaneously changes the electromagnetic scattering characteristics of the entire aperture, allowing the main lobe of its radiation pattern to be precisely aligned in the direction of the target signal, while simultaneously forming a specified depth null in all known and predicted interference directions.

[0056] Step 6: Communication Execution and Closed-Loop Feedback. After the cognitive reconfigurable metasurface aperture 100 is reconstructed and the spatial channel is "cleaned," the communication transceiver subsystem 400 immediately executes the communication task through this optimized channel. During the communication process, the communication transceiver subsystem 400 continuously monitors the actual performance parameters of the communication link, such as bit error rate and signal-to-noise ratio, and provides real-time feedback of this measured data to the resource optimization and configuration module 330 of the cognitive engine subsystem 300. This feedback information is used to verify the effectiveness of the current aperture configuration and serves as a basis for dynamically adjusting the objective function weights or constraints in subsequent optimization cycles. After this step is completed, the system immediately returns to the first step and continuously repeats the entire process, forming a complete and uninterrupted adaptive closed loop of cognition-decision-execution-feedback, ensuring that the communication system always maintains optimal anti-interference performance and communication efficiency in a dynamically changing and complex electromagnetic environment. Example

[0057] In order to more specifically illustrate the technical effects of the technical solution of the present invention, a specific application scenario will be described in detail below. Figure 8 The scenario involves a drone platform performing a high-data-rate image transmission mission in a complex urban canyon environment. Its data link operates in the X-band. In this environment, there are two typical interference signals: one is located on top of a building on the ground, a broadband noise source with a fixed direction but a broad spectrum (Source 1); the other is an interference source from an unknown aircraft in the adjacent airspace (Source 2).

[0058] In this embodiment, the specific engineering specifications of the cognitive reconfigurable metasurface aperture 100 are as follows: its core is a planar array of 48x48 metasurface units 110, totaling 2,304 independently controllable phase control units. The system operates at a center frequency of 10 GHz. The metasurface units 110 utilize a symmetrical "I"-shaped metal patch structure, with a gallium arsenide hyperabrupt junction active tuning device 112 precisely integrated at the center break of the patch. When a reverse bias voltage of -10 volts to 0 volts is applied, the diode's junction capacitance can continuously vary between 0.2 pF and 2.5 pF. This range provides a dynamic adjustment range of over 340 degrees of reflection phase for each unit at a 10 GHz operating frequency. The dielectric substrate 140 is made of polytetrafluoroethylene (PTFE), a material with excellent high-frequency characteristics, with a relative dielectric constant of 2.2 and a loss tangent of 0.0009, to minimize dielectric loss.

[0059] The broadband multi-channel receiver subsystem 200 is configured to continuously monitor a wide frequency band from 8 GHz to 12 GHz. The analog-to-digital converter array utilizes a four-channel synchronous sampling architecture, with each channel achieving an effective sampling rate of up to 40 GS / s and a 14-bit quantization bit count, ensuring comprehensive, high-fidelity capture of environmental signals across the entire X-band.

[0060] The hardware configuration of the cognitive engine subsystem 300 adopts a heterogeneous computing architecture to achieve optimal performance: the real-time spectrum recognition module 310 is implemented on a Xilinx RFSoC series FPGA, which integrates a high-performance analog-to-digital converter array and powerful programmable logic resources, making it an ideal choice for performing front-end real-time signal processing; the situation assessment and prediction module 320 is implemented on an NVIDIA Jetson AGX Xavier embedded GPU module, leveraging its powerful parallel computing capabilities to execute complex situation analysis and machine learning algorithms; and the resource optimization and allocation module 330 runs on an Intel Core i9 series multi-core processor, using its high main frequency and powerful general computing capabilities to quickly solve optimization problems.

[0061] In this scenario, the optimization method of the present invention is specifically performed according to the following process: In the first step, after the system is powered on, the cognitive reconfigurable metasurface aperture 100 is first configured in the initial state of omnidirectional reception, and the broadband multi-channel receiver subsystem 200 begins to continuously collect electromagnetic signals in the 8-12 GHz frequency band and digitize them into data streams.

[0062] In the second step, the real-time spectrum awareness module 310 running on the FPGA receives the digital signal stream. Its internal polyphase filter bank core outputs the environmental spectrum in real time with a frequency resolution of 10 kHz. The energy detection logic quickly identifies three significant signal entities in the spectrum: the drone's own telemetry, control, and image transmission uplink signal, a friendly target signal, located at 10.1 GHz; a strong noise signal (interference source 1) incident from the ground (-30° azimuth, -75° elevation) occupying the 9.8-9.9 GHz frequency band; and a signal (interference source 2) emitting rapid, irregular hopping within the 10.2-11.5 GHz frequency band from the air (+60° azimuth, +10° elevation). The real-time spectrum awareness module 310 generates a fully parameterized environmental state vector for each of these three signals and outputs it to the upper layer.

[0063] In the third step, the situation assessment and prediction module 320 running on the image processor receives the environmental state vector. By comparing it with a pre-stored friendly signal feature library, the module identifies the 10.1 GHz signal as the target communication signal. Based on its power and bandwidth characteristics, the ground noise signal is identified as threat source No. 1. Based on its broadband, fast frequency hopping behavior, the airborne signal is identified as intelligent jamming threat source No. 2. The module immediately initializes a six-dimensional Kalman filter for threat source No. 2, using subsequent continuously incoming arrival angle information to track its trajectory and, combined with its current motion state, predicts its possible spatial position within the next 50 millisecond decision time window. See [Note: The original text appears to be corrupted and should be omitted.] Figure 9 This diagram schematically illustrates how a Kalman filter smooths the trajectory points (measurements) of a mobile interference source and generates a predicted location. Simultaneously, the LSTM network in the module analyzes the frequency hopping pattern of source 2 for subsequent strategy identification. Ultimately, the module generates a threat assessment matrix containing the specific parameters and predicted locations of two high-threat interference sources.

[0064] In the fourth step, the resource optimization and configuration module 330 running on the image processor receives the threat assessment matrix and predicted environmental conditions, as well as the communication quality constraint of "bit error rate below 10^-6" required for the drone's image transmission mission. The module then constructs a genetic algorithm task: the chromosome length is 2304, with each gene representing the phase value of a metasurface unit (quantized to 8 bits, corresponding to 256 phase states). The fitness function is set as: the antenna gain in the direction of the target signal minus the weighted gain in the direction of threat source #1 minus the weighted gain in the predicted direction of threat source #2. The genetic algorithm is initiated and, after approximately 200 generations of population evolution, which takes approximately 30 milliseconds, converges to the optimal solution, namely, the optimal phase configuration for the 2304 units. This configuration is encoded into the metasurface control word.

[0065] In the fifth step, the metasurface control word is written to the metasurface control matrix 130 at high speed via a high-speed serial peripheral interface bus. The digital-to-analog converter array refreshes the bias voltage of all 2304 varactor diodes within a few microseconds. The physical properties of the cognitive reconfigurable metasurface aperture 100 are instantly reconfigured. Figure 6 and Figure 7 , is sensitive to interference sources 1 and 2 coming from different directions. Figure 7 In the image, a high-gain main lobe pointing to our ground station is formed at 10.1 GHz. At the same time, precise nulls with a depth of more than 40 decibels are formed in the fixed direction of interference source 1 [azimuth angle -30°, pitch angle -75°] and the predicted position direction of interference source 2.

[0066] In the sixth step, the communication transceiver subsystem 400 began high-speed image transmission using this "cleaned" spatial channel. Its internal bit error rate monitoring module measured the actual link bit error rate to be stable at approximately 10^-7, exceeding the mission requirement of 10^-6. This performance parameter was fed back to the cognitive engine subsystem 300 in real time. The system then seamlessly entered the next cognitive cycle, re-probing the environment to respond to the continued maneuvers of Threat Source No. 2, thereby achieving continuous, proactive, and intelligent adaptation to the dynamic and complex electromagnetic environment.

[0067] Comparative Example 1 To further highlight the technical advantages of the present invention, a comparative example is provided. Comparative Example 1 employs a conventional multi-beam antenna system. This system utilizes traditional digital beamforming (DBF) technology, capable of forming a main beam directed toward the target signal and adaptively nulling interference from a known direction and fixed frequency. However, this system lacks the broadband environmental awareness capabilities described in the present invention. Its operating bandwidth is limited to the communications frequency band, and it lacks a cognitive engine for dynamic tracking, prediction, and behavioral analysis of interference sources.

[0068] In a drone urban canyon scenario identical to Example 1, the system in Comparative Example 1 was able to detect the 10.1 GHz target signal and form a main beam. It also detected a portion of the in-band energy from a fixed-direction broadband noise interference source (Source 1) and attempted to place a null in its center direction (-30° in azimuth, -75° in elevation). However, due to the lack of broadband panoramic perception, its understanding of Source 1's spectral characteristics was incomplete, and the nulling effect was limited. More importantly, the system in Comparative Example 1 was completely unable to effectively address Source 2, which rapidly hops and maneuvers within the 10.2-11.5 GHz frequency band. Its narrowband receiver was unable to capture out-of-band frequency-hopping signals and had no way of knowing their presence. Even if some of the frequency-hopping signal fell within the band, its static nulling algorithm was unable to track the rapid maneuvers and frequency changes of the interference source. Therefore, when Source 2's frequency hops into the communication band, the communication link suffers a severe impact.

[0069] Performance indicators Example 1 (Invention) Technical effect description SINR at the receiving end -3.5 dB +18.2 dB The invention significantly improves SINR by accurately predicting and deeply suppressing double interference. BER of the communication link About 5x10⁻² (communication interruption) High SINR directly translates into extremely low BER, ensuring successful task execution. <![CDATA[约 1x10⁻ 7 (High-quality communication)]]> Null depth for interference source 1 28 dB 43 dB The invention's wideband awareness and optimization algorithm achieves a deeper null. Null depth for interference source 2 Cannot form an effective null 41 dB (at the predicted location) The invention's unique prediction and tracking capability is the key to dealing with intelligent dynamic interference. Adaptation time to dynamic interference N / A (cannot adapt) The complete cognitive closed loop ensures sub-second fast adaptive response. < 50 ms ​ It can be clearly seen from the data in Table 1 that the anti-interference microwave communication system and its optimization method described in the present invention, compared with traditional technologies, show overwhelming advantages when facing complex, dynamic, and intelligent interference environments. The fundamental reason for this is that the present invention deeply integrates the cognitive reconfigurable metasurface aperture of the physical layer with a cognitive engine with perception, evaluation, prediction, and decision-making capabilities, and constructs a high-speed closed-loop adaptive optimization process, thereby upgrading the communication system from a passive signal transceiver to an intelligent entity that can actively understand, predict, and reshape its electromagnetic environment. This paradigm shift from static defense to cognitive adaptation is the core of the technical solution of the present invention, making it extremely technologically advanced and widely applicable to military and civilian applications.

[0070] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. An anti-interference microwave communication system, comprising a cognitive reconfigurable metasurface aperture (100), a broadband multi-channel receiver subsystem (200), and a communication transceiver subsystem (400), wherein: The cognitive reconfigurable metasurface aperture (100) is used to interact with electromagnetic waves and reconstruct its electromagnetic response characteristics according to control instructions. The broadband multi-channel receiver subsystem (200) is connected to the cognitive reconfigurable metasurface aperture (100) and is used to capture and digitize broadband electromagnetic environment signals. The communication transceiver subsystem (400) is connected to the cognitive reconfigurable metasurface aperture (100) and is used to perform communication tasks after the cognitive reconfigurable metasurface aperture (100) is reconstructed. The system is characterized in that it further includes: a cognitive engine subsystem (300), wherein the cognitive engine subsystem (300) is in communication with the broadband multi-channel receiver subsystem (200), the cognitive reconfigurable metasurface aperture (100), and the communication transceiver subsystem (400); The cognitive engine subsystem (300) is configured to receive and process the digital signal representing the current electromagnetic environment output by the broadband multi-channel receiver subsystem (200), evaluate and predict the electromagnetic environment situation based on the processing result, and generate optimal control instructions for driving the cognitive reconfigurable metasurface aperture (100) to reconstruct the physical characteristics in combination with the communication task requirements provided by the communication transceiver subsystem (400) to suppress interference signals and enhance target signals in the spatial domain.

2. The anti-interference microwave communication system according to claim 1, characterized in that: The cognitive reconfigurable metasurface aperture (100) comprises: A two-dimensional array consisting of a plurality of metasurface units (110), wherein the metasurface units (110) are arranged on a dielectric substrate (140), and each metasurface unit (110) is composed of a metal patch (111) of a specific geometric shape and an active tuning device (112) embedded therein, wherein the active tuning device (112) is used to change the electromagnetic response characteristics of the metasurface unit (110) in which it is located; A metasurface control matrix (130) is connected to the cognitive engine subsystem (300) and is used to receive the optimal control instruction and convert the instruction into a precise driving signal applied to the active tuning device (112), thereby independently controlling the reflection phase or transmission phase of each metasurface unit (110), thereby reconstructing the radiation pattern of the entire aperture, including the main lobe direction, the null position, the side lobe level and the polarization mode.

3. The anti-interference microwave communication system according to claim 1, characterized in that: The broadband multi-channel receiver subsystem (200) includes: a low noise amplifier for initially amplifying a broadband radio frequency signal received by the cognitive reconfigurable metasurface aperture (100); a multi-channel down-conversion module connected to the output end of the low-noise amplifier and used to down-convert the amplified broadband radio frequency signal to a unified intermediate frequency; and a high-speed analog-to-digital converter array connected to the output end of the multi-channel down-conversion module and used to synchronously sample and quantize the intermediate frequency signal of each channel to generate a broadband digital intermediate frequency signal stream with a high dynamic range that can fully characterize the current electromagnetic environment, and output it to the cognitive engine subsystem (300).

4. The anti-interference microwave communication system according to claim 1, characterized in that: The cognitive engine subsystem (300) further includes: A real-time spectrum recognition module (310) is used to process the broadband digital intermediate frequency signal stream in real time to detect signal entities in the environment and extract their physical parameters to generate an environment state vector; A situation assessment and prediction module (320) is used to receive the environmental state vector, perform threat identification and level classification on signal entities, and track and predict the behavior of dynamic interference sources to generate a global threat assessment matrix and predict environmental state; The resource optimization and configuration module (330) is used to receive the global threat assessment matrix, the predicted environmental state and the communication task requirements, and generate the optimal control instructions by optimizing multiple objectives.

5. The anti-interference microwave communication system according to claim 4, characterized in that: The real-time spectrum recognition module (310) performs high-resolution spectrum analysis on the broadband digital intermediate frequency signal stream by using an overlap-add fast Fourier transform algorithm to generate a real-time environmental spectrum graph; Detecting independent signal entities on the real-time environmental spectrum using an adaptive threshold decision algorithm based on energy detection; and for each detected signal entity, calling a set of parameter estimation algorithm libraries to extract its key physical parameters, the key physical parameters including at least angle of arrival, center frequency, signal bandwidth, signal power, and polarization state; The full set of parameters of each signal entity is encapsulated into a structured environment state vector and output to the situation assessment and prediction module (320).

6. The anti-interference microwave communication system according to claim 5, characterized in that: The situation assessment and prediction module (320) maintains a multi-dimensional, time-series evolving electromagnetic environment model in an internal memory, and continuously updates the model using the received environment state vector flow; Based on a preset threat assessment rule library, signal entities in the electromagnetic environment model are divided into target communication signals, interference signals, and other signals, and the threat level of the interference signal is calculated according to its parameters to generate the global threat assessment matrix; For signal entities identified as mobile interference sources, a Kalman filter is activated to track their arrival angle and output a predicted value for the position of the interference source in the next decision cycle, which is included in the predicted environmental state; and for signal entities identified as intelligent interference sources, a long short-term memory recurrent neural network is used to analyze their historical behavior patterns to identify their interference strategies, and the identification results are used to update the global threat assessment matrix.

7. The anti-interference microwave communication system according to claim 6, characterized in that: The resource optimization and configuration module (330) constructs a multi-objective optimization problem, wherein the objective function of the optimization problem is defined as maximizing the signal to interference noise ratio of the target communication link, while minimizing the gain of the directional pattern in the current and predicted arrival angle directions of all high-threat-level interference sources to below a preset deep nulling threshold, and using the communication task requirements as constraints; and solves the multi-objective optimization problem using a genetic algorithm, wherein each chromosome represents a set of phase control values ​​applied to all metasurface units (110), and the fitness function is determined by the predicted signal to interference noise ratio value calculated based on the chromosome configuration; When the genetic algorithm converges, the obtained optimal chromosome is decoded into a set of precise digital phase values, and the set of digital phase values ​​is encapsulated into the metasurface control word and sent to the cognitive reconfigurable metasurface aperture (100).

8. The anti-interference microwave communication system according to claim 1, characterized in that: The communication transceiver subsystem (400) continuously monitors actual performance parameters of the communication link during the execution of the communication task, wherein the actual performance parameters include real-time bit error rate and signal-to-noise ratio; the resource optimization and configuration module (330) is used to feed back the actual performance parameters to the cognitive engine subsystem (300) in real time, and the feedback information is used to verify the validity of the current aperture configuration and serve as a basis for dynamically adjusting the objective function weight or constraint conditions in subsequent optimization cycles, thereby forming a complete cognitive closed loop.

9. An optimization method for an anti-interference microwave communication system, characterized in that: The following steps are involved: Environmental detection and broadband digitization, through the cognitive reconfigurable metasurface aperture (100), continuously captures the electromagnetic environment signals in the space where the system is located and converts them into high-fidelity broadband digital intermediate frequency signal streams; Real-time spectrum recognition and state vector generation: performing high-resolution spectrum analysis on the broadband digital intermediate frequency signal stream, detecting independent signal entities in the spectrum, accurately extracting the parameters of each signal entity, and encapsulating them into a structured environmental state vector for output. The parameters include frequency, power, and angle of arrival. Situation assessment and dynamic threat prediction: updating the internal time-series electromagnetic environment model, identifying and grading threats for all signal entities to generate a threat assessment matrix, and simultaneously tracking and predicting the trajectory of identified mobile or intelligent interference sources to form a predicted environmental state; Multi-objective optimization and aperture configuration strategy generation, constructing a multi-objective optimization model with maximizing the target signal signal-to-interference-noise ratio as the core and suppressing all known and predicted interferences as the constraint, and solving the model until a set of optimal phase control values ​​applied to each metasurface unit (110) in the cognitive reconfigurable metasurface aperture (100) is found, the set of phase control values ​​being encoded as a metasurface control word; Adaptively reconstructing the physical characteristics of the aperture, transmitting the generated metasurface control word to the cognitive reconfigurable metasurface aperture (100) to drive its physical characteristics to change so that the main lobe of its radiation pattern is aligned with the target signal direction, and at the same time forming a deep null in all known and predicted interference directions; Communication execution and closed-loop feedback: after the cognitive reconfigurable metasurface aperture (100) is reconstructed, the communication task is performed through the optimized spatial channel, while the actual performance parameters of the communication link are continuously monitored, and these parameters are fed back in real time for dynamic adjustment of the subsequent optimization process.

10. The method according to claim 9, characterized in that When conducting situation assessment and dynamic threat prediction, it also includes: For mobile interference sources, a Kalman filter is used to track their trajectory and predict their angle of arrival in the next time period based on their continuous angle of arrival measurements. Intelligent interference sources use long-short-term memory recurrent neural networks to identify their interference strategies by analyzing their historical frequency hopping sequences or sweep characteristics; The multi-objective optimization and aperture configuration strategy generation further includes: using a genetic algorithm to solve the multi-objective optimization model, encoding a sequence representing a complete set of phase control values ​​as a chromosome in the genetic algorithm, and using the predicted signal-to-interference-noise ratio calculated based on the set of phase configurations as the fitness function of the chromosome, and performing a global search through selection, crossover, and mutation genetic operators until the optimal solution is found.

Images