Signal safety and APT protection system based on multi-modal fusion and power regulation and control

Abstract

The invention discloses a signal security and APT protection system based on multi-mode fusion and power regulation and control, and belongs to the technical field of wireless communication and network security fusion. The system aims at solving the problems that a traditional conformal antenna is rigid in power regulation and control and poor in environmental adaptability, APT attacks extend to a radio frequency layer, and existing protection lacks multi-layer deep linkage, and comprises six core modules, and multi-modal features are extracted through radio frequency sensing, network behavior monitoring and environment sensing modules; a graph neural network and Transformer / reinforcement learning fused AI decision engine carries out unified modeling, outputs a power distribution matrix and a safety response strategy, drives a conformal array antenna to dynamically adjust the phase and the power, executes protection operations such as dynamic beam interference and adaptive power adjustment, adapts to multiple scenes such as vehicle-mounted scenes, unmanned aerial vehicles and smart homes, and improves the safety and reliability of the conformal array antenna. And the safety, the environmental adaptability and the power utilization efficiency of wireless communication are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of wireless communication and network security convergence technology, and in particular to a signal security and APT protection system based on multimodal fusion and power regulation. Background Technology

[0002] With the rapid development of multi-source wireless communication, such as 5G, V2X, and drone relay networks, the signal space is becoming increasingly complex. While traditional conformal antennas offer advantages in concealment and lightweight design, their power regulation capabilities are limited, failing to achieve environmental adaptation and security layer integration. On the other hand, Advanced Persistent Threat (APT) attacks have gradually extended from the network protocol layer to the radio frequency (RF) physical layer, utilizing low power consumption, short bursts, or carrier hiding to achieve covert communication. Existing protection methods mainly focus on network-side traffic analysis and host protection, lacking a real-time detection mechanism deeply integrated with the RF layer. Furthermore, changes in environmental temperature, humidity, dielectric stress, and curvature can all affect the performance of conformal antennas, leading to signal leakage, power imbalance, or directional distortion, thereby impacting system communication security. Therefore, a comprehensive system integrating AI power regulation, environmental adaptive compensation, and multimodal APT identification is urgently needed. Summary of the Invention

[0003] The purpose of this invention is to provide a signal security and APT protection system based on multi-mode fusion and power regulation, which solves the problems of limited power regulation capability and poor environmental adaptability of traditional conformal antennas, and the lack of deep linkage between the radio frequency layer, network layer and environmental layer in existing protection against APT attacks extending to the radio frequency physical layer.

[0004] To achieve the above objectives, this invention provides a signal security and APT protection system based on multimodal fusion and power regulation, including a conformal array antenna module, a radio frequency sensing and feature extraction module, a network behavior monitoring module, an environment sensing module, an AI power allocation and fusion decision engine, and a signal security and response module. The system achieves deep fusion protection of the signal layer, network layer, and environment layer through multimodal feature fusion and AI-driven dynamic regulation, specifically: The radio frequency sensing and feature extraction module, network behavior monitoring module, and environment sensing module respectively extract radio frequency fingerprint features, network behavior features, and environmental state features and transmit them to the AI ​​power allocation and fusion decision engine. The AI ​​power allocation and fusion decision engine performs unified modeling of multimodal features and outputs a power allocation matrix and security response strategy. The signal security and response module performs protection operations according to the power allocation matrix and security response strategy, and the conformal array antenna module performs dynamic phase and power adjustments according to the power allocation matrix. The control chain of the system is verified through the TPM / TEE module to ensure command reliability and tamper resistance.

[0005] Preferably, the conformal array antenna module is a multi-element flexible bonding array, made of flexible materials and processes, and bonded to the surface of a complex curved carrier. Each array element integrates an independent digitally controlled phase shifter and a programmable power amplifier at its rear end, enabling independent, high-precision, microsecond-level control of the phase and amplitude of the transmitted / received signals of each array element. The conformal array antenna module also integrates a tunable filter structure, real-time impedance detection, and an adjustable matching network, possessing array notch filtering and impedance dynamic matching functions. The array notch filtering function is used for active absorption and reflection of specific threat frequency bands, and the impedance dynamic matching function is used to adjust the matching state according to carrier deformation, environmental changes, and operating frequency.

[0006] Preferably, the radio frequency sensing and feature extraction module uses Fast Fourier Transform (FFT), Short-Time Fourier Transform (STFT), and Wavelet Transform for time-frequency domain preprocessing to extract multi-dimensional radio frequency fingerprint features. The multi-dimensional radio frequency fingerprint features include amplitude spectrum features, phase spectrum features, cyclic spectrum features, transient features, and sparse representation coefficients. The transient features include rising / falling edges and overshoot features of the signal during power-on / off and modulation switching. The sparse representation coefficients are obtained by sparse decomposition of the signal on an overcomplete dictionary.

[0007] Preferably, the network behavior monitoring module collects network traffic and protocol features based on the NetFlow / IPFIX standard and extracts multi-dimensional network behavior features, including session topology features, packet interval distribution and timing features, and protocol deep parsing features.

[0008] Preferably, the environmental perception module collects multi-dimensional environmental parameters, including temperature, humidity, stress, and electromagnetic noise; the module transmits the collected environmental parameters to the AI ​​power allocation and fusion decision engine in real time to realize antenna performance compensation, beam pointing correction, and contextual correlation analysis of abnormal communication.

[0009] Preferably, the AI ​​power allocation and fusion decision engine adopts a hybrid architecture of graph neural network and Transformer; the graph neural network is used to model the network session topology and spatial relationships between antenna elements, and to mine complex correlations in the topology; the Transformer encoder is used to process radio frequency feature sequences, environmental parameter time series data and graph features extracted by the graph neural network, and to capture global correlations between long-distance dependencies and multimodal features; the input of the engine is radio frequency feature vector, network behavior feature map and environmental state vector, and the output is an N×M complex power allocation matrix and a security response strategy decision vector, where N is the number of array elements and M is the channel.

[0010] Preferably, the protection operations of the signal security and response module include: Dynamic beam jamming: Adjust the array beam shape according to the power allocation matrix to form nulls aligned with the jamming direction and emit suppressive jamming signals towards the attack source; Adaptive power adjustment: Optimizes the total transmit power and power allocation of each link according to channel quality and communication requirements to achieve low probability of intercept communication; Link isolation: Isolating malicious terminals from external connections through either logical or physical means; Intelligent spectrum suppression: The command antenna array forms a transmit beam null point and adjusts impedance in the threat frequency band to achieve energy absorption; RF snapshot recording and tracing: Records raw I / Q data, spectrum diagrams and environmental context before and after the threat trigger moment, and generates a tracing report.

[0011] Preferably, the AI ​​power allocation and fusion decision engine adopts a joint architecture of multi-task reinforcement learning and graph neural network, and dynamically adjusts the transmit power and phase parameters of each array element in different channels according to the electromagnetic environment complexity, interference signal strength, signal direction of arrival, communication density and security level.

[0012] Preferably, the system is a hardware and software integrated reconfigurable platform. The conformal array antenna module supports reconfigurable array units and adopts a PIN diode and MEMS adjustable impedance structure. The system also integrates a local high-speed ADC sampling module and an FPGA front-end processing module to realize closed-loop control of signal perception, feature extraction, AI decision-making and security protection.

[0013] Therefore, this invention adopts the aforementioned signal security and APT protection system based on multimodal fusion and power regulation. Through multimodal feature fusion and AI-driven dynamic regulation architecture, it achieves deep collaborative protection of the signal layer, network layer, and environment layer. Its conformal array antenna module, with its flexible bonding design, microsecond-level high-precision regulation, and impedance dynamic matching function, effectively adapts to complex carriers and environmental changes, solving the pain points of rigid power regulation and susceptibility to environmental influences in traditional conformal antennas. Multi-module collaborative extraction of multi-dimensional features of RF fingerprints, network behavior, and environmental status, combined with an AI decision engine based on a hybrid architecture of graph neural networks and Transformers, enables accurate identification and real-time response to APT attacks. Through multi-faceted protection operations such as dynamic beam interference and adaptive power adjustment, coupled with TPM / TEE command verification and integrated hardware and software reconfigurable design, it ensures both low interception probability and anti-interference capability, while also achieving threat tracing and anti-tampering. It is suitable for multiple scenarios such as vehicles, drones, and smart homes, improving the security, environmental adaptability, and power utilization efficiency of wireless communication systems.

[0014] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0015] Figure 1 This is the system architecture diagram of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0018] Example 1 This invention provides a signal security and APT protection system based on multimodal fusion and power regulation, with the following architecture: Figure 1 As shown, the system includes a conformal array antenna module, a radio frequency sensing and feature extraction module, a network behavior monitoring module, an environmental perception module, an AI power allocation and fusion decision engine, and a signal security and response module. The system achieves deep fusion protection across the signal layer, network layer, and environment layer through multimodal feature fusion and AI-driven dynamic control. Specifically, the radio frequency sensing and feature extraction module, the network behavior monitoring module, and the environmental perception module extract radio frequency fingerprint features, network behavior features, and environmental state features, respectively, and transmit them to the AI ​​power allocation and fusion decision engine. The AI ​​power allocation and fusion decision engine performs unified modeling of the multimodal features and outputs a power allocation matrix and a security response strategy. The signal security and response module executes protection operations according to the power allocation matrix and the security response strategy, and the conformal array antenna module dynamically adjusts the phase and power according to the power allocation matrix. The system's control chain is verified through a TPM / TEE module to ensure command reliability and tamper resistance. The system is a reconfigurable hardware and software integrated platform. The conformal array antenna module supports reconfigurable array elements and adopts a PIN diode and MEMS adjustable impedance structure. The system also integrates a local high-speed ADC sampling module and an FPGA front-end processing module to realize closed-loop control of signal perception, feature extraction, AI decision-making and security protection.

[0019] The conformal array antenna module is a multi-element flexible bonding array, made of flexible materials and processes, and bonded to the surface of a complex curved carrier. Each array element integrates an independent digitally controlled phase shifter and a programmable power amplifier at its rear end, enabling independent, high-precision, microsecond-level control of the phase and amplitude of the transmitted / received signals of each array element. The conformal array antenna module also integrates a tunable filter structure, real-time impedance detection, and an adjustable matching network, and has array notch filtering and impedance dynamic matching functions. The array notch filtering function is used to actively absorb and reflect specific threat frequency bands, while the impedance dynamic matching function is used to adjust the matching state according to carrier deformation, environmental changes, and operating frequency.

[0020] The radio frequency sensing and feature extraction module uses Fast Fourier Transform (FFT), Short Time Fourier Transform (STFT), and Wavelet Transform for time-frequency domain preprocessing to extract multi-dimensional radio frequency fingerprint features. These multi-dimensional radio frequency fingerprint features include amplitude spectrum features, phase spectrum features, cyclic spectrum features, transient features, and sparse representation coefficients. Transient features include rising / falling edges and overshoot features during power-on / off and modulation switching. The sparse representation coefficients are obtained by sparsely decomposing the signal on an overcomplete dictionary.

[0021] The network behavior monitoring module collects network traffic and protocol characteristics based on the NetFlow / IPFIX standard, and extracts multi-dimensional network behavior features, including session topology features, packet interval distribution and timing features, and protocol deep parsing features.

[0022] The environmental perception module collects multi-dimensional environmental parameters, including temperature, humidity, stress, and electromagnetic noise. The module transmits the collected environmental parameters to the AI ​​power allocation and fusion decision engine in real time to realize antenna performance compensation, beam pointing correction, and contextual correlation analysis of abnormal communication.

[0023] The AI ​​power allocation and fusion decision engine adopts a hybrid architecture of graph neural networks and Transformers. Graph neural networks are used to model network session topology and spatial relationships between antenna elements, uncovering complex correlations in the topology. Transformer encoders are used to process RF feature sequences, environmental parameter time-series data, and graph features extracted by graph neural networks, capturing global correlations between long-distance dependencies and multimodal features. The engine's inputs are RF feature vectors, network behavior feature maps, and environmental state vectors, and its outputs are an N×M complex power allocation matrix and a security response strategy decision vector, where N is the number of array elements and M is the number of channels. The AI ​​power allocation and fusion decision engine adopts a joint architecture of multi-task reinforcement learning and graph neural networks, dynamically adjusting the transmit power and phase parameters of each array element in different channels based on electromagnetic environment complexity, interference signal strength, signal direction of arrival, communication density, and security level.

[0024] The protection operations of the signal security and response module include: Dynamic beam jamming: Adjust the array beam shape according to the power allocation matrix to form nulls aligned with the jamming direction and emit suppressive jamming signals towards the attack source; Adaptive power adjustment: Optimizes the total transmit power and power allocation of each link according to channel quality and communication requirements to achieve low probability of intercept communication; Link isolation: Isolating malicious terminals from external connections through either logical or physical means; Intelligent spectrum suppression: The command antenna array forms a transmit beam null point and adjusts impedance in the threat frequency band to achieve energy absorption; RF snapshot recording and tracing: Records raw I / Q data, spectrum diagrams and environmental context before and after the threat trigger moment, and generates a tracing report.

[0025] The following section explains how this system works in specific application scenarios.

[0026] Example 2 Vehicle-mounted multi-signal conformal antenna system: The vehicle-mounted multi-signal conformal antenna system is a core technology solution for V2X communication security in intelligent connected vehicles. With the development of 5G / 6G communication technology and the advancement of autonomous driving, V2X communication scenarios such as vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-pedestrian (V2P) have placed higher demands on communication security and power efficiency. Existing vehicle-mounted antenna systems suffer from problems such as fixed power distribution, poor environmental adaptability, and insufficient safety protection capabilities, making it difficult to cope with complex and ever-changing traffic and electromagnetic environments.

[0027] This embodiment employs an AI power allocation engine based on a graph neural network and Transformer fusion architecture to achieve real-time perception and dynamic response to traffic environment and electromagnetic interference. The system integrates data from vehicle-mounted radar, cameras, GPS, and other devices through multimodal sensor fusion technology, combined with radio frequency sensing and feature extraction modules, enabling it to identify security threats in different communication scenarios and adaptively adjust antenna array power allocation strategies.

[0028] The working principle of the vehicle-mounted multi-signal conformal antenna system is based on a collaborative mechanism of multi-modal fusion perception and AI intelligent decision-making. The system collects physical parameters such as temperature, humidity, and vibration in real time, as well as dynamic information such as vehicle speed, position, and driving direction, through the vehicle's environmental perception module. This environmental data is fed back to the AI ​​power allocation and fusion decision engine for antenna array fitting compensation and power allocation strategy optimization.

[0029] The radio frequency (RF) sensing and feature extraction module is responsible for preprocessing and extracting features from the received signal. The system employs various signal processing techniques such as FFT / STFT / wavelet transform to analyze signal characteristics from multiple dimensions, including the time domain, frequency domain, and time-frequency domain. Specifically, by extracting RF fingerprint features, the system can identify the hardware characteristics of different communication devices, enabling device authentication and secure access control. RF fingerprints encompass multiple dimensions, including carrier frequency offset, nonlinear distortion, and frequency response.

[0030] The network behavior monitoring module collects network traffic data based on the NetFlow / IPFIX protocol, extracting information such as session topology, protocol characteristics, and traffic patterns. Through deep session tracing technology, the system establishes a normal operation model and compares it with actual network activity. Once an anomaly is detected, it immediately triggers an alarm or automatically takes security measures.

[0031] The AI ​​power allocation and fusion decision engine employs a unified modeling architecture that combines graph neural networks and Transformers. Graph neural networks are used to model network topology and node relationships, while Transformers handle temporal features and long-range dependencies. Through multimodal data fusion, the engine can comprehensively analyze environmental perception data, radio frequency characteristics, network behavior, and other information to output the optimal power allocation matrix and security response strategy.

[0032] The signal security and response module performs dynamic beamforming and interference suppression. When a security threat is detected, the system adjusts the phase and amplitude distribution of the antenna array to form a directional beam for interference suppression or signal enhancement. Simultaneously, the system records radio frequency snapshot data, including the signal's time-domain waveform, spectral characteristics, and spatial distribution, for subsequent security incident tracing and forensic analysis.

[0033] Its power allocation mechanism is as follows: The system adjusts the transmission power of each antenna unit in real time according to traffic conditions and electromagnetic interference to achieve the dual goals of V2X communication security and energy efficiency optimization.

[0034] In normal communication scenarios, the system employs a priority-based power allocation strategy. Based on the security level of V2X communication, communications are categorized into three levels: emergency safety messages (such as forward accident warnings), important business messages (such as traffic signal information), and general data messages. Emergency safety messages have the highest priority, and the system allocates the maximum available power to them to ensure reliable message transmission. Important business messages have the next highest priority, while general data messages have the lowest priority.

[0035] When electromagnetic interference is detected, the system employs a game theory-based power control algorithm. The algorithm models the interference environment as a multi-agent game scenario, where each communication node acts as an agent, maximizing its communication utility function by optimizing its own power allocation. The utility function comprehensively considers factors such as signal strength, interference level, and power consumption, and an iterative optimization algorithm solves for the Nash equilibrium solution to achieve the globally optimal power allocation.

[0036] When encountering an APT attack, the system activates a security response mode. First, it uses radio frequency fingerprinting technology to identify the attacker, then employs dynamic beam jamming technology to suppress the attack signal. Simultaneously, the system automatically reduces the power output of non-critical communications, concentrating limited power resources on ensuring secure V2X communications.

[0037] The system possesses learning and adaptive capabilities. By continuously collecting and analyzing communication data, the AI ​​engine constantly optimizes the power allocation strategy, improving the system's adaptability and robustness in different environments. Especially in complex urban environments, the system can dynamically adjust the antenna beam pointing and power distribution based on factors such as building distribution, road conditions, and traffic flow, ensuring communication quality and security.

[0038] Example 3 Unmanned Aerial Vehicles (UAVs) / Shipborne Platforms: UAVs and shipborne platforms face severe communication security challenges in complex electromagnetic environments. UAV platforms need to operate under various weather conditions while simultaneously dealing with electronic interference and interception threats from adversaries. Shipborne platforms, on the other hand, face the complex electromagnetic environment on board ships, the influence of marine climate, and potential electronic warfare threats. Traditional antenna systems are insufficient to meet the comprehensive requirements of these platforms for low probability of intercept, high anti-jamming capabilities, and power domain protection.

[0039] This embodiment designs a conformal antenna system with low intercept communication capability and power domain protection for the specific application scenarios of UAVs / shipborne platforms. The system adopts a multi-element flexible bonding array design, which can adapt to the complex curved surfaces of the platform, and has phase control and power amplification functions, supporting array notch filtering and dynamic impedance matching.

[0040] Low probability of intercept (LPI) communication employs multiple techniques to reduce the likelihood of signal interception and identification by the enemy, ensuring the stealth and security of communication. First, frequency hopping technology: through random frequency hopping sequences and precise time synchronization, the system can transmit signals on different frequencies in a very short time, making it difficult for the enemy to track and intercept. The frequency hopping pattern is generated using pseudo-random sequences, possessing good randomness and unpredictability. Second, power control technology: based on communication distance and channel conditions, the transmit power is dynamically adjusted to ensure the receiver can correctly demodulate the signal while minimizing the signal's spatial propagation range. The power control algorithm, based on channel estimation and distance measurement, can accurately calculate the minimum required transmit power. Simultaneously, the system employs power density control technology to distribute the signal power over a wider frequency range, reducing the power density per unit bandwidth. Third, signal design technology: using spread spectrum modulation, the original signal is extended into a wider frequency band, reducing the signal's spectral density. The spreading factor can be dynamically adjusted according to security requirements, maximizing spreading gain while ensuring communication quality. Meanwhile, the system employs low-probability detection signal waveform designs, such as noise-like signals and chaotic sequences, to further reduce signal detectability. Fourthly, regarding antenna technology, a conformal array design is used, distributing antenna elements across the platform surface to reduce concentrated scattering sources and lower the radar cross section (RCS). Through intelligent beamforming technology, the system can concentrate energy in the target direction while simultaneously creating low-gain or zero-gain regions in other directions, reducing spatial signal leakage.

[0041] The electromagnetic environment faced by UAVs / shipborne platforms is complex and ever-changing, and anti-interference technologies mainly include various methods in the frequency domain, time domain, and spatial domain.

[0042] In the frequency domain, the system employs adaptive frequency hopping and spectrum sensing technologies. By monitoring spectrum usage in real time, the system can avoid interfering frequency bands and select clean frequency resources for communication. When the spectrum sensing module detects interference signals, the system automatically adjusts its frequency hopping pattern to avoid interfering frequency bands.

[0043] In the time domain, the system employs burst communication and pulse compression techniques. By compressing information and transmitting it within an extremely short time, the system can reduce the signal's exposure time in space, lowering the probability of interception. Pulse compression, through matched filtering, enables the recovery of the original signal at the receiving end while maintaining low peak power.

[0044] In the spatial domain, the system employs intelligent beamforming and adaptive null formation techniques. By adjusting the phase and amplitude of each element in the antenna array, the system can form a high-gain beam pointing towards the target while simultaneously creating nulls in the interference direction, effectively suppressing interference signals. The null formation algorithm is based on the minimum variance distortionless response (MVDR) criterion, enabling it to maintain the integrity of the target signal while suppressing interference.

[0045] Power domain protection is another important feature of the system. The system can monitor the power intensity of the received signal in real time and automatically activate protection mechanisms when an abnormally high power signal is detected. Protection measures include reducing the receiver gain, shutting down some receiver channels, and activating the limiter. Simultaneously, the system possesses power inversion capabilities, enabling it to infer the location and power of the transmitting source based on the power characteristics of the received signal, providing a basis for subsequent countermeasures.

[0046] The system is designed for electromagnetic compatibility, enabling stable operation in complex electromagnetic environments. Through shielding, filtering, and grounding techniques, the system effectively suppresses the effects of both internal and external interference. The shielding design employs a multi-layered shielding structure. The filtering technology utilizes multi-stage LC filters to effectively suppress interference in specific frequency bands.

[0047] Example 4 Smart Home Terminal Array: Smart home terminal devices are numerous, including smart speakers, smart cameras, smart sensors, and smart appliances. These devices interconnect through wireless communication networks. However, traditional smart home antenna systems suffer from problems such as fixed power distribution, severe interference, and insufficient privacy protection capabilities, making it difficult to meet users' dual needs for communication quality and privacy security.

[0048] This embodiment addresses the unique characteristics of smart home environments by designing an adaptive power allocation smart home terminal array antenna system. The system employs conformal array antenna technology, seamlessly integrating with the home environment without compromising aesthetics. Through an AI power allocation algorithm, the system automatically adjusts the antenna array's power allocation strategy based on environmental changes and user needs, achieving the goals of minimizing interference and maximizing privacy protection.

[0049] The adaptive power allocation strategy of the smart home terminal array is based on a deep understanding of the home environment and user behavior. The system collects environmental parameters such as temperature, humidity, light, sound, and motion in real time through the environmental sensing module, and combines them with network behavior monitoring data to build environmental models and user behavior models.

[0050] The system employs a power allocation algorithm based on reinforcement learning. The agent learns to take the optimal power allocation action under different states by interacting with the environment, maximizing long-term cumulative rewards. The reward function comprehensively considers multiple factors such as communication quality, interference level, power consumption, and privacy protection, achieving multi-objective optimization.

[0051] In terms of interference suppression, the system employs spatial multiplexing technology and intelligent beamforming to reduce interference between devices. The system can identify the location and communication needs of different devices, allocating an independent beam to each device to avoid signal overlap. Simultaneously, the system uses power control technology to dynamically adjust the transmission power based on device distance and channel conditions, minimizing interference while ensuring communication quality.

[0052] In terms of privacy protection, the system employs multiple technical measures. First, data encryption: all communication data is transmitted using encrypted algorithms. Second, device authentication: the system uses radio frequency fingerprint recognition technology to ensure that only authorized devices can access the network. Third, privacy-aware power control: the system dynamically adjusts the device's power output based on user activity patterns and location information to reduce signal leakage.

[0053] The system possesses learning and memory capabilities. By continuously collecting and analyzing user behavior data, the system can learn users' daily activity patterns, predict user needs, and adjust power allocation strategies in advance. For example, when the system detects that a user is entering the bedroom to rest, it will automatically reduce the power of devices in the bedroom area while maintaining the minimum operating power of necessary equipment.

[0054] Therefore, this invention adopts the aforementioned signal security and APT protection system based on multimodal fusion and power regulation. Through multimodal feature fusion and AI-driven dynamic regulation architecture, it achieves deep collaborative protection of the signal layer, network layer, and environment layer. Its conformal array antenna module, with its flexible bonding design, microsecond-level high-precision regulation, and impedance dynamic matching function, effectively adapts to complex carriers and environmental changes, solving the pain points of rigid power regulation and susceptibility to environmental influences in traditional conformal antennas. Multi-module collaborative extraction of multi-dimensional features of RF fingerprints, network behavior, and environmental status, combined with an AI decision engine based on a hybrid architecture of graph neural networks and Transformers, enables accurate identification and real-time response to APT attacks. Through multi-faceted protection operations such as dynamic beam interference and adaptive power adjustment, coupled with TPM / TEE command verification and integrated hardware and software reconfigurable design, it ensures both low interception probability and anti-interference capability, while also achieving threat tracing and anti-tampering. It is suitable for multiple scenarios such as vehicles, drones, and smart homes, improving the security, environmental adaptability, and power utilization efficiency of wireless communication systems.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A signal security and APT protection system based on multimodal fusion and power regulation, characterized in that, The system includes a conformal array antenna module, a radio frequency sensing and feature extraction module, a network behavior monitoring module, an environmental sensing module, an AI power allocation and fusion decision engine, and a signal security and response module. Through multi-modal feature fusion and AI-driven dynamic control, the system achieves deep fusion protection across the signal layer, network layer, and environment layer, specifically: The radio frequency sensing and feature extraction module, network behavior monitoring module, and environment sensing module respectively extract radio frequency fingerprint features, network behavior features, and environmental state features and transmit them to the AI ​​power allocation and fusion decision engine; the AI ​​power allocation and fusion decision engine performs unified modeling of multimodal features and outputs a power allocation matrix and a security response strategy; The signal security and response module performs protection operations according to the power allocation matrix and security response strategy, and the conformal array antenna module performs dynamic phase and power adjustments according to the power allocation matrix; the control chain of the system is verified by the TPM / TEE module to ensure command reliability and tamper resistance.

2. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The conformal array antenna module is a multi-element flexible bonding array, made of flexible materials and processes, and bonded to the surface of a complex curved carrier. Each array element integrates an independent digitally controlled phase shifter and a programmable power amplifier at its rear end, enabling independent, high-precision, microsecond-level control of the phase and amplitude of the transmitted / received signals of each array element. The conformal array antenna module also integrates a tunable filter structure, real-time impedance detection, and an adjustable matching network, possessing array notch filtering and dynamic impedance matching functions. The array notch filtering function is used for active absorption and reflection of specific threat frequency bands, and the dynamic impedance matching function is used to adjust the matching state according to carrier deformation, environmental changes, and operating frequency.

3. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The radio frequency sensing and feature extraction module employs Fast Fourier Transform (FFT), Short-Time Fourier Transform (STFT), and Wavelet Transform for time-frequency domain preprocessing to extract multi-dimensional radio frequency fingerprint features. These multi-dimensional radio frequency fingerprint features include amplitude spectrum features, phase spectrum features, cyclic spectrum features, transient features, and sparse representation coefficients. The transient features include rising / falling edges and overshoot features of the signal during power-on / off and modulation switching. The sparse representation coefficients are obtained by sparsely decomposing the signal on an overcomplete dictionary.

4. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The network behavior monitoring module collects network traffic and protocol features based on the NetFlow / IPFIX standard and extracts multi-dimensional network behavior features, including session topology features, packet interval distribution and timing features, and protocol deep parsing features.

5. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The environmental perception module collects multi-dimensional environmental parameters, including temperature, humidity, stress, and electromagnetic noise. The module transmits the collected environmental parameters to the AI ​​power allocation and fusion decision engine in real time to realize antenna performance compensation, beam pointing correction, and contextual correlation analysis of abnormal communication.

6. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The AI ​​power allocation and fusion decision engine adopts a hybrid architecture of graph neural network and Transformer. The graph neural network is used to model the network session topology and spatial relationships between antenna elements, and to mine complex correlations in the topology. The Transformer encoder is used to process radio frequency feature sequences, environmental parameter time series data and graph features extracted by the graph neural network, and to capture global correlations between long-distance dependencies and multimodal features. The input of the engine is radio frequency feature vector, network behavior feature map and environmental state vector, and the output is an N×M complex power allocation matrix and a security response strategy decision vector, where N is the number of array elements and M is the number of channels.

7. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The protection operations of the signal security and response module include: Dynamic beam jamming: Adjust the array beam shape according to the power allocation matrix to form nulls aligned with the jamming direction and emit suppressive jamming signals towards the attack source; Adaptive power adjustment: Optimizes the total transmit power and power allocation of each link according to channel quality and communication requirements to achieve low probability of intercept communication; Link isolation: Isolating malicious terminals from external connections through either logical or physical means; Intelligent spectrum suppression: The command antenna array forms a transmit beam null point and adjusts impedance in the threat frequency band to achieve energy absorption; RF snapshot recording and tracing: Records raw I / Q data, spectrum diagrams and environmental context before and after the threat trigger moment, and generates a tracing report.

8. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The AI ​​power allocation and fusion decision engine adopts a joint architecture of multi-task reinforcement learning and graph neural network. Based on the complexity of the electromagnetic environment, the strength of the interference signal, the direction of the signal wave, the communication density, and the security level, it dynamically adjusts the transmission power and phase parameters of each array element in different channels.

9. The signal security and APT protection system based on multimodal fusion and power regulation according to claim 1, characterized in that, The system is a hardware and software integrated reconfigurable platform. The conformal array antenna module supports reconfigurable array elements and adopts a PIN diode and MEMS adjustable impedance structure. The system also integrates a local high-speed ADC sampling module and an FPGA front-end processing module to realize closed-loop control of signal perception, feature extraction, AI decision-making and security protection.

Images