A control system for a controllable anti-recording device

Abstract

The application discloses a controllable recording prevention device control system, and relates to the technical field of information security protection.A four-element cross array MEMS microphone array is used in the multi-dimensional audio detection module, which can identify recording equipment characteristic signals in a 20Hz to 20kHz frequency band.A self-adaptive comb interference spectrum is generated by the interference generation module based on DDS technology, and the communication and voice frequency bands are dynamically avoided.The heterogeneous computing main control module integrates CPU and DSP, and is provided with a multi-level decision tree algorithm.A composite energy management module combines piezoelectric and electromagnetic dual-mode energy collection, and an anti-interference communication module uses a UWB and Bluetooth 5.3 hybrid link, supports dynamic frequency hopping and LDPC coding.The application combines a multi-dimensional microphone array with an intelligent algorithm to realize high-precision recording equipment detection and positioning, and the system has high integration degree and can be quickly deployed in conference rooms, secure locations and the like to provide all-round protection for information security.

Description

Technical Field

[0001] The present invention relates to the technical field of information security protection, and in particular to a control system of a controllable anti-recording device. Background Art

[0002] With the rapid development of information technology, the prevalence of smart mobile devices and miniature recording devices has posed increasingly severe challenges to information security. In highly confidential settings such as government agencies, financial institutions, and research institutes, unauthorized recording can lead to the leakage of confidential information and significant losses. Traditional anti-recording technologies rely primarily on metal detectors or manual inspections, but these methods have significant limitations: they cannot detect recording functions built into electronic devices, have limited ability to identify miniature recording devices (such as button recorders and pinhole cameras), and require continuous manual intervention, resulting in low efficiency.

[0003] Detection technologies based on acoustic principles have emerged, identifying recording devices by analyzing unusual frequency components or signal characteristics in ambient sound. However, existing acoustic detection systems mostly use a single microphone, which cannot accurately determine the direction of the sound source. They are also easily affected by background noise in complex acoustic environments, resulting in a high false alarm rate. Furthermore, some anti-recording devices use full-band interference technology. While this effectively suppresses recording devices, it also severely impacts normal voice communications, making them incapable of meeting and lecture scenarios where voice clarity is crucial.

[0004] In recent years, intelligent anti-recording systems based on signal processing and machine learning have gradually developed, but technical bottlenecks remain. For example, existing systems are unable to adequately recognize new recording devices (such as those using low-noise ADCs or specialized sampling frequencies) and lack adaptive learning mechanisms to adapt to the evolving threat landscape. Suboptimal energy management schemes result in short battery life, making them incapable of meeting long-term monitoring requirements. Furthermore, regarding communication security, traditional encryption methods are vulnerable to attacks and cannot guarantee the security of remote control and data transmission. Therefore, developing a high-precision, intelligent, low-power anti-recording system with multiple security features is of great practical significance. Summary of the Invention

[0005] The present invention proposes a control system for a controllable anti-recording device to solve the problems mentioned in the above-mentioned prior art.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A control system for a controllable anti-recording device includes the following modules:

[0008] Multi-dimensional audio detection module: a four-element cross array layout is formed by a micro-electromechanical system microphone array, combined with a time-frequency domain joint analysis algorithm, the audio signal is first converted to the frequency domain through short-time Fourier transform, and then the wavelet packet decomposition is used to extract the detail features; connected with the heterogeneous computing host module through the high-speed data bus, real-time transmission of audio feature vectors and recording device orientation information;

[0009] Interference generation module: a wideband signal source based on direct digital frequency synthesis technology, real-time sensing of available frequency bands through dynamic spectrum access technology, introducing a spectrum utilization optimization formula , wherein is the spectrum utilization rate, is the bandwidth of the ith available frequency band, is the energy utilization efficiency of the frequency band, n is the total number of available frequency bands, is the adjustment coefficient, and are the interference signal and the target frequency band center frequency respectively, is the maximum frequency range of the system, is the total available bandwidth, is the dynamic spectrum access optimization factor, connected with the heterogeneous computing host module through the control interface, receiving the target frequency band, interference intensity control parameters;

[0010] Heterogeneous computing host module: using a heterogeneous processor to integrate CPU and special signal processing unit, built-in multi-feature fusion matching algorithm, storing recording device feature parameters through template library, configuring capacitive sensor array to support five working mode gesture switching, with edge parameter updating capability, receiving multi-dimensional audio detection module information, sending control parameters to interference generation module, and interacting with anti-interference communication module through encrypted communication interface;

[0011] Composite energy management module: integrating piezoelectric and electromagnetic dual-mode energy harvesting units, introducing an energy management efficiency formula , wherein is the energy management efficiency, and are the piezoelectric and electromagnetic unit collected power respectively, is the energy conversion efficiency, is the system loss power, is the total input energy, the module is matched with a lithium-sulfur battery with an energy density of 300 Wh / kg and a Qi protocol wireless charging coil, and the dual-mode energy collaborative management is realized through formula optimization, and each module is adaptively powered and energy consumption is monitored through the power management bus;

[0012] Anti-interference communication module: It adopts ultra-wideband UWB and Bluetooth 5.3 hybrid communication link, data transmission adopts 2000 hops / second dynamic frequency hopping technology and LDPC coding, and communication encryption adopts 128-bit symmetric encryption algorithm. It interacts with the heterogeneous computing main control module through an encrypted communication interface to realize data upload, command reception and multi-device collaboration.

[0013] Furthermore, it also includes a voiceprint separation interference module: this module is connected to the multi-dimensional audio detection module and the interference generation module through a high-speed data bus: the multi-dimensional audio detection module transmits the extracted audio feature vector ring to the voiceprint separation interference module in real time; the voiceprint separation interference module adopts the variable step size LMS algorithm in the adaptive filtering technology, dynamically adjusts the filtering parameters through voice activity detection, separates the environmental voice and the recording device signal in real time, and introduces the voiceprint separation degree formula ,in is the voiceprint separation, is the pure voice power, and are the interference signal and noise power respectively, λ is the phase influence factor, and They are the original speech and interference wave phases respectively, and voiceprint separation and interference control are achieved through formula optimization.

[0014] Furthermore, it also includes a physical layer encryption module: this module is connected to the heterogeneous computing main control module and the anti-interference communication module through an encrypted data interface: the heterogeneous computing main control module transmits the detected recording evidence data (including device type, timestamp, and characteristic parameters) to the physical layer encryption module; the physical layer encryption module is based on the Cai circuit design of the chaotic signal generator, and the generated pseudo-random sequence passes the NISTSP800-22 test, introducing the chaotic encryption entropy formula ,in is the encryption entropy value, n is the symbol set size of the chaotic sequence, is the probability distribution of sequence symbols, i is the state index, λ is the chaos complexity factor, and They are synchronization time and bit period respectively. The module performs real-time stream encryption on the recorded evidence and uses stream cipher mode and audio feature point coding to achieve physical watermark embedding.

[0015] Furthermore, the microphone array in the multi-dimensional audio detection module uses a cross-correlation algorithm to perform real-time self-calibration, uses a MUSIC algorithm to perform spatial spectrum estimation, and determines the position of the recording device by searching for spatial spectrum peaks.

[0016] Furthermore, the interference signal in the interference generation module is generated by combining direct digital frequency synthesis with digital up-conversion technology, generating a sine wave signal through a lookup table, and outputting it after digital-to-analog conversion and low-pass filtering to generate a frequency hopping interference signal.

[0017] Furthermore, the signal processing algorithm in the heterogeneous computing main control module adopts a multi-level feature extraction architecture, first extracting audio features through Mel-frequency cepstral coefficients, and then using decision trees for classification and recognition. The built-in feature template library supports online updates and optimizes the recognition model through incremental learning.

[0018] Furthermore, the energy collection unit in the composite energy management module adopts a complementary design, with piezoelectric materials responsible for high-frequency energy collection and electromagnetic coils responsible for medium and low-frequency energy collection. The energy management chip is used to achieve adaptive switching of dual-channel inputs and maximum power point tracking.

[0019] Furthermore, the UWB link in the anti-interference communication module adopts bilateral two-way ranging technology, calculates the distance by measuring the round-trip time of the signal, and realizes three-dimensional positioning by combining at least three anchor points. The Bluetooth link is used for device status monitoring and parameter configuration in low-power state.

[0020] Furthermore, the adaptive filtering algorithm in the voiceprint separation interference module adjusts the filter coefficients through the minimum mean square error criterion, adopts a variable step size strategy to accelerate the convergence speed, automatically increases the step size when voice is present, and reduces the step size in a noisy environment.

[0021] Furthermore, the pseudo-random sequence generated by the chaotic circuit in the physical layer encryption module is used as a key stream and is bit-by-bit XOR encrypted with the audio data. During the encryption process, a digital watermark based on the device's unique identification and time information is embedded, and the watermark information is embedded in the audio feature points through quantized index modulation.

[0022] Compared with the existing technology, the beneficial effects of the present invention are:

[0023] In terms of detection accuracy, the system utilizes a four-element cross-array microphone array combined with the MUSIC algorithm to precisely locate recording devices. It can simultaneously identify multiple devices and determine their orientation, effectively avoiding the positioning ambiguity associated with traditional single-microphone systems. A joint time-frequency domain analysis algorithm captures the unique electrical characteristics of recording devices, significantly reducing false alarm rates and maintaining high reliability even in complex environments.

[0024] The design of the interference generation module balances interference effectiveness with normal voice communication. Dynamic spectrum access technology enables real-time spectrum sensing of the ambient spectrum, precisely avoiding voice and civilian communication bands. This technology prevents recording equipment from being jammed while preserving clear voice intelligibility, making it suitable for scenarios requiring normal communication, such as meetings and negotiations. The combination of adaptive comb jamming and chaotic spread spectrum signals enables customized jamming strategies for different types of recording equipment, improving jamming efficiency.

[0025] The energy efficiency management and communication security design of the system further enhances the practicability and reliability. The composite energy collection module collects environmental energy in dual-mode of piezoelectricity and electromagnetism, significantly prolongs the device endurance time, and reduces the maintenance cost. The low-power design enables the device to maintain stable performance in a long working state. The anti-interference communication module adopts UWB and Bluetooth hybrid networking to realize high-precision positioning and secure data transmission, supports multi-device collaborative work, and expands the protection range. The physical layer encryption technology provides reliable protection for the recording evidence, effectively prevents data tampering and illegal acquisition, and safeguards information security. These innovative designs make the technology of the present application have a wide application prospect in confidential places, important meetings and other scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 A schematic block diagram of a control system of a controllable anti-recording device according to the present application is shown in the figure;

[0027] Figure 2 A frequency spectrum utilization rate optimization comparison combination diagram of the interference module according to the present application is shown in the figure;

[0028] Figure 3 A comparison diagram of the endurance time of the traditional lithium battery and the dual-energy endurance of the system in different working modes is shown in the figure;

[0029] Figure 4 A combination diagram for evaluating the effect of the voiceprint separation interference module is shown in the figure. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0031] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0032] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the said features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below with reference to the accompanying drawings.

[0033] Reference Figures 1 to 4 :A control system for a controllable anti-recording device, comprising the following modules:

[0034] The multi-dimensional audio detection module utilizes a four-element cross-array MEMS microphone array, using Knowles SPM0404LR5H-MIC elements. Its built-in differential amplifier effectively suppresses common-mode noise, achieving a sensitivity of -42dBV / Pa and a frequency response range of 20Hz-20kHz. Precisely matched PCB trace lengths achieve an 8mm element spacing with an error within ±0.1mm, ensuring spatial consistency of the array.

[0035] The module uses a cross-correlation calibration algorithm for real-time self-calibration. Using a sliding window, it calculates the cross-correlation function of adjacent array element signals, completing phase difference compensation within 5ms with an accuracy of less than 0.5°, ensuring array signal synchronization. The joint time-frequency domain analysis process involves performing a 1024-point Hamming window short-time Fourier transform on the 48kHz sampled, 24-bit quantized audio signal, shifting the frame by 512 points. This generates a spectrum with a time resolution of 23.4375ms and a frequency resolution of 46.875Hz. A three-layer db4 wavelet packet decomposition is then performed to extract a 128-dimensional feature vector, containing the approximate coefficients and the energy distribution of each detail coefficient, enabling multi-dimensional feature extraction of the audio signal.

[0036] For multi-source localization, the MUSIC algorithm is used for spatial spectrum estimation. By constructing an 8×8 covariance matrix and separating the signal and noise subspaces through eigenvalue decomposition, the golden section method is used to accelerate spectral peak search, completing azimuth estimation within ±7.8° within 15ms. This allows for simultaneous direction-of-arrival estimation of multiple signal sources, enabling simultaneous localization of multiple recording devices. A multi-dimensional analysis is used to extract features specific to different recording devices: mobile phone recordings typically exhibit ADC conversion noise peaks in the 8-12kHz range, with sampling clock harmonics concentrated at 44.1kHz and its multiples. Professional recording devices exhibit flatter spectral characteristics in the high-frequency range (15-20kHz), enabling device type identification.

[0037] Interference Generation Module: This module utilizes the AD9854 DDS chip as a broadband signal source. Its 180MHz system clock supports 1mHz frequency resolution and enables nanosecond frequency switching via the SPI interface. It utilizes direct digital frequency synthesis technology in conjunction with a digital signal processor to generate an adaptive comb interference spectrum. Dynamic spectrum access technology allows real-time detection of available frequency bands with 10kHz resolution, avoiding communication and voice bands.

[0038] Introducing spectrum utilization optimization formula ,in Represents spectrum utilization, which is a key indicator for measuring the module's spectrum utilization efficiency; is the bandwidth of the i-th available frequency band, reflecting the available width of each frequency band; is the frequency band energy utilization efficiency, which reflects the effectiveness of frequency band energy utilization; n is the total number of available frequency bands, It is the frequency offset adjustment coefficient, which is used to adjust the impact caused by frequency offset; and are the interference signal and target frequency band center frequencies respectively, is the maximum frequency range of the system, is the total available bandwidth, Optimization factor for dynamic spectrum access.

[0039] The chaotic spread spectrum signal is generated using Logistic mapping, which is a simple and widely used chaotic mapping model that can generate sequences with good chaotic properties. Its expression is , where x nThe value of the nth iteration, μ is a control parameter, when μ takes a certain range of values, the system will show chaotic state. Here μ is 3.99, the generated sequence has the required chaotic characteristics. The generated sequence is converted to an analog signal by an 8-bit DAC, and then smoothed by a fourth-order low-pass filter, with a bandwidth of 4000 kHz and a power spectral density of -10 dBm / Hz. The adaptive comb interference mechanism monitors the spectrum of the recording device in real time. When a specific frequency component is detected for more than 30 ms, a 25 dB deep notch is generated at ± 500 Hz, implemented by a biquad IIR filter, with a center frequency update period of 20 ms and a stopband attenuation rate of ≥ 60 dB / octave. The phase-coded pulse train generation uses a 13-bit Barker code to modulate an 8 kHz carrier, with a pulse width of 100 μs and a duty cycle of 20%, and the baseband signal is shifted to the target frequency band by digital up-conversion technology.

[0040] The dynamic spectrum access process scans the 20 Hz-20 kHz frequency band with 10 kHz resolution, dwells at each frequency point for 5 ms, and judges the occupation state by energy detection. The spectrum usage database records civilian communication frequency bands (such as 3.5 GHz) and voice-sensitive frequency bands (300 Hz-3.4 kHz). The particle swarm optimization algorithm is used to allocate interference power, and at the same time, the module generates a sine wave signal through a lookup table, which is converted to an analog signal and filtered by a low-pass filter, supporting a frequency agility speed of ≤ 1 μs, and can generate a frequency hopping interference signal, which suppresses the signal-to-noise ratio of the recording device by more than 20 dB. Actual measurement shows that the module suppresses the signal-to-noise ratio of the Samsung Galaxy S23 recording function by 22 dB, and in a 10 square meter conference room environment, the isolation between the interference signal and the voice frequency band is ≥ 30 dB, and the PESQ score is 3.8, meeting the normal conversation requirements.

[0041] The heterogeneous computing master module adopts a heterogeneous architecture of STM32H750XB dual-core Cortex-M7 (400 MHz) and ADSP-21489 DSP, realizes data interaction through a high-speed AXI bus, and shares a 1M B SRAM cache 102 type recording device feature template. The signal processing flow adopts a three-stage pipeline architecture: the first stage is completed by the DSP, which calculates the 13-order Mel frequency cepstrum coefficient (MFCC) and adopts fast discrete cosine transform (DCT) optimization, and the processing time of each frame is ≤2 ms; MFCC, Mel-Frequency Cepstral Coefficients, is a feature parameter widely used in speech and audio processing, which simulates the perceptual characteristics of the human auditory system to different frequency sounds, converts the audio signal to the Mel frequency scale for analysis, and can effectively extract the feature information of the audio; the second stage is completed by the CPU to extract a 24-dimensional feature vector, including spectral centroid, bandwidth, skewness, and other statistical features; and the third stage is classified and recognized through a three-stage decision tree (depth 6), and the decision threshold of each layer is 0.7 / 0.85 / 0.95, and the decision delay is controlled within 30 ms.

[0042] The decision tree classification confidence formula is introduced as follows: , wherein C is the comprehensive classification confidence, T i is the i-th decision tree threshold, S i is the decision score of the corresponding level, and k is the slope parameter (k=10 is taken).

[0043] The 8x8 capacitive sensor array adopts mutual capacitance detection principle, recognizes gestures by detecting the capacitance change between electrodes, and supports five gestures such as sliding, clicking, and long pressing. The working mode switching logic is as follows: when double-finger sliding is detected, the intelligent mode is switched to the strong interference mode, and the system automatically increases the output power to 120 mA; single-finger clicking switches to the mute mode, which closes the interference signal but keeps the detection function, and the power consumption is reduced to 35 mA.

[0044] The edge end incremental learning adopts an online gradient descent algorithm, triggers model updating once every 100 new samples, and the updating time is 7.2 seconds. The adaptive learning rate formula is introduced as follows: , wherein is the learning rate of the t-th update, is the initial learning rate (0.01 is taken), is the decay coefficient (0.001 is taken), and n is the number of updates. Through the formula, the model can still maintain a noise suppression capability of 16 dB in a noisy restaurant environment (background noise 70 dB).

[0045] The signal processing algorithm adopts a multi-level feature extraction architecture, first extracts audio features through mel frequency cepstral coefficients, then uses decision trees for classification and recognition, and supports online updating of the built-in feature template library through incremental learning to continuously optimize the recognition model.

[0046] The composite energy management module: the energy collection unit adopts a complementary design of PZT-5A piezoelectric ceramic sheet (10mm×10mm×0.5mm) and a 5mm diameter 200 turn electromagnetic coil. The piezoelectric ceramic sheet is made of material with d33 coefficient ≥300pC / N, which can generate 50μW power under 70dB, 1kHz noise excitation, and is mainly responsible for high frequency energy collection; the electromagnetic coil has an inductance of 5mT, and with the help of customized Fe-Si-Al alloy magnetic core material, the magnetic flux density is increased to 8mT, which outputs 30μW under 5mT, 50Hz magnetic field environment, and is responsible for medium and low frequency energy collection.

[0047] The dual-channel energy is processed by LTC3588-1 (piezoelectric rectification) and MAX6660 (electromagnetic conversion) chips respectively, and then input to the TPS62842 energy management chip to realize maximum power point tracking (MPPT) with tracking efficiency ≥95%, achieve adaptive switching of dual-channel input, and ensure efficient use of energy. To further optimize energy management, the energy management efficiency formula is introduced , wherein is the energy management efficiency, which is a key indicator of the energy management level of the module; and are the piezoelectric and electromagnetic unit collection powers respectively, reflecting the energy collection capabilities of the two units; is the energy conversion efficiency, which reflects the effectiveness of the energy conversion process; is the system loss power, is the total input energy. The energy storage system is composed of a 2200mAh lithium-sulfur battery (energy density 300Wh / kg) and a 0.22F super capacitor (equivalent series resistance ≤50mΩ). When the system starts, the super capacitor provides a peak current of 150mA within 10ms to ensure that the DSP wakes up quickly; when working normally, the battery supplies power at a constant current of 85mA; when the detected environmental noise ≥65dB and lasts for more than 10 seconds, it automatically switches to the energy collection priority mode, which prioritizes charging the super capacitor. This module is equipped with a lithium-sulfur battery with an energy density of 300Wh / kg and a Qi protocol wireless charging coil, and in intelligent mode, it can last for 15 hours, with a sleep power consumption as low as 8μA, which can maintain the system clock running for more than 30 days, and the equivalent series resistance of the super capacitor is ≤50mΩ, supporting millisecond-level startup. The charging system supports 5V / 1A wired fast charging (1.5 hours to full) and Qi protocol wireless charging, with a charging distance of up to 10mm.

[0048] Anti-interference communication module: DW1000 UWB chip is used, which works in 3.5-6.5GHz frequency band, uses double-sided two-way ranging (TWR) mode, calculates distance through sending-receiving-reply timestamp, and the ranging accuracy can reach ±4.8cm. With microstrip patch antenna, its gain is 3dBi, and the beam width is 120°, which supports multi-device positioning with the help of time division multiplexing mechanism. Combined with at least three anchor points, three-dimensional positioning can be realized, and the positioning accuracy reaches centimeter level. The nRF52840 Bluetooth 5.3 chip supports 2Mbps data rate, and the transmission distance reaches 100 meters when using LongRange mode, which is used for device state monitoring and parameter configuration in low power consumption state. Through Mesh networking protocol, 8 devices work cooperatively, the network topology is star structure, the master device is responsible for data aggregation and decision distribution, so that UWB and Bluetooth link cooperate to achieve wide range coverage and accurate positioning.

[0049] Dynamic frequency hopping technology works at a rate of 2000 hops / second, and the frequency hopping sequence is generated by a 16-bit linear feedback shift register (LFSR). (LFSR, Linear Feedback Shift Register, is a kind of timing logic circuit widely used in digital circuits. It is composed of shift register and feedback logic, which generates a specific sequence by continuously shifting and performing XOR operation on some bits of the register according to the feedback logic. In this scenario, it is used to generate frequency hopping sequence to realize dynamic frequency hopping function) The polynomial is , covering 79 channels of 2.4GHz ISM band, enhancing communication anti-interference ability. LDPC encoding complies with IEEE802.11n standard, code rate 1 / 2, code length 1024, using sum-product algorithm (SPA) iterative decoding, the bit error rate is ≤10 −6 at 5dB SNR, ensuring data transmission accuracy. Communication encryption uses SM4 algorithm (128-bit key), when generating key, first extract device unique ID (64-bit) and current timestamp (32-bit), generate 256-bit intermediate value by SHA-256 hash, and intercept the first 128 bits as encryption key, updated every 3 minutes, to ensure communication security. Remote control APP sends configuration instructions through BLE channel, parameter update delay is 95ms, supports real-time spectrum display (resolution 1Hz), interference parameter adjustment (frequency range, intensity, etc.) and device state monitoring (battery power, temperature, etc.) functions.

[0050] The present invention also includes a voiceprint interference separation module: it utilizes an adaptive filtering system based on the variable-step-size LMS (VS-LMS) algorithm, which adjusts filter coefficients according to the minimum mean square error criterion. This variable step-size strategy automatically increases the step-size when speech is present, accelerating convergence and rapidly tracking signal changes; it reduces the step-size in noisy environments to improve filter stability, thereby achieving a balance between filtering performance and stability, effectively separating speech signals from recording device signals. The system is equipped with a 128-order transversal filter, and the step-size factor μ is dynamically adjusted based on the voice activity detection (VAD) results. VAD uses a dual-threshold method, identifying speech as occurring when the short-term energy exceeds 30dB and the zero-crossing rate is between 10 and 100 times per frame. If the VAD probability is ≥ 0.8, μ is set to 0.05; if the VAD probability is < 0.3, μ is set to 0.001.

[0051] Introducing the voiceprint separation formula ,in Voiceprint separation is a key indicator for measuring the ability to separate speech and interference signals; is the pure voice power, and are the interference signal and noise power respectively, λ is the phase influence factor, and are the phases of the original speech and the interference wave respectively. This formula optimizes voiceprint separation by integrating multiple factors.

[0052] Reference signal generation is based on multi-channel adaptive beamforming, using the MUSIC algorithm to estimate the recording device's position. Specifically, after using the MUSIC algorithm to estimate the recording device's position, the system uses this position information to accurately calculate the spatial relationship between each microphone channel and the recording device. Based on this information, the weights of each microphone channel are adjusted according to a specific weight calculation model. This model comprehensively considers factors such as the distance and angle between the microphone and the recording device, as well as the attenuation during signal propagation. This ensures that each channel's signal is appropriately weighted in subsequent processing, laying the foundation for generating a high-quality anti-phase signal. The microphone channel weights are then adjusted to generate an anti-phase signal with a phase error within 4.2°, which interferes with the recording device. Dynamic mask generation is performed in the 8kHz-16kHz frequency band, using piecewise linear interpolation to divide the frequency band into eight subbands (each 1kHz wide). The subband attenuation coefficients are dynamically adjusted based on the spectral characteristics of the recording device, with a mask bandwidth adjustment step of 100Hz. This approach ensures effective interference without disrupting normal conversation, achieving efficient voiceprint separation and interference control.

[0053] The present invention also includes a physical layer encryption module: a chaotic signal generator is constructed based on an improved Cai circuit, with precisely matched circuit parameters: R0 = 10kΩ (accuracy ±0.1%), R1 = 1.2kΩ, C1 = 10nF (temperature coefficient ±10ppm / °C), C2 = 100nF, and inductance L = 1mH (Q value ≥ 100). The nonlinear resistor is implemented by a special dual op amp circuit, and the characteristic curve is: , where m1=0.7 and m2=0.8. The generated chaotic sequence is sampled by an 8-bit ADC, and the Lyapunov exponent is 0.85, which has good randomness and ergodicity. The chaotic encryption entropy formula is introduced. ,in is the encryption entropy value, n is the symbol set size of the chaotic sequence, is the probability distribution of sequence symbols, i is the state index, λ is the chaos complexity factor, and The synchronization time and bit period are respectively. This formula is used to optimize the encryption process to ensure security and real-time performance.

[0054] During encryption, the 48kHz sampled, 24-bit quantized audio data is first framed (1024 points per frame). A pseudo-random sequence generated by a chaotic circuit is used as the key stream and bit-by-bit XORed with the audio data to produce the ciphertext. Simultaneously, the device ID (8 bytes) and a timestamp (4 bytes) based on the device's unique identifier are embedded in the 2-5kHz mid-frequency band using Quantization Index Modulation (QIM) with an embedding strength of 15dB. A quantization step size of ±1 / 3 is used to ensure robustness. This digital watermark is resistant to compression and clipping. Decryption extracts the watermark through correlation detection to verify data integrity. The module uses a stream cipher mode and audio feature point encoding to embed the physical watermark. It uses a hardware clock to synchronize timestamps within 1ms, achieving a data transmission rate of ≥128Mbps after encryption. The USB 3.0 interface supports 480Mbps high-speed data export, but encryption increases file size. The storage format is WAV with an encryption header containing metadata such as the encryption algorithm version and key expiration date.

[0055] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A control system for a controllable anti-recording device, characterized in that: Includes the following modules: Multi-dimensional audio detection module: This module uses a micro-electromechanical system microphone array in a four-element cross array layout. Combined with a time-frequency domain joint analysis algorithm, it first converts the audio signal into the frequency domain through short-time Fourier transform, and then uses wavelet packet decomposition to extract detailed features. Connected to the heterogeneous computing main control module via a high-speed data bus; Interference generation module: A broadband signal source built based on direct digital frequency synthesis technology uses dynamic spectrum access technology to sense available frequency bands in real time and dynamically avoid communication and voice bands. It connects to the heterogeneous computing main control module through a control interface to receive target frequency bands and interference intensity control parameters. Heterogeneous computing master control module: This module uses a heterogeneous processor to integrate the CPU and signal processing unit, has a built-in multi-feature fusion matching algorithm, stores recording device feature parameters through a template library, and is equipped with a capacitive sensor array to support five working modes and gesture switching. It receives information from the multi-dimensional audio detection module, sends control parameters to the interference generation module, and interacts with the anti-interference communication module through an encrypted communication interface. Composite Energy Management Module: Integrates piezoelectric and electromagnetic dual-mode energy harvesting units, paired with a lithium-sulfur battery and Qi protocol wireless charging coil; adaptively supplies power to each module and monitors energy consumption through a power management bus; Anti-interference communication module: uses ultra-wideband (UWB) and Bluetooth 5.3 hybrid communication links, dynamic frequency hopping technology and LDPC coding for data transmission, and a 128-bit symmetric encryption algorithm for communication encryption; interacts with the heterogeneous computing main control module through an encrypted communication interface; Voiceprint separation interference module: This module is connected to the multi-dimensional audio detection module and the interference generation module through a high-speed data bus: the multi-dimensional audio detection module transmits the extracted audio feature vector ring to the voiceprint separation interference module in real time; the voiceprint separation interference module adopts the variable step size LMS algorithm in the adaptive filtering technology, dynamically adjusts the filtering parameters through voice activity detection, separates the ambient voice and the recording device signal in real time, and introduces the voiceprint separation degree formula ,in is the voiceprint separation, is the pure voice power, and are the interference signal and noise power respectively, λ is the phase influence factor, and The phases of the original speech and interference wave are used to achieve voiceprint separation and interference control through formula optimization; Physical layer encryption module: This module is connected to the heterogeneous computing main control module and the anti-interference communication module through an encrypted data interface: the heterogeneous computing main control module transmits the detected recording evidence data to the physical layer encryption module; The physical layer encryption module is based on the chaotic signal generator designed by Cai's circuit. The generated pseudo-random sequence passes the NISTSP800-22 test and introduces the chaotic encryption entropy formula. ,in is the encryption entropy value, n is the symbol set size of the chaotic sequence, is the probability distribution of sequence symbols, i is the state index, λ is the chaos complexity factor, and The module performs real-time stream encryption on recorded evidence and uses stream cipher mode and audio feature point coding to embed physical watermarks. The interference signal in the interference generation module is generated by combining direct digital frequency synthesis and digital up-conversion technology, generating a sine wave signal through a lookup table, and then outputting it after digital-to-analog conversion and low-pass filtering to generate a frequency hopping interference signal; The spectrum utilization optimization formula in the interference generation module is: ,in is the spectrum utilization, is the bandwidth of the i-th available frequency band, is the energy utilization efficiency of the frequency band, n is the total number of available frequency bands, is the adjustment coefficient, and are the interference signal and target frequency band center frequencies respectively, is the maximum frequency range of the system, is the total available bandwidth, Optimization factor for dynamic spectrum access.

2. The control system of a controllable anti-recording device according to claim 1, characterized in that: The microphone array in the multi-dimensional audio detection module uses a cross-correlation algorithm for real-time self-calibration, uses a MUSIC algorithm for spatial spectrum estimation, and determines the position of the recording device by searching for spatial spectrum peaks.

3. The control system of the controllable anti-recording device according to claim 1, characterized in that: The signal processing algorithm in the heterogeneous computing main control module adopts a multi-level feature extraction architecture, first extracting audio features through Mel-frequency cepstral coefficients, and then using a decision tree for classification and recognition. The built-in feature template library supports online updates and optimizes the recognition model through incremental learning.

4. The control system of a controllable anti-recording device according to claim 1, characterized in that: The energy collection unit in the composite energy management module adopts a complementary design. The piezoelectric material is responsible for high-frequency energy collection, and the electromagnetic coil is responsible for medium and low-frequency energy collection. The energy management chip realizes adaptive switching of dual inputs and maximum power point tracking. The energy management efficiency formula in the composite energy management module is: ,in For energy management efficiency, and Collect power for piezoelectric and electromagnetic units respectively, is the energy conversion efficiency, is the system power loss, is the total input energy.

5. The control system of a controllable anti-recording device according to claim 1, characterized in that: The UWB link in the anti-interference communication module adopts bilateral two-way ranging technology, calculates the distance by measuring the round-trip time of the signal, and realizes three-dimensional positioning by combining at least three anchor points. The Bluetooth link is used for device status monitoring and parameter configuration in low-power state.

6. The control system of a controllable anti-recording device according to claim 1, characterized in that: The adaptive filtering algorithm in the voiceprint separation interference module adjusts the filter coefficients through the minimum mean square error criterion, adopts a variable step size strategy to accelerate the convergence speed, automatically increases the step size when there is voice, and reduces the step size in a noisy environment.

7. The control system of the controllable anti-recording device according to claim 1, characterized in that: The pseudo-random sequence generated by the chaotic circuit in the physical layer encryption module is used as a key stream and is encrypted bit by bit with the audio data. A digital watermark based on the unique device identification and time information is embedded in the encryption process. The watermark information is embedded in the audio feature points through quantization index modulation.

Images