An intelligent recognition device and method for electronic devices based on millimeter-wave sensing

Abstract

The present invention discloses an intelligent recognition device and method for electronic devices based on millimeter-wave sensing. The method obtains the non-linear response signal reflected by the electronic device by setting a simple mmWave sensing detection device, performs wavelet signal analysis after preprocessing the non-linear response signal, extracts features based on the characteristic values of each dimension of the signal, and then uses an intelligent algorithm to realize the recognition of the electronic device based on millimeter-wave sensing. The method proposed by the present invention for intelligent analysis and recognition using the unique non-linear response signal reflected by the electronic device to mmWave sensing is simple and effective, with high accuracy, precision, stability in recognition and low cost.

Description

Technical Field

[0001] The present invention belongs to the field of millimeter-wave sensing, and particularly relates to an intelligent identification device and method for electronic devices based on millimeter-wave sensing. Background Art

[0002] Millimeter wave (mmWave) is an electromagnetic wave with a wavelength in the range of 1 mm to 10 mm, which is in the wavelength range where microwaves and far-infrared waves overlap. Due to the short wavelength and narrow beam of millimeter waves, they have good directivity. Therefore, they can penetrate non-metallic materials, image metal objects, measure unique data information such as physical distance, speed, and angle, and at the same time have the ability to reflect different targets, and can detect specific characteristics of different objects within the detection range. Millimeter waves have strong measurement resolution and high accuracy.

[0003] Currently, the detection and identification of electronic devices mainly use X-ray devices, which are bulky and costly. The identification method based on millimeter-wave sensing is mainly used for the detection and tracking of humans or non-electronic objects and cannot identify objects. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention proposes an intelligent identification device and method for electronic devices based on millimeter-wave sensing. By using the intelligent analysis result of the non-linear response signal of an object detected by millimeter-wave sensing, this method can not only detect electronic devices but also intelligently identify them. The device is portable, has high accuracy, accuracy, stability in identification, and low cost.

[0005] The object of the present invention is achieved by the following technical solutions:

[0006] An intelligent identification device for electronic devices based on millimeter-wave sensing, the device includes a front-end millimeter-wave detector and an intelligent identification module for electronic devices;

[0007] The front-end millimeter-wave detector includes a radio frequency unit and a baseband unit; the radio frequency unit is used to realize signal transmission and reception, that is, modulate, amplify, and transmit the signal output by the baseband unit, receive, amplify, and demodulate the radio frequency signal; the baseband unit realizes modulation and demodulation of baseband signals, generates millimeter-wave signals in the 24 GHz band, acquires the reflected signals of electronic devices, and sends the processed reflected signals to the intelligent identification module for electronic devices;

[0008] The intelligent identification module for electronic devices is used to receive the reflected signals of electronic devices sent by the front-end millimeter-wave detector, preprocess them to remove interference and noise, and then extract effective features through wavelet analysis, so as to identify the classification of electronic devices according to the eigenvalue of the signal.

[0009] An intelligent recognition method for electronic devices based on millimeter-wave sensing, the method specifically includes the following steps:

[0010] S1: Send millimeter-wave signals through the transmitting antenna of the front-end millimeter-wave detector, and capture the reflected signals of the object through the receiving antenna. The reflected signals include linear response signals and non-linear response signals r(t);

[0011] S2: Preprocess the non-linear response signal r(t), that is, perform smoothing filtering on it through a filtering method to remove interference and noise in the signal, and perform signal demodulation to remove the DC component of the signal to obtain an effective non-linear response signal y(t);

[0012]

[0013] Among them, is the approximation signal, is the detail signal; represents the mother wavelet function, satisfying dynamic scaling and translation; a0, a, and b are the scale factor and the corresponding translation parameters respectively, F W (a0, b), F W (a, b) are the correlation coefficients;

[0014] S3: Through wavelet-based signal analysis, decompose the non-linear response signal into an approximation signal and a detail signal, that is, the low-frequency part and the high-frequency part of the signal; and when analyzing the wavelet, in order for the inverse transform to exist, it is necessary to satisfy the admissible condition, that is

[0015]

[0016] Among them, is φ(ω), the Fourier transform of the function, is the constant of the corresponding wavelet; ω is the integration variable;

[0017] S4: Define the quantitative values of statistical features in the time domain and frequency domain, and extract the signal features of the decomposed non-linear response signal; S5: According to the eigenvalue of the signal, identify and classify the electronic device through a classification algorithm.

[0018] Furthermore, the quantitative values of the statistical features in the time domain defined in S4 are specifically as follows:

[0019] Mean value:

[0020] Standard deviation: Skewness coefficient: Kurtosis coefficient: Root mean square amplitude: Minimum value: Highest value: The quantitative values of the frequency-domain statistical features are specifically as follows:

[0021] Mean value:

[0022] Standard deviation: Skewness coefficient: Kurtosis coefficient: Peak factor: Flatness: Where N is the total amount of data when calculating the statistical feature values, x(i) is the time-domain data value corresponding to the i-th data, and y(i) is the frequency-domain data value corresponding to the i-th data.

[0023] Furthermore, in the S5, the electronic device is identified and classified by the SVM or KNN classification algorithm.

[0024] Furthermore, the filtering method in the S2 is the Savizky-Golay filtering method.

[0025] The beneficial effects of the present invention are as follows:

[0026] The intelligent identification device and method for electronic devices based on millimeter-wave sensing of the present invention obtain the non-linear response signal reflected by the electronic device through a simple millimeter-wave sensing detection device, perform wavelet signal analysis after preprocessing the non-linear response signal, and extract features based on the characteristic values of each dimension of the signal, thereby realizing the intelligent identification of electronic devices based on millimeter-wave sensing. This method proposes to use the unique non-linear response signal reflected by the electronic device to the mmWave sensing for intelligent analysis and identification. The method is simple and effective, with high accuracy, accuracy, and stability in identification, and low cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the intelligent identification device for electronic devices based on millimeter-wave sensing of the present invention;

[0028] Figure 2 It is a flowchart of the intelligent identification method for electronic devices based on millimeter-wave sensing of the present invention;

[0029] Figure 3 It is a performance graph of the identification method under different numbers of electronic devices. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] The present invention will be described in detail below according to the drawings and preferred embodiments. The purpose and effects of the present invention will become more apparent. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0031] AsFigure 1 As shown in the figure, the intelligent identification device for electronic devices based on millimeter-wave sensing of the present invention includes a front-end millimeter-wave detector and an intelligent identification module for electronic devices.

[0032] The front-end millimeter-wave detector transmits continuous millimeter-wave signals through the transmitting antenna Tx, and processes or demodulates the signals reflected by the object obtained through the receiving antenna Rx. The front-end millimeter-wave detector includes a radiofrequency (RF) and a baseband (BB) unit. The RF unit is used to realize signal transmission and reception, that is, modulate, amplify and transmit the signals output by the baseband unit, receive, amplify and demodulate the RF signals; the baseband unit realizes modulation and demodulation of baseband signals, generates millimeter-wave signals in the 24 GHz frequency band, acquires the reflected signals of the electronic devices, and sends the processed reflected signals to the intelligent identification module for electronic devices.

[0033] The intelligent identification module for electronic devices is used to receive the reflected signals of the electronic devices sent by the front-end millimeter-wave detector, preprocess them to remove interference and noise, and then extract effective features through wavelet analysis, so as to identify the classification of the electronic devices according to the characteristic values of the signals.

[0034] When the front-end millimeter-wave detector continuously transmits millimeter-wave signals with a transmission frequency in the 24 GHz frequency band at the Tx antenna end, the objects within the millimeter-wave coverage range, especially electronic devices, can generate two types of RF signals for the reflection of millimeter-wave signals. One is a linear response signal, the main carrier frequency of which is the same as the signal sent by the detector, and the change in phase is related to the distance, shape and size of the object, but the linear response signal cannot reflect the characteristics of the object. The other is a non-linear response signal, which comes from the reflection of the electronic device to the detected millimeter-wave signal. When the electronic device enters the millimeter-wave coverage range, components such as chips, connectors and metal wirings on the printed circuit board (PCB) in the electronic device are equivalent to an antenna array under the resolution of millimeter waves. At the same time, components such as inductance (L), capacitance (C) and resistance (R) related to the antenna array become processors for millimeter-wave signals. Therefore, the expression of the non-linear response signal r(t) reflected by the electronic device can be obtained as follows:

[0035]

[0036] where z(t) is the reflected signal, is the non-linear modulation function of the PCB board, is the complex power series of the non-linear system, represents the convolution calculation, h f (t) is an ideal band-pass filter with a carrier bandwidth.

[0037] After the millimeter-wave signal modulated by the PCB of the electronic device is radiated from the electronic device, it will be captured by the Rx end of the millimeter-wave detector. This non-linear response signal contains the physical characteristics of the electronic device. By performing signal processing and analysis based on intelligent algorithms, the electronic device is identified.

[0038] Therefore, as Figure 2 shown, the intelligent identification method of an electronic device based on millimeter-wave sensing according to the present invention specifically includes the following steps:

[0039] S1: Transmit a millimeter-wave signal through the transmitting antenna of the front-end millimeter-wave detector, and capture the reflected signal of the object through the receiving antenna. This reflected signal includes a linear response signal and a non-linear response signal r(t);

[0040] S2: Preprocess the non-linear response signal r(t), that is, perform smoothing filtering on it through a filtering method to remove interference and noise in the signal, and perform signal demodulation to remove the DC component of the signal to obtain an effective non-linear response signal y(t);

[0041]

[0042] Among them, is the approximation signal, is the detail signal; represents the mother wavelet function, which satisfies dynamic scaling and translation; a0, a, and b are the scale factor and the corresponding translation parameters respectively, F W (a0, b), F W (a, b) are the correlation coefficients;

[0043] S3: Through wavelet-based signal analysis, decompose the non-linear response signal into an approximation signal and a detail signal, that is, the low-frequency part and the high-frequency part of the signal; and when analyzing the wavelet, in order for the inverse transform to exist, it is necessary to satisfy the admissible condition, that is

[0044]

[0045] Among them, is φ(ω), the Fourier transform of the function, is the constant of the corresponding wavelet; ω is the integration variable.

[0046] S4: Define the quantitative statistical feature values in the time domain and frequency domain, and extract the signal features from the decomposed non-linear response signal. The signal characteristics are obtained by calculating the statistical feature values of the signal. As one of the implementation methods, a total of 13 quantitative statistical feature values in the time domain and frequency domain are defined. These 13 feature quantitative values represent the features of the non-linear response signal from different angles. The feature quantitative values include the mean, standard deviation, skewness coefficient, kurtosis coefficient, root mean square amplitude, minimum value, and maximum value in the time domain of the signal, and the mean, standard deviation, skewness coefficient, kurtosis coefficient, peak factor, and flatness in the frequency domain of the signal. Among them, the skewness coefficient is a measure of the symmetry of the signal, the kurtosis coefficient estimates whether the signal is a heavy-tailed distribution or a light-tailed distribution in a normal distribution, and the flatness indicates the degree to which they approach the Euclidean space of the same dimension. Specifically as follows:

[0047] The specific time-domain statistical feature quantitative values are as follows:

[0048] Mean:

[0049] Standard deviation: Skewness coefficient: Kurtosis coefficient: Root mean square amplitude: Minimum value: Maximum value: The specific frequency-domain statistical feature quantitative values are as follows:

[0050] Mean:

[0051] Standard deviation: Skewness coefficient: Kurtosis coefficient: Peak factor: Flatness: Where N is the total amount of data when calculating the statistical feature values, x(i) is the time-domain data value corresponding to the i-th data, and y(i) is the frequency-domain data value corresponding to the i-th data.

[0052] S5: According to the feature values of the signal, identify and classify the electronic device through a classification algorithm. The classification algorithm is preferably SVM (Support Vector Machines) or KNN (k-Nearest Neighbor).

[0053] The effects of the recognition method of the present invention will be described below through embodiments.

[0054] The SVM algorithm and the KNN algorithm are used to identify and compare electronic devices. Since the non-linear response signal is collected as the feature vector for electronic device identification, both the SVM algorithm and the KNN algorithm can achieve very good identification performance. Through the ROC (Receiver Operating Characteristic) curve, the equal error rate (EER) value of the SVM algorithm is 0.0044, and the EER value of the KNN algorithm is 0.0647. The lower the equal error rate value, the higher the accuracy of the identification system.

[0055] As Figure 3 shown, in the case of different numbers of electronic devices, the performance of the intelligent identification device and method for electronic devices based on millimeter-wave sensing is evaluated in terms of three aspects: accuracy, precision, and recall. Figure 3 It can be seen that the algorithm has excellent performance and high stability, that is, the accuracy, precision, and recall are all maintained between 99.1% and 100% when detecting different types of device numbers.

[0056] Those of ordinary skill in the art can understand that the above are only preferred examples of the invention and are not used to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions described in the foregoing examples or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, etc. made within the spirit and principle of the invention shall be included within the protection scope of the invention.

Claims

1. An intelligent recognition device for electronic equipment based on millimeter-wave sensing, characterized in that, The device includes a front-end millimeter-wave detector and an intelligent electronic device identification module; The front-end millimeter-wave detector includes a radio frequency unit and a baseband unit; the radio frequency unit is used to realize signal transmission and reception, that is, modulate, amplify and transmit the signal output by the baseband unit, receive, amplify and demodulate the radio frequency signal; the baseband unit realizes modulation and demodulation of baseband signals, generates millimeter-wave signals in the 24 GHz frequency band, acquires the reflected signals of electronic devices, and sends the processed reflected signals to the intelligent electronic device identification module; The reflected signals include linear response signals reflected by electronic devices within the millimeter-wave coverage range and non-linear response signals containing the physical characteristics of electronic devices generated by modulating the millimeter-wave signals by the PCB boards in the electronic devices; the intelligent electronic device identification module is used to receive the electronic device reflected signals sent by the front-end millimeter-wave detector, preprocess them, remove interference and noise, decompose the non-linear response signals reflecting different electronic devices, and then through wavelet analysis, decompose the non-linear response signals into approximation signals and detail signals, extract the effective physical characteristics of electronic devices, and thus identify and classify electronic devices according to the eigenvalue of the signals.

2. An intelligent recognition method for electronic devices based on millimeter-wave sensing, characterized in that, The method specifically includes the following steps: S1: Send millimeter-wave signals through the transmitting antenna of the front-end millimeter-wave detector, and capture the reflected signals of the object through the receiving antenna. The reflected signals include linear response signals reflected by electronic devices within the millimeter-wave coverage range and non-linear response signals r(t) containing the physical characteristics of electronic devices generated by modulating the millimeter-wave signals by the PCB boards in the electronic devices; where z(t) is the reflected signal, is the non - linear modulation function of the PCB board, is the complex power series of the non - linear system, represents the convolution calculation, h f (t) is the ideal band - pass filter of the carrier bandwidth; S2: Preprocess the non-linear response signal r(t), that is, perform smoothing filtering on it by a filtering method to remove interference and noise in the signal, and perform signal demodulation to remove the DC component of the signal to obtain an effective non-linear response signal y(t); Among them, is the approximate signal, is the detail signal; represents the mother wavelet function, satisfying dynamic scaling and translation; a0, a, and b are the scale factor and the corresponding translation parameters, respectively, and F W (a0, b), F W (a, b) are the correlation coefficients; S3: Decompose the non-linear response signal into an approximation signal and a detail signal, that is, the low-frequency part and the high-frequency part of the signal, through wavelet-based signal analysis; and when analyzing the wavelet, in order for the inverse transform to exist, it is necessary to satisfy the admissible condition, that is Among them, is the Fourier transform of the φ(ω) function, is the constant of the corresponding wavelet; ω is the integration variable; S4: Define the statistical feature quantitative values in the time domain and frequency domain, and extract the signal features of the decomposed non-linear response signal; S5: Identify and classify electronic devices through a classification algorithm according to the eigenvalue of the signal.

3. The intelligent recognition method for electronic devices based on millimeter-wave sensing according to claim 2, wherein, The specific time-domain statistical feature quantitative values defined in S4 are as follows: Mean: Standard deviation: Skewness coefficient: Kurtosis coefficient: Root mean square amplitude: Minimum value: Highest value: The specific frequency-domain statistical feature quantitative values are as follows: Mean: Standard deviation: Skewness coefficient: Kurtosis coefficient: Peak factor: Flatness: Where N is the total amount of data when calculating the statistical feature value, x(i) is the time-domain data value corresponding to the i-th data, and y(i) is the frequency-domain data value corresponding to the i-th data.

4. The intelligent recognition method of an electronic device based on millimeter-wave sensing according to claim 3, characterized in that In S5, the electronic devices are identified and classified through SVM or KNN classification algorithms.

5. The intelligent recognition method for electronic devices based on millimeter-wave sensing according to claim 2, wherein The filtering method in S2 is the Savizky-Golay filtering method.

Images