Unmanned aerial vehicle cluster polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction

Abstract

The invention provides an unmanned aerial vehicle cluster polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction, and the system integrates five functional subsystems, namely an environment perception subsystem, a topology reconstruction subsystem, a polarization regulation and control subsystem, a space-time coding subsystem and a resource scheduling subsystem, and carries out the unified coordination through a decision control center based on a multi-objective optimization algorithm. And closed-loop adaptive control is formed. The system can accurately sense interference source information, channel state and cluster dynamic state in real time; according to a sensing result and a task demand, dynamically optimizing and adjusting a communication topological structure of the cluster, and actively avoiding a strong interference area; meanwhile, the polarization state of the signal is adaptively selected and controlled so as to match a channel and suppress interference; and a reconfigurable space-time coding scheme is adopted, and coding parameters are flexibly adjusted to adapt to changing channel and interference conditions. And the communication anti-interference capability, the data transmission reliability and the energy utilization efficiency of the unmanned aerial vehicle cluster in a complex electromagnetic environment are obviously enhanced.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a polarization-space-time coding anti-interference communication system for an unmanned aerial vehicle cluster based on dynamic topology reconstruction. Background Art

[0002] In complex mission scenarios such as modern military conflict, disaster response, smart city infrastructure inspections, and refined environmental monitoring, unmanned aerial vehicle (UAV) swarms have become an indispensable key technology platform due to their distributed collaboration, wide-area coverage, and rapid deployment capabilities. The anti-interference performance of UAV swarm communication systems is a key element in ensuring stable system operation and efficient mission completion.

[0003] At present, there are some representative solutions to the anti-interference problem of drone cluster communications: dynamic spectrum access solutions based on cognitive radio, physical layer anti-interference solutions based on MIMO-OFDM, and distributed self-healing routing solutions based on mesh topology.

[0004] Although the above technologies have improved the anti-interference capabilities of drone swarms to a certain extent, in the face of increasingly complex battlefield electromagnetic environments and intelligent interference methods, there are still the following significant limitations: 1. Insufficient adaptability of communication strategies: In the face of fast-varying channels and agile and intelligent interference sources, the anti-interference effects of most existing solutions will deteriorate sharply, and they lack the ability to respond quickly and accurately. 2. Decoupling of network topology and physical layer anti-interference: Topology management (mainly focusing on connectivity and routing efficiency) and physical layer anti-interference technology (focusing on signal processing) are often designed as independent modules, lacking cross-layer collaborative design. As a result, it is impossible to dynamically adjust the topology structure to avoid interference areas according to the physical layer interference conditions, or to optimize the physical layer parameters according to topology changes, resulting in suboptimal configuration of resources (such as energy, spectrum, and computing power) and loss of overall anti-interference performance. 3. Insufficient utilization of the polarization dimension: Most existing solutions ignore or do not fully utilize the polarization information of electromagnetic waves. Even if some schemes involve polarization, they are mostly limited to static polarization diversity reception and fail to implement adaptive polarization control (at both the transmitter and receiver) based on the real-time interference polarization characteristics. This makes it impossible to effectively combat polarization-agile interference or utilize the polarization dimension for interference suppression. 4. Fixed space-time coding schemes: Once traditional space-time coding schemes (such as fixed-rate Alamouti codes) are selected, their coding structure and parameters are generally not changed. It is impossible to dynamically adjust the coding rate, diversity order, or switch between diversity and multiplexing based on real-time channel quality, interference type and intensity, and cluster topology connectivity, making it difficult to achieve the optimal balance between anti-interference and transmission efficiency in all scenarios. 5. Significant energy efficiency issues: Many traditional anti-interference methods, such as blindly increasing transmit power, increasing error correction coding redundancy, and maintaining a high-density connection topology, significantly increase the energy consumption of drone nodes. For drone platforms with limited energy reserves, especially those that require long-term hovering operations, high energy consumption becomes a key bottleneck restricting their sustained combat capabilities. 6. Computational complexity and real-time performance conflict: Some advanced anti-interference algorithms with superior theoretical performance (such as complex iterative interference cancellation and high-dimensional space-time-frequency joint processing) are computationally intensive, placing extremely high demands on the processing power and real-time performance of the UAV's onboard computing platform. Real-time processing is difficult to achieve on resource-constrained UAVs, resulting in actual deployment results far below theoretical expectations. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention aims to provide a polarization-space-time coding anti-interference communication system for drone swarms based on dynamic topology reconstruction to address the problems raised in the above background technology. The present invention can accurately perceive interference source information, channel status, and cluster dynamics in real time; dynamically optimize and adjust the cluster's communication topology based on the perception results and mission requirements, and actively avoid areas of strong interference; and adopt a reconfigurable space-time coding scheme to flexibly adjust coding parameters to adapt to changing channel and interference conditions. This system can significantly enhance the communication anti-interference capability, data transmission reliability, and energy utilization efficiency of drone swarms in complex electromagnetic environments.

[0006] In order to achieve the above-mentioned objectives, the present invention is implemented through the following technical solutions: a UAV cluster polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction, including an environmental perception subsystem, a topology reconstruction subsystem, an adaptive polarization control subsystem, a reconfigurable space-time coding subsystem, a multi-dimensional resource joint scheduling subsystem and a decision control center. The environmental perception subsystem is responsible for comprehensive and real-time perception of the communication environment in which the UAV cluster is located; the topology reconstruction subsystem dynamically optimizes the communication topology structure of the cluster according to the environmental perception results and task requirements; the adaptive polarization control subsystem dynamically adjusts the polarization state of the receiving and transmitting signals according to the polarization characteristics of the channel and interference; the reconfigurable space-time coding subsystem adaptively selects and configures the space-time coding scheme according to the channel quality, interference characteristics and business requirements; the multi-dimensional resource joint scheduling subsystem uniformly and efficiently coordinates and allocates the energy, computing and communication resources within the cluster under the premise of meeting task constraints.

[0007] Furthermore, the environmental perception subsystem includes a multi-mode interference perception and positioning module, a high-precision CSI estimation module and a cluster self-status monitoring module; the topology reconstruction subsystem includes a topology quality assessment module, an intelligent topology generation module, and a smooth topology execution and control module; the adaptive polarization regulation subsystem includes a polarization state real-time detection module, a polarization strategy optimization module, and a reconfigurable polarization antenna control module; the reconfigurable space-time coding subsystem includes a coding mode and parameter selection module, a joint polarization-space-time codeword generation module, and a coding and signal processing execution module; the multi-dimensional resource joint scheduling subsystem includes an energy efficiency perception and optimization management module, a distributed computing resource collaborative allocation module, and a cross-layer communication resource joint optimization module; the decision control center includes situational perception and fusion, strategy generation based on deep reinforcement learning, execution control and feedback.

[0008] Furthermore, the multi-mode interference perception and positioning module adopts a hybrid model based on deep learning, in which a convolutional neural network (CNN) is first used to extract static spatial features from the spectrum or constellation diagram of the received signal; then a recurrent neural network (RNN) or Transformer model is combined to process the time series of the static features to identify dynamic interference patterns such as sweeps or pulses; finally, the interference type probability is output through a classifier (including an SVM or fully connected layer).

[0009] Furthermore, the decision control center is implemented in a distributed or centralized manner, integrating information from each subsystem, running the core multi-objective optimization decision algorithm based on the global situation and preset task objectives, generating collaborative control instructions and sending them to each execution subsystem to achieve closed-loop intelligent regulation of the entire system. The multi-objective optimization decision algorithm includes a policy network based on deep reinforcement learning (DRL).

[0010] Furthermore, it also includes building a unified, closed-loop adaptive control framework across the physical layer and network layer to achieve collaborative, dynamic, and intelligent optimization of the three key degrees of freedom in drone cluster communications: topology structure, signal polarization state, and space-time coding mode, in order to cope with complex electromagnetic interference.

[0011] Furthermore, the energy topology generation module adopts a genetic algorithm combined with reinforcement learning (RL-GA), wherein the reinforcement learning agent dynamically adjusts the crossover operator, mutation operator and fitness function weight of the genetic algorithm according to the environmental perception results.

[0012] Furthermore, the coding mode and parameter selection module is used to maintain a performance-complexity database, storing the theoretical / simulated performance and computational complexity of different STC modes under different channel conditions and interference types, where the channel conditions include high / low SNR and correlated / non-correlated channels, and the theoretical / simulated performance includes BER and throughput.

[0013] Furthermore, the smooth topology execution and control module adopts an artificial potential field method or a model predictive control (MPC) algorithm to plan a smooth flight trajectory for each UAV to avoid collisions, so as to achieve a transition from the current topology to the target topology.

[0014] Beneficial effects of the present invention:

[0015] 1. This UAV swarm polarization-space-time coding (STCC) anti-interference communication system based on dynamic topology reconstruction intelligently and collaboratively controls network topology (interference avoidance), signal polarization (interference suppression), and space-time coding (diversity / error correction), forming a multi-dimensional, collaborative anti-interference mechanism at the network topology, signal physical dimension, and coding levels. Simulation and hardware-in-the-loop experiments demonstrate that, in the face of strong adaptive interference scenarios, this system reduces the average bit error rate (BER) of the communication link by more than two orders of magnitude, and increases the effective data throughput by approximately 3-5 times, compared to baseline systems using only traditional STBC or frequency hopping technology. This system is capable of maintaining effective connections even in environments where traditional solutions completely disrupt communications.

[0016] 2. The dynamic topology reconstruction mechanism of the UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction gives the network self-healing capabilities at the physical level, while adaptive polarization and space-time coding enhance the robustness of the link itself. Multi-layer collaboration ensures that even when some nodes fail or the link suffers from sudden strong interference, the system can still quickly adjust and maintain the continuity of critical services. Comparative experimental data show that in a simulated complex electromagnetic environment, the end-to-end packet delivery success rate (PDR) of this system is stably maintained at above 95%, while the PDR of traditional fixed strategy solutions may drop below 60%.

[0017] 3. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction uses precise environmental perception and intelligent resource scheduling. The system can dynamically allocate energy on demand, for example, reducing power when interference is low, adopting high-order modulation and low-redundancy coding; in the case of strong interference, it prioritizes avoiding / suppressing interference through topology and polarization adjustment rather than blindly increasing power. Comprehensive evaluation shows that when achieving the same communication reliability (such as BER <10 -4 ), the average energy consumption of the system of the present invention is reduced by 40%-60% compared with traditional power control or fixed coding schemes, significantly extending the combat radius and mission endurance of the UAV cluster.

[0018] 4. The DRL-based decision-making core of this invention enables the system to learn and adapt online to unknown or rapidly changing electromagnetic environments. The system's average response time to the appearance, disappearance, type change, or movement of interference sources is reduced to hundreds of milliseconds, significantly faster than traditional rule-based or slow-adaptive solutions. This rapid adaptive capability ensures stable system performance in dynamic, non-stationary, and complex environments.

[0019] 5. This invention balances performance and complexity in its algorithmic design. For example, it employs hierarchical decision-making (DRL is responsible for macro-strategy, while subsystems optimize specific parameters), distributed computing (some perception and optimization tasks are performed locally on nodes), and lightweight models (such as pruned neural networks). This ensures that the core algorithm can run in real time on the drone-mounted embedded platform, and that the end-to-end control latency meets the requirements of most tasks (such as real-time video transmission and collaborative control).

[0020] 6. This invention utilizes a clear functional module division and standardized subsystem interfaces, facilitating deployment, integration, and upgrades across diverse hardware platforms and drone clusters of varying sizes. The core optimization algorithm exhibits excellent scalability, adapting to changes in the number of nodes. Its effectiveness has been demonstrated in simulation clusters ranging from 4 to 64 nodes, with performance scaling significantly with the number of nodes. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a system architecture diagram of the present invention;

[0022] Figure 2 This is a timing diagram of the workflow of the closed-loop adaptive control system of the present invention;

[0023] Figure 3 A logic flow chart for dynamic topology reconstruction and anti-interference of the present invention;

[0024] Figure 4 Reconstruct a three-dimensional schematic diagram of the anti-interference effect for dynamic topology;

[0025] Figure 5 Schematic diagram of the principle and effect of adaptive polarization control;

[0026] Figure 6 This is a diagram showing the structure and performance advantages of reconfigurable polarization-space-time coding;

[0027] Figure 7 This is a diagram of the comprehensive performance evaluation and comparison experiment of the system of the present invention. DETAILED DESCRIPTION

[0028] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0029] Example 1

[0030] See also Figures 1 to 7 The present invention provides the following technical solution: a UAV cluster polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction. It mainly consists of the following components that cooperate with each other:

[0031] 1. Environmental perception subsystem: responsible for comprehensive and real-time perception of the communication environment in which the drone cluster is located.

[0032] Include:

[0033] 1.1 Multi-mode interference perception and positioning module: Integrates multiple detection methods (such as broadband perception based on energy detection, signal recognition based on cyclic spectrum characteristics, and positioning based on multi-UAV collaborative direction finding) to achieve accurate identification of interference source types (narrowband, broadband, swept frequency, pulse, noise frequency modulation, polarization interference, etc.) and estimation of spatial location.

[0034] 1.2 High-precision Channel State Information (CSI) Estimation Module: Utilizes pilot signals, feedback mechanisms, and possible channel reciprocity to accurately estimate channel parameters of the inter-node communication link, such as the channel gain matrix, phase, delay spread, Doppler shift, and polarization channel parameters (such as cross-polarization discrimination XPD).

[0035] 1.3 Cluster status monitoring module: collects and updates the status information of each drone node in the cluster, such as position, speed, attitude, remaining energy, available computing resources, current communication load, etc., in real time.

[0036] 2. Topology Reconstruction Subsystem: Responsible for dynamically optimizing the cluster's communication topology based on environmental perception results and task requirements. It includes:

[0037] 2.1 Topology Quality Assessment Module: This module quantitatively evaluates the overall performance of the current topology structure based on multi-dimensional indicators such as connectivity, algebraic connectivity (robustness), average path length (efficiency), link interference level, and node energy balance.

[0038] 2.2 Intelligent topology generation module: It adopts an improved heuristic optimization algorithm (such as genetic algorithm or ant colony optimization combined with reinforcement learning) to maximize the topology quality evaluation index. Under the constraints of connectivity, energy consumption, etc., it generates the optimal or suboptimal target topology structure (i.e., the connection relationship between nodes).

[0039] 2.3 Smooth topology execution and control module: Converts the generated target topology into specific drone motion control instructions (adjusting position to establish / disconnect links) and communication link management instructions (activating / deactivating logical links), and designs a smooth transition mechanism to avoid business interruption caused by drastic topology changes.

[0040] 3. Adaptive polarization control subsystem: Responsible for dynamically adjusting the polarization state of the transmitted and received signals based on the polarization characteristics of the channel and interference. It includes:

[0041] 3.1 Polarization state real-time detection module: Utilizes orthogonal dual-polarization or fully polarized antenna arrays to measure the polarization state of the received signal and potential interference signals in real time, usually described by Stokes parameters or polarization ellipse parameters.

[0042] 3.2 Polarization Strategy Optimization Module: Based on the detected signal and interference polarization states, channel polarization parameters (such as XPD), and interference suppression / signal enhancement goals, an optimization model for polarization selection is established to determine the optimal transmit polarization mode (such as linear polarization, circular polarization, elliptical polarization and their directions) and receive polarization matching / filtering strategy.

[0043] 3.3 Reconfigurable polarized antenna control module: By controlling the feeding network or structural parameters of the reconfigurable polarized antenna unit (such as antennas based on PIN diodes, MEMS switches or liquid crystal materials), the selected transmit / receive polarization state can be accurately achieved.

[0044] 4. Reconfigurable Space-Time Coding Subsystem: Responsible for adaptively selecting and configuring space-time coding schemes based on channel quality, interference characteristics, and service requirements. Includes:

[0045] 4.1 Coding Mode and Parameter Selection Module: This module maintains a library of various space-time coding schemes (e.g., STBC at different rates, V-BLAST with different diversity / multiplexing tradeoffs, and layered space-time code (LSTC). It selects the most appropriate coding mode and its parameters (e.g., coding rate and modulation order) based on the real-time CSI, the identified interference type (e.g., whether the interference is spatially correlated), the required data rate, and reliability requirements.

[0046] 4.2 Joint Polarization-Space-Time Codeword Generation Module: This module combines the selected space-time coding scheme with the polarization strategy to generate jointly optimized polarization-space-time coding codewords. For example, different polarization schemes can be used on different antennas, or information bits can be mapped to combined symbols in the space-time and polarization domains.

[0047] 4.3 Coding and signal processing execution module: implements real-time information mapping, coding, modulation, and signal distribution and transmission in multiple antennas and multi-polarization dimensions of the selected coding scheme.

[0048] 5. Multi-dimensional resource joint scheduling subsystem: responsible for unified and efficient coordination and allocation of energy, computing and communication (spectrum, power, time slot) resources within the cluster under the premise of meeting task constraints.

[0049] Include:

[0050] 5.1 Energy Efficiency Perception and Optimization Management Module: monitors the energy consumption of each node, predicts the remaining working time, and optimizes the flight path, communication power, and computing task allocation based on task priority and node contribution, to maximize the overall working life of the cluster or the working time of key nodes.

[0051] 5.2 Distributed computing resource collaborative allocation module: Dynamically allocates CPU, GPU, FPGA and other computing resources based on the computing needs and real-time requirements of the perception, decision-making, control and other algorithms carried by each node to ensure the timely completion of key tasks (such as real-time interference identification, topology optimization, encoding and decoding).

[0052] 5.3 Cross-layer communication resource joint optimization module: Collaboratively optimizes parameters of the physical layer (power, polarization, STC mode), MAC layer (time slot, frequency allocation), and network layer (routing selection) to maximize system throughput, minimize end-to-end delay, or maximize anti-interference margin.

[0053] 6 Decision control center (logic unit): as the "brain" of the system, usually distributed or centralized

[0054] This is achieved through the use of a ground station or central node. It is responsible for integrating information from each subsystem, running the core multi-objective optimization decision-making algorithm (such as a policy network based on deep reinforcement learning (DRL)) based on the global situation and preset mission objectives, and generating coordinated control instructions to be issued to each execution subsystem, thus achieving closed-loop intelligent control of the entire system.

[0055] The key to this invention's innovation lies in its ability to achieve, for the first time, a tightly coupled collaborative optimization of five previously independent anti-interference dimensions: environmental perception, topology, polarization state, space-time coding, and resource allocation. This results in a closed-loop adaptive control system. This cross-layer, cross-domain joint decision-making mechanism addresses the suboptimal nature of existing technologies, often caused by decoupling and insufficient coordination between modules. This results in a subtle improvement in overall performance.

[0056] This embodiment also provides the internal implementation mechanism and mutual relationship of the above system, which are as follows:

[0057] 1Environmental perception subsystem

[0058] 1.1 Multi-mode interference perception and positioning module:

[0059] 1.1.1 Interference Identification: A hybrid model based on deep learning is used. First, a convolutional neural network (CNN) is used to extract preliminary features from the received signal's spectrum, constellation diagram, or time-domain waveform. A recurrent neural network (RNN) or Transformer is then used to process the signal's time series characteristics, capturing dynamic interference patterns such as sweeps and pulses. Finally, a classifier (such as a support vector machine (SVM) or fully connected layer) is used to output the probability of the interference type.

[0060] Model f recog (S rx ;Θ DNN ) Output interference category I type , where S rx is the received signal sample, Θ DNN are the trained network parameters.

[0061] 1.1.2 Interference Location: For a specific interference source identified, an observation network is formed using at least three drones with direction-finding capabilities. A multi-station angle of arrival (AOA) / time difference of arrival (TDOA) / received signal strength difference (RSSD) fusion positioning algorithm based on maximum likelihood estimation (MLE) or extended Kalman filtering (EKF) is used. The positioning error covariance matrix ploc can be approximated by the inverse of the Fisher information matrix J: p loc ≈J —1 Fisher information matrix J is a function of the position coordinates of the interference source to be located, and its elements depend on the geometric position of each UAV node and the statistical variance of the measurement values ​​(AOA, TDOA, RSSD)

[0062] 1.2 High-precision CSI estimation module:

[0063] 1.2.1 Channel parameter estimation: Design orthogonal pilot sequences and estimate the multiple-input multiple-output (MIMO) channel matrix H(t)∈C using channel estimation techniques (such as least squares LS and minimum mean square error MMSE). Nr×Nt , where N t ,N r are the number of transmitting and receiving antennas, h ij (t) represents the complex channel gain from transmitting antenna j to receiving antenna i at time t.

[0064] 1.2.2 Polarization Channel Parameter Estimation: Using dual-polarization pilots, estimate the polarization channel matrix:

[0065]

[0066] where hxyrepresents the channel coefficient from y-polarized transmission to x-polarized reception (v: vertical, h: horizontal). The cross-polarization discrimination (XPD) is defined as the ratio of co-polarized received power to cross-polarized received power, and its calculation formula is:

[0067]

[0068] where h vv and h hh are the co-polarized channel coefficients for vertical-vertical and horizontal-horizontal, respectively, and h vh and h hv are the cross-polarized channel coefficients for horizontal-to-vertical and vertical-to-horizontal, respectively.

[0069] 1.3 Cluster Self-state Monitoring Module: Each UAV obtains its own state s self,i (t) = {p i (t), v i (t), E i (t), c load,i (t),...} through internal sensors (GPS / INS, power meter, CPU load monitor), and periodically broadcasts or reports to the decision center (or neighbor nodes) through the control channel.

[0070] 2 Topology Reconstruction Subsystem

[0071] 2.1 Topology Quality Evaluation Module: Define the comprehensive topology fitness function Fitness(G(t)), which is the weighted sum of multiple sub-indicators: Fitness(G) = w con f con (G) + w rob f rob (G) + w eff f eff (G) + w jam f jam (G) + w eng f eng (G)... f rob Evaluate robustness (such as the algebraic connectivity λ2(L G ) of the network, i.e., the second smallest eigenvalue of the graph Laplacian matrix L G ), f eff Evaluate communication efficiency (such as average hop count), f jam Evaluate interference resistance (such as average signal-to-interference-and-noise ratio (SINR) or the degree of avoiding interference areas), f eng Evaluate energy balance (such as the inverse of the variance of node residual energy). The weights w k can be dynamically adjusted according to task priorities.

[0072] 2.2 Intelligent Topology Generation Module: Modeling topology optimization as a constrained optimization problem:

[0073] max G ∈GFitneSS(G)subjectto:connectivity(G)≥c min ,

[0074] Maxpower(G)≤p limit ,colliSionAvoidance(G) is solved using an improved reinforcement learning guided genetic algorithm (RL-GA). In the improved RL-GA algorithm, the reinforcement learning agent dynamically adjusts the crossover rate and mutation rate of the genetic algorithm according to the current environmental state (such as interference distribution and node energy), and adjusts the weights (w k ) to accelerate algorithm convergence and escape local optima. The RL agent dynamically adjusts the GA's crossover and mutation operators and fitness function weights based on the current environmental state (interference distribution, node energy, etc.), accelerating convergence to a high-quality solution. G is the set of all feasible topologies.

[0075] 2.3 Smooth topology execution and control module: After calculating the target topology G, the target position p of each UAV i is generated target,i Or speed command v cmd,i It uses artificial potential field methods or model predictive control (MPC) to plan smooth and safe flight trajectories, while coordinating with the MAC layer to establish and tear down links.

[0076] 3 Adaptive polarization control subsystem

[0077] 3.1 Polarization state real-time detection module: The receiving end uses a fully polarized antenna to receive the signal and calculates the normalized Stokes parameters S = [S1, S2, S3] through the signal processor T ,in:

[0078]

[0079] Among them, I H ,I V ,I +45 ,I -45 , I R ,I L Represent the signal power received on the horizontal, vertical, +45°, -45°, right-hand, and left-hand polarization channels respectively; I represents the power of the corresponding polarization channel. The total power S0 used for normalization is usually the sum of the powers of the orthogonal components, for example, S0 = I H +I V .

[0080] 3.2 Polarization Strategy Optimization Module: The goal is to find the optimal transmit polarization p t and receiving polarization p r (expressed by Jones vector) to maximize the received signal-to-interference-and-noise ratio (SINR). The optimization problem can be expressed as:

[0081]

[0082] Among them, p t and p r are the Jones vectors for transmission and reception respectively; H pol is the polarization channel matrix of the desired signal; P s is the signal transmission power; N0 is the noise power; K is the number of interference sources; P I,k 、 and are the power of the kth interference source, the polarization channel matrix, and the transmit polarization, respectively. This optimization can be solved by eigenvalue decomposition or gradient-based optimization methods.

[0083] 3.3 Reconfigurable polarized antenna control module: According to the optimization result P t or P r , calculate the control signals required to drive the reconfigurable antenna (such as the bias voltage of the PIN diode and the MEMS switch state), and achieve precise control of the polarization ellipse parameters (axis ratio AR, tilt angle τ).

[0084] 4Reconfigurable Space-Time Coding Subsystem

[0085] 4.1 Coding Mode and Parameter Selection Module: Maintains a performance-complexity database that stores the theoretical / simulated performance (BER, throughput) and computational complexity of different STC modes (such as Alamouti, GoldenCode, LSTC) under different channel conditions (high / low SNR, correlated / uncorrelated channels) and interference types. Based on real-time estimated CSI, SINR, interference characteristics I type and service QoS requirements R req , select the utility function U(Mode,paramSIcSI,SINR,I type ,R req ) Maximum modes and parameters.

[0086] 4.2 Joint Polarization-Space-Time Codeword Generation Module: Mapping Information Symbols S1, S2, ... to Joint Polarization-Space-Time Codewords Where Nt is the number of physical antennas, T is the coding time block length, and p is the number of polarization states used (usually 2). For example, a 2×2×2 polarization-space-time code: (vertical polarization), (Horizontal polarization) Here, 4 symbols are transmitted simultaneously, utilizing both space-time and polarization dimensions.

[0087] 4.3 Encoding and signal processing execution module: the generated X PSTC The codeword passes through the modulator, pulse shaping filter, and drives N t The receiver then performs a corresponding joint polarization-space-time decoding algorithm (such as detection based on the maximum likelihood criterion).

[0088] 5. Multi-dimensional resource joint scheduling subsystem

[0089] 5.1 Energy Efficiency Perception and Optimization Management Module: Establishing Node Energy Consumption Model E consumed,i (t)= Efly,i (t)+E comm,i (p tx ,Mode)+E comp,i (Task). Based on the remaining energy E rem,i (t) and task value V i , using Lyapunov optimization or prediction-based control methods to dynamically adjust the transmit power ptx, communication mode Mode, computing task offloading decisions, etc. to maximize the long-term cumulative task value or network life ∑ iErem,i (Tend).

[0090] 5.2 Distributed computing resource collaborative allocation module: The computing tasks in the cluster are modeled as a directed acyclic graph (DAG). A distributed task scheduling algorithm (such as one based on game theory or heuristic search) is used to allocate subtasks to nodes with sufficient computing power. aVail,j And satisfy the delay constraint D req,k The optimization goal is to minimize the total task completion time or maximize the parallelism.

[0091] 5.3 Cross-layer communication resource joint optimization module: Construct a cross-layer optimization problem to jointly optimize the power p ij , frequency band f ij , time slot t ij , routing path sd ,Polarization P,i,P r,j, STC Mode ij The goal is to maximize the total network throughput∑ s,d R sd Or meet the SINR requirement of a specific link sINR ij ≥y th , while satisfying constraints such as power, spectrum, and interference. Decomposition techniques (such as dual decomposition) or meta-heuristic algorithms are used to solve the problem.

[0092] 6Decision control center (logic unit)

[0093] 6.1 Situational Awareness and Fusion: Integrate real-time information uploaded by each subsystem (interference map M int (t), channel graph M CSI (t), topological state G(t), node state s cluster (t)), construct the global situation map S(t).

[0094] 6.2 Policy Generation Based on Deep Reinforcement Learning: Taking the system state S(t) as the input of the DRL agent. The action space of the agent Contains control parameters for each subsystem (such as target topology parameters, polarization configuration index, STC mode index, resource allocation ratio). The reward function R(t) is designed to comprehensively reflect the anti-interference effect, communication performance, energy efficiency and task completion: R(t) = α1.ΔBER(t) + α2.Throughput(t) - α3.Energy(t) + α4.Tasksuccess(t) where α k Is the weight coefficient. Actor-Critic algorithms (such as DDPG, PPO) are used to train the policy network π(A|S; θ π ) and the value network V(S;θ v ), so that it can output the action that maximizes the long-term cumulative reward in the current state.

[0095] 6.3 Execution Control and Feedback: Optimal Action Output by DRL This is broken down into specific control instructions for the topology, polarization, STC, and resource scheduling subsystems, and then issued for execution. The execution results are monitored simultaneously, generating feedback to update the status at the next moment, forming a closed-loop control system.

[0096] Figure 1 The overall architecture of the proposed system is shown, consisting of five core functional subsystems (environmental perception, topology reconstruction, polarization control, space-time coding, and resource scheduling) and a core decision-making control center. Arrows represent information and control flows, highlighting the closed-loop feedback and collaborative working relationship between each subsystem and the decision-making center. The decision-making control center makes intelligent decisions based on deep reinforcement learning (DRL).

[0097] Figure 2 The system's workflow is detailed, exemplifying the characteristics of closed-loop adaptive control. The environmental perception subsystem continuously provides input, and the decision-making control center makes decisions based on the DRL model, issuing joint control instructions to the execution subsystems. Each subsystem then responds with feedback, and the decision-making center evaluates the results based on a reward function and updates the DRL model, continuously optimizing the system's strategy to adapt to the dynamically changing environment.

[0098] Figure 3 The core logic of the dynamic topology reconstruction subsystem in response to interference changes is demonstrated. When a significant interference change is detected and the current topology is insufficiently robust, the system initiates a topology optimization algorithm (such as RL-GA) to generate a new target topology and smoothly executes the reconstruction process to avoid interference and improve link quality.

[0099] Figure 4 Comparison of the layout and link status of the drone cluster before (left) and after (right) topology reconstruction. Red asterisks represent the location of interference sources, and red translucent spheres indicate their effective interference range. In the initial topology, some drones were close to the interference sources, causing interference to the communication links between them (red dashed lines). After dynamic topology reconstruction, the drones actively adjusted their positions to avoid the interference area, significantly reducing the number of interfered links in the reconstructed topology. Most links (green solid lines) resumed normal communication, thereby improving the robustness and reliability of the entire cluster network.

[0100] Figure 5 Description: The left figure uses the Poincare sphere to illustrate basic polarization states (H: horizontal, V: vertical, + / -45°: diagonal linear, RHC / LHC: left-handed circular polarization). The middle figure illustrates the effects of polarization matching and mismatching: when the receiving antenna polarization (green dashed line) matches the polarization of the desired signal (blue solid line), the received signal power is maximized while minimizing the power of the orthogonal interference signal (red dashed line). The right figure shows the adaptive receive polarization strategies that the system can adopt to address different types of interference, such as canceling co-polarization interference, using polarization matching to receive signals for cross-polarization interference, and employing diversity reception for non-polarization interference, to achieve optimal interference mitigation.

[0101] Figure 6 Description: The upper left figure shows the block diagram of the joint polarization-space-time coding transmitter in this invention, highlighting the integration of coding and polarization control, as well as the control signal from the decision center. The upper right figure shows an example codeword structure for a two-antenna, two-polarization, two-time-slot joint polarization-space-time block code (Pol-STBC). This codeword structure utilizes both spatial and polarization domains to transmit four data symbols within two time slots, improving spectrum efficiency and diversity potential. The figure below compares the performance of different schemes using simulated BER curves. As can be seen, compared to schemes with no diversity (SISO), spatial diversity only (STBC), or polarization diversity only (PolDiv), the proposed joint polarization-space-time coding scheme (Proposed Joint Pol-STBC) utilizes a higher diversity order, achieving significant BER improvement at the same signal-to-noise ratio (SNR). Even under strong interference conditions (jammed curve, with reduced effective signal-to-interference-and-noise ratio (SINR)), this joint scheme maintains relatively good performance, demonstrating its superior interference mitigation capabilities.

[0102] Figure 7 Four sub-graphs show the performance comparison of different indicators between the joint optimization scheme proposed in the present invention and several other baseline schemes that only adopt partial optimization strategies (or no optimization). Upper left graph: Under different interference intensities, the scheme of the present invention (ProposedJointOpt) always maintains the highest system throughput, especially under high interference. Upper right graph: Under medium interference conditions, the energy efficiency (the number of bits transmitted per unit energy) of the scheme of the present invention increases fastest with the signal-to-noise ratio (SNR), and is optimal at each SNR point. Lower left graph: The scheme of the present invention can maintain the highest link reliability (close to 100%) under various interference conditions, which is significantly better than other schemes. Lower right graph: The radar chart conducts a normalized comprehensive evaluation of each scheme from the three dimensions of throughput, energy efficiency, and reliability, clearly showing that the scheme of the present invention exhibits the best balance and overall performance in all optimization objectives.

[0103] The key to this invention's innovation lies in its ability to achieve, for the first time, a tightly coupled collaborative optimization of five previously independent anti-interference dimensions: environmental perception, topology, polarization state, space-time coding, and resource allocation. This results in a closed-loop adaptive control system. This cross-layer, cross-domain joint decision-making mechanism addresses the suboptimal nature of existing technologies, often caused by the decoupling and insufficient coordination of modules. This results in a subtle improvement in overall performance.

[0104] Example 2

[0105] Simplified topology management solutions based on tiering or clustering

[0106] Description: For large-scale drone swarms, real-time optimization of the global topology can be prohibitively complex. This solution divides the swarm into several subclusters, each employing a relatively stable local topology (such as a mesh or star topology). Inter-cluster communication occurs through elected cluster heads (CHs). Topology management focuses on dynamic cluster division, cluster head election, and inter-cluster routing maintenance.

[0107] Implementation:

[0108] Clustering algorithm: An improved algorithm based on node energy, location, and connectivity can be used, such as Energy-Aware Hierarchical Clustering (EAHC) or Load-Balanced Clustering Algorithm (LBCA).

[0109] Cluster head election: The cluster head is elected periodically or on an event-triggered basis, taking into account factors such as the node's remaining energy, computing power, average distance to cluster members, and link quality.

[0110] Interference perception and cluster adjustment: When a node or cluster head in a cluster senses strong interference, it can trigger fine-tuning of the cluster topology (such as disconnecting the interfered link) or re-clustering and cluster head election to avoid the interference area.

[0111] Advantages and Disadvantages: This approach reduces the computational complexity of global topology optimization and improves scalability. However, it may sacrifice some global optimization performance. Cluster head nodes can easily become bottlenecks or single points of failure, and the anti-interference ability of inter-cluster communication may be weaker than that of global optimization solutions.

[0112] Example 3

[0113] Adaptive frequency hopping and power control scheme based on interference perception

[0114] Description: As a simplified alternative or supplement to the polarization and space-time coding techniques in the present invention, this solution focuses more on adaptive adjustment in the frequency and power domains to combat interference.

[0115] Implementation:

[0116] Wideband spectrum sensing: Uses methods such as energy detection and cyclic spectrum analysis to quickly scan the operating frequency band and identify interfered sub-bands and available "spectrum holes."

[0117] Intelligent Adaptive Frequency Hopping (IAFH): Dynamically selects the frequency hopping sequence based on real-time spectrum sensing results, giving priority to communicating at frequencies with low interference power. The frequency hopping pattern is difficult to predict and track.

[0118] Interference-Aware Power Control (JAPC): Based on the selected frequency, the transmit power is finely adjusted according to the interference level and channel conditions at that frequency to ensure that the receiving end achieves the target SINR while avoiding excessive interference to neighboring nodes and saving energy.

[0119] Combined with topology: The quality of the available hopping frequency set or the required transmit power of a link can be used as one of the indicators for topology evaluation and route selection.

[0120] Advantages and Disadvantages: Relatively simple to implement, with low antenna hardware requirements (no reconfigurable polarization antennas are required). Effective against narrowband or partial-band interference. However, it is less effective than the combined approach of the present invention against wideband interference, co-channel interference, or interference that is separable in the spatial or polarization dimensions. Energy efficiency may be lower than this approach because higher power may be required to overcome residual interference.

[0121] The two alternative solutions of Example 2 and Example 3 can be used as simplified versions of the complete solution of the present invention or alternative solutions in specific scenarios based on specific application requirements, cost constraints, hardware platform capabilities, and the expected complexity of the interference environment. They can also be combined with certain modules in the present invention (for example, the topology management of Alternative Solution 1 can be combined with the polarization / STC module of the present invention) to form a hybrid anti-interference system.

[0122] The basic principles, main features and advantages of the present invention are shown and described above. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention.

[0123] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A polarization-space-time coding anti-interference communication system for UAV swarms based on dynamic topology reconstruction, characterized by: It includes an environmental perception subsystem, a topology reconstruction subsystem, an adaptive polarization control subsystem, a reconfigurable space-time coding subsystem, a multi-dimensional resource joint scheduling subsystem, and a decision-making control center. The environmental perception subsystem is responsible for comprehensive and real-time perception of the communication environment in which the UAV cluster is located; the topology reconstruction subsystem dynamically optimizes the communication topology of the cluster based on the environmental perception results and mission requirements; the adaptive polarization control subsystem dynamically adjusts the polarization state of the receiving and transmitting signals according to the polarization characteristics of the channel and interference; the reconfigurable space-time coding subsystem adaptively selects and configures the space-time coding scheme according to the channel quality, interference characteristics, and business requirements; The multi-dimensional resource joint scheduling subsystem coordinates and allocates energy, computing and communication resources within the cluster in a unified and efficient manner while meeting task constraints.

2. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction according to claim 1 is characterized by: The environmental perception subsystem includes a multi-mode interference perception and positioning module, a high-precision CSI estimation module and a cluster self-status monitoring module; the topology reconstruction subsystem includes a topology quality assessment module, an intelligent topology generation module, and a smooth topology execution and control module; the adaptive polarization regulation subsystem includes a polarization state real-time detection module, a polarization strategy optimization module, and a reconfigurable polarization antenna control module; the reconfigurable space-time coding subsystem includes a coding mode and parameter selection module, a joint polarization-space-time codeword generation module, and a coding and signal processing execution module; the multi-dimensional resource joint scheduling subsystem includes an energy efficiency perception and optimization management module, a distributed computing resource collaborative allocation module, and a cross-layer communication resource joint optimization module; the decision control center includes situational perception and fusion, strategy generation based on deep reinforcement learning, execution control and feedback.

3. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction according to claim 2 is characterized by: The multi-mode interference perception and positioning module adopts a hybrid model based on deep learning, in which a convolutional neural network (CNN) is first used to extract static spatial features from the spectrum or constellation diagram of the received signal; then a recurrent neural network (RNN) or Transformer model is combined to process the time series of the static features to identify dynamic interference patterns such as sweeps or pulses; finally, a classifier (including an SVM or fully connected layer) outputs the probability of the interference type.

4. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction according to claim 3 is characterized by: The decision control center is implemented in a distributed or centralized manner, integrating information from each subsystem, running the core multi-objective optimization decision algorithm based on the global situation and preset task objectives, generating collaborative control instructions and sending them to each execution subsystem to achieve closed-loop intelligent regulation of the entire system. The multi-objective optimization decision algorithm includes a policy network based on deep reinforcement learning (DRL).

5. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction according to claim 2 is characterized by: It also includes building a unified, closed-loop adaptive control framework across the physical layer and network layer to achieve collaborative, dynamic, and intelligent optimization of the three key degrees of freedom in drone cluster communications: topology structure, signal polarization state, and space-time coding mode, in order to cope with complex electromagnetic interference.

6. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction according to claim 2 is characterized by: The energy topology generation module adopts a genetic algorithm combined with reinforcement learning (RL-GA), wherein the reinforcement learning agent dynamically adjusts the crossover operator, mutation operator and fitness function weight of the genetic algorithm according to the environmental perception results.

7. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction according to claim 2 is characterized by: The coding mode and parameter selection module is used to maintain a performance-complexity database, storing the theoretical / simulated performance and computational complexity of different STC modes under different channel conditions and interference types, where the channel conditions include high / low SNR and correlated / non-correlated channels, and the theoretical / simulated performance includes BER and throughput.

8. The UAV swarm polarization-space-time coding anti-interference communication system based on dynamic topology reconstruction according to claim 2 is characterized by: The smooth topology execution and control module adopts the artificial potential field method or the model predictive control (MPC) algorithm to plan a smooth flight trajectory for each UAV to avoid collision, so as to achieve the transition from the current topology to the target topology.

Images