Multi-band mobile phone signal detection and alarm method and system

Abstract

The application provides a multi-band mobile phone signal detection and alarm method and system, relates to the technical field of communication safety monitoring, and comprises the following steps: collecting electromagnetic signals of a target area and preprocessing; obtaining a signal spectrum through fast Fourier transform; extracting mobile phone signal characteristic frequency band information; judging a communication state according to the uplink and downlink frequency band signal strength ratio; calculating the position of a mobile phone signal source and recording a moving track based on a signal strength difference positioning algorithm; judging a use scene according to the track data and the time sequence relationship of the communication state; evaluating a threat level and sending alarm information. The application can realize real-time monitoring and accurate positioning of the mobile phone use behavior in the target area, and improves the safety control efficiency.

Description

Technical Field

[0001] The present invention relates to communication security monitoring technology, and in particular to a multi-band mobile phone signal detection and alarm method and system. Background Art

[0002] Controlling mobile phone communications in specific locations is a crucial aspect of security. Existing mobile phone signal detection equipment typically uses a signal strength threshold method. This method cannot distinguish between uplink and downlink signal characteristics, making it easy to misidentify base station signals as mobile phone signals. Furthermore, effective identification is difficult when multiple mobile phone signal sources are present simultaneously.

[0003] Traditional mobile phone signal positioning methods primarily rely on single signal characteristics for calculation, failing to fully utilize the correlation characteristics of uplink and downlink during mobile phone communication, resulting in insufficient positioning accuracy. The reliability of positioning results is significantly reduced, especially in complex environments with large signal strength fluctuations or obstructions.

[0004] Existing systems for monitoring mobile phone signals lack the ability to dynamically analyze mobile phone usage behavior. They are unable to intelligently identify scenarios based on signal characteristics and location changes, making it difficult to differentiate between mobile phone usage behaviors with different threat levels. Therefore, a monitoring and early warning method that can accurately identify mobile phone signals, precisely locate them, and intelligently assess threat levels is urgently needed. Summary of the Invention

[0005] The embodiments of the present invention provide a multi-band mobile phone signal detection and alarm method and system, which can solve the problems in the prior art.

[0006] A first aspect of an embodiment of the present invention provides a multi-band mobile phone signal detection and alarm method, comprising:

[0007] Collecting electromagnetic signals in a target area and performing preprocessing to obtain preprocessed electromagnetic signals;

[0008] Performing a fast Fourier transform on the preprocessed electromagnetic signal to obtain a signal spectrum, extracting characteristic frequency band information of the mobile phone signal from the signal spectrum, wherein the characteristic frequency band information of the mobile phone signal includes an uplink frequency band signal strength value and a downlink frequency band signal strength value, and determining the communication status of the mobile phone based on the ratio of the uplink frequency band signal strength value to the downlink frequency band signal strength value;

[0009] Obtain signal strength data within the target area. Based on the mobile phone's communication status and signal strength data, use the signal strength difference positioning algorithm to calculate the real-time location coordinates of the mobile phone signal source, and record the changes in the location coordinates over time to obtain movement trajectory data.

[0010] According to the time sequence relation of the mobile trajectory data and the mobile phone communication state, the current mobile phone use scene is judged, the mobile phone use scene is evaluated according to the pre-set scene threat level parameter, and the alarm information is sent to the monitoring terminal according to the threat level evaluation result.

[0011] In an alternative embodiment,

[0012] The electromagnetic signals in the target area are collected and preprocessed to obtain preprocessed electromagnetic signals, including:

[0013] The mobile phone signal full-band electromagnetic signals in the target area are collected, the signal strength of the electromagnetic signals is detected in real time, the signal gain is automatically adjusted according to the signal strength, the amplitude of the electromagnetic signals is maintained within a pre-set dynamic range, and the gain-adjusted signals are obtained;

[0014] The gain-adjusted signals are subjected to multi-layer wavelet decomposition to obtain wavelet coefficients of different scales, the local variance of the wavelet coefficients is calculated, the adaptive threshold is determined according to the local variance, the wavelet coefficients are subjected to soft threshold processing using the adaptive threshold, and the denoised signals are obtained through wavelet reconstruction;

[0015] The denoised signals are divided into a plurality of time windows, the signals in each time window are subjected to Fourier transform to obtain frequency spectra, the frequency spectra are combined to obtain time-frequency spectra, the mean square error of each time-frequency point in the time-frequency spectrum is calculated, the noise power spectrum of each time-frequency point is estimated based on the mean square error, the corresponding Wiener filter coefficients are calculated according to the noise power spectrum, and the denoised signals are subjected to filtering processing using the Wiener filter coefficients to obtain the preprocessed electromagnetic signals.

[0016] In an alternative embodiment,

[0017] The preprocessed electromagnetic signals are subjected to fast Fourier transform to obtain signal frequency spectra, and the mobile phone signal feature band information is extracted from the signal frequency spectra, including:

[0018] The preprocessed electromagnetic signals are subjected to wavelet decomposition using an orthogonal wavelet basis function to obtain a plurality of frequency band coefficients, the energy values of the frequency bands are calculated according to the frequency band coefficients, the energy values are compared with pre-set frequency band thresholds to determine target frequency bands, the phase of the frequency band coefficients of the target frequency bands is compensated, and the compensated frequency band coefficients are subjected to Fourier transform to obtain target frequency spectrum signals;

[0019] The target frequency spectrum signals are divided into uplink sub-frequency bands and downlink sub-frequency bands, the probability density functions of the uplink sub-frequency bands and the downlink sub-frequency bands are constructed respectively, the adaptive threshold is calculated based on the probability density functions to obtain the optimal segmentation point, and the adaptive threshold is used as a signal peak value detection threshold;

[0020] performing morphological filtering on the spectrum signals of the uplink sub-band and the downlink sub-band, performing peak value searching on the filtered spectrum signals by using the signal peak value detection threshold, obtaining signal peak value positions, calculating a signal bandwidth according to the signal peak value positions, and setting an integer multiple of the signal bandwidth as an energy aggregation interval;

[0021] performing energy density integration on the spectrum in the optimized integration interval to obtain uplink frequency band signal strength values and downlink frequency band signal strength values.

[0022] In an optional embodiment,

[0023] determining the mobile phone communication state according to a ratio of the uplink frequency band signal strength values and the downlink frequency band signal strength values includes:

[0024] performing time domain sliding processing on the uplink frequency band signal strength values and the downlink frequency band signal strength values, setting a decreasing weight coefficient according to the order of sampling time, multiplying the decreasing weight coefficient by the signal strength values at corresponding time points and accumulating to obtain smoothed uplink signal strength values and smoothed downlink signal strength values;

[0025] calculating a ratio of the smoothed uplink signal strength values and the smoothed downlink signal strength values, extracting fluctuation amplitudes and change rates of the ratio, constructing a feature matrix according to the fluctuation amplitudes and the change rates, and performing singular value decomposition on the feature matrix to obtain main feature components;

[0026] establishing a decision function based on the main feature vectors, calculating an optimal segmentation point of the decision function by using the maximum inter-class variance method, and taking the optimal segmentation point as a communication state discrimination threshold;

[0027] dividing a value range of the signal strength ratio into a normal communication interval and an abnormal communication interval according to the discrimination threshold, calculating the ratio of the smoothed uplink signal strength values and the smoothed downlink signal strength values in real time, determining a normal communication state when the current ratio falls into the normal communication interval, determining a link fault type according to a relative position relationship between the current ratio and the normal communication interval when the current ratio falls into the abnormal communication interval, and outputting a corresponding communication state determination result.

[0028] In an optional embodiment,

[0029] obtaining signal strength data in a target area, calculating real-time position coordinates of a mobile phone signal source by using a signal strength difference positioning algorithm based on the mobile phone communication state and the signal strength data, and recording changes of the position coordinates over time to obtain movement trajectory data includes:

[0030] Set up multiple signal acquisition nodes in the target area to collect signal strength data in multiple directions;

[0031] Calculating a ratio of a short-time standard deviation to a long-time standard deviation of the signal strength data according to a pre-acquired mobile phone communication state, and screening a valid signal detector according to the ratio;

[0032] Calculating a signal strength change trend collected by an effective signal detector, determining a smoothing window length based on the change trend, determining a spatial weight based on the spatial distribution position of the signal detector, and performing weighted smoothing on the spatial weight and signal strength data to obtain a smoothed signal strength value;

[0033] Select any two valid signal detectors to form a detector pair, calculate the difference in smoothed signal strength values ​​between the detector pairs, dynamically adjust the path loss index based on the mobile phone communication status, and calculate the distance difference between the detector pairs;

[0034] Calculating the signal correlation and the stability of the signal strength difference of the detector pair to obtain a credibility score, constructing a hyperbolic equation based on the distance difference, using the credibility score as the initial weight of the hyperbolic equation, and solving the hyperbolic equation system in an iterative manner to obtain the real-time position of the signal source;

[0035] The moving speed and direction are calculated according to the real-time position of the signal source within the continuous time window, and the abnormal position is constrained and corrected in combination with the mobile phone communication status to obtain a smooth moving trajectory.

[0036] In an optional embodiment,

[0037] Calculating the signal correlation of the detector pair and the stability of the signal strength difference to obtain a credibility score, constructing a hyperbolic equation based on the distance difference, using the credibility score as the initial weight of the hyperbolic equation, and solving the hyperbolic equation system in an iterative manner to obtain the real-time position of the signal source includes:

[0038] Obtaining a signal strength sequence of the detector pair, calculating a cross-correlation coefficient of the signal strength sequence and a standard deviation of a signal strength difference sequence, and performing an adaptive weighted combination of the cross-correlation coefficient and the inverse of the standard deviation to obtain an initial credibility score of the detector pair;

[0039] Calculate the directional angle of the line connecting the detector pair, determine the spatial distribution weight based on the distribution of the directional angle, and multiply the spatial distribution weight by the initial credibility score to obtain a revised credibility score for the detector pair;

[0040] weighting the distance differences between the detector pairs using the modified credibility score, constructing a hyperbolic equation system based on the weighted distance differences, and solving the hyperbolic equation system using a constrained weighted least squares method to obtain an initial position estimate;

[0041] Obtaining the signal source position of the last time, combining the maximum moving speed constraint to determine the position constraint region of the current time, and performing adaptive step iterative search in the position constraint region to obtain the current iteration position, taking the initial position estimate value as the starting point.

[0042] Based on the current iteration position, the theoretical distance difference value of the detector pair is calculated, the deviation of the theoretical distance difference value from the measured distance difference value is taken as the positioning residual, the real-time confidence score is obtained by dynamically attenuating and updating the correction confidence score according to the positioning residual, and the weight of the hyperbolic equation set is updated using the real-time confidence score, and the iteration is continued until the position estimate value converges to obtain the real-time position of the signal source.

[0043] In an optional embodiment,

[0044] According to the time sequence relationship between the moving trajectory data and the mobile phone communication state, the current mobile phone use scene is judged, the threat level of the mobile phone use scene is evaluated according to the pre-set scene threat level parameter, and the alarm information is sent to the monitoring terminal according to the threat level evaluation result, including:

[0045] Obtaining the mobile phone's moving trajectory data and communication state data, extracting the motion features and spatial distribution features in the moving trajectory data, extracting the communication behavior features and signal quality features in the communication state data, and generating time sequence feature data;

[0046] The time sequence feature data is segmented by time window, the correlation degree between the features in each time window is calculated, a feature correlation matrix is constructed, and the feature correlation matrix is matched with a pre-set mobile phone use scene model to obtain the current mobile phone use scene;

[0047] The time sequence feature data is compared and analyzed with the pre-set normal scene features corresponding to the current mobile phone use scene, a feature deviation sequence is generated, the feature deviation sequence is time sequence clustered to obtain an abnormal event cluster, and an abnormal event feature is generated based on the time span and the number of occurrences of the abnormal event cluster;

[0048] According to the abnormal event feature, a threat score is calculated, a pre-set threat level parameter corresponding to the current mobile phone use scene is weighted to obtain a real-time threat level, when the real-time threat level exceeds a warning threshold, the abnormal event feature triggering the warning is taken as a monitoring template, the number of matched abnormal events is counted in an observation period, and when the number of matched abnormal events exceeds a set threshold and the threat level continues to rise, an alarm information is sent to the monitoring terminal.

[0049] In a second aspect of the embodiment of the application, a multi-band mobile phone signal detection and alarm system is provided, comprising:

[0050] The first unit is configured to collect electromagnetic signals in a target area and pre-process the electromagnetic signals to obtain pre-processed electromagnetic signals.

[0051] The second unit is configured to perform fast Fourier transform on the pre-processed electromagnetic signals to obtain signal frequency spectrum, extract mobile phone signal characteristic frequency band information from the signal frequency spectrum, wherein the mobile phone signal characteristic frequency band information includes uplink frequency band signal strength value and downlink frequency band signal strength value, and determine the mobile phone communication state according to the ratio of the uplink frequency band signal strength value and the downlink frequency band signal strength value.

[0052] The third unit is configured to obtain signal strength data in the target area, calculate real-time position coordinates of the mobile phone signal source based on the mobile phone communication state and the signal strength data by using a signal strength difference positioning algorithm, and record the change of the position coordinates over time to obtain movement trajectory data.

[0053] The fourth unit is configured to determine the current mobile phone use scenario according to the time sequence relationship between the movement trajectory data and the mobile phone communication state, perform threat level evaluation on the mobile phone use scenario according to pre-set scenario threat level parameters, and send alarm information to a monitoring terminal according to the threat level evaluation result.

[0054] In a third aspect of the embodiment of the present application, an electronic device is provided, comprising:

[0055] a processor;

[0056] a memory for storing processor-executable instructions;

[0057] The processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0058] In a fourth aspect of the embodiment of the present application, a computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0059] In the embodiment, by collecting and analyzing electromagnetic signals in a target area, accurate detection and alarm of multi-frequency band mobile phone signals are achieved, different mobile phone communication systems and communication states can be effectively identified, and the accuracy and comprehensiveness of mobile phone signal detection are improved. By using a signal strength difference positioning algorithm and combining uplink and downlink frequency band signal strength ratio analysis, the position of a mobile phone signal source can be calculated in real time and movement trajectory can be recorded, effectively solving the technical problem that traditional methods cannot accurately locate a mobile phone signal source, and significantly improving positioning accuracy and real-time performance. Based on the time sequence relationship between movement trajectory data and communication state, intelligent identification and threat level evaluation of mobile phone use scenarios are achieved, and hierarchical alarm information is automatically sent according to the safety risks of different scenarios, so that the system has higher intelligent level and practicality, and meets the differentiated needs in different security monitoring environments. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 A flowchart of a multi-band mobile phone signal detection and alarm method according to an embodiment of the present application is shown in

[0061] Figure 2 A signal strength ratio distribution and communication state determination diagram is shown in

[0062] Figure 3 A multi-dimensional reliability evaluation and iterative optimization positioning technology performance comparison and analysis diagram is shown in DETAILED DESCRIPTION

[0063] In order to make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in connection with the drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0064] The technical solutions of the present application will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described in some embodiments.

[0065] Figure 1 A flowchart of a multi-band mobile phone signal detection and alarm method according to an embodiment of the present application is shown in Figure 1 The method includes:

[0066] Collecting electromagnetic signals in a target area and pre-processing to obtain pre-processed electromagnetic signals;

[0067] Performing fast Fourier transform on the pre-processed electromagnetic signals to obtain signal frequency spectrum, extracting mobile phone signal characteristic frequency band information from the signal frequency spectrum, the mobile phone signal characteristic frequency band information including uplink frequency band signal strength value and downlink frequency band signal strength value, and determining the mobile phone communication state according to the ratio of the uplink frequency band signal strength value and the downlink frequency band signal strength value;

[0068] Obtaining signal strength data in the target area, calculating real-time position coordinates of the mobile phone signal source using a signal strength difference positioning algorithm based on the mobile phone communication state and the signal strength data, and recording the change of the position coordinates with time to obtain movement trajectory data;

[0069] According to the time sequence relationship between the mobile trajectory data and the mobile phone communication state, the current mobile phone use scene is judged, the mobile phone use scene is evaluated according to the pre-set scene threat level parameter, and the alarm information is sent to the monitoring terminal according to the threat level evaluation result.

[0070] In an optional embodiment, the electromagnetic signals in the target area are collected and preprocessed to obtain the preprocessed electromagnetic signals, including:

[0071] The mobile phone signal full-band electromagnetic signals in the target area are collected, the signal strength of the electromagnetic signals is detected in real time, the signal gain is automatically adjusted according to the signal strength, the amplitude of the electromagnetic signals is maintained within a pre-set dynamic range, and the gain-adjusted signals are obtained.

[0072] The gain-adjusted signals are subjected to multi-layer wavelet decomposition to obtain wavelet coefficients of different scales, the local variance of the wavelet coefficients is calculated, the adaptive threshold is determined according to the local variance, the wavelet coefficients are subjected to soft threshold processing using the adaptive threshold, and the denoised signals are obtained through wavelet reconstruction.

[0073] The denoised signals are divided into a plurality of time windows, the signals in each time window are subjected to Fourier transform to obtain frequency spectra, the frequency spectra are combined to obtain time-frequency spectra, the mean square error of each time-frequency point in the time-frequency spectrum is calculated, the noise power spectrum of each time-frequency point is estimated based on the mean square error, the corresponding Wiener filter coefficients are calculated according to the noise power spectrum, and the denoised signals are subjected to filtering processing using the Wiener filter coefficients to obtain the preprocessed electromagnetic signals.

[0074] In one embodiment, an electromagnetic signal collection device is deployed inside the target area, which is equipped with a wideband receiving antenna and a high-performance digital receiver. The collection device sets the frequency range to 700MHz-2700MHz, covering the main mobile phone communication frequency bands including GSM, CDMA, WCDMA, LTE and 5GNR, etc. The collection device captures the full-band electromagnetic signals in real time through scanning of the target area, the sampling rate is set to 10MHz, and the quantization accuracy is 16 bits to ensure the integrity and accuracy of the collected data.

[0075] Real-time signal strength detection is realized during the collection process, and the signal strength is monitored by calculating the root mean square value of the signal. When the signal strength is detected to be lower than-90dBm, the preamplifier gain is automatically increased, and the gain step is 5dB; when the signal strength exceeds-30dBm, the gain is automatically reduced to prevent signal saturation. In this way, the signal amplitude can be maintained within the dynamic range of [-80dBm, -20dBm], and the gain-adjusted signals are obtained.

[0076] The gain-adjusted signal is subjected to wavelet denoising processing. The db4 wavelet basis is used to decompose the signal into 5 layers, obtaining wavelet coefficients of different scales. For each layer of wavelet coefficients, a sliding window size of 128 sampling points is set, and the window overlap rate is 50%. The local variance of the wavelet coefficients is calculated in each window, and the calculation method is to divide the sum of the squares of the wavelet coefficients in the window by the window length. According to the calculated local variance, the threshold value is determined using an adaptive threshold estimation method based on Bayesian criteria. Experimental data show that, under the condition of a signal-to-noise ratio of about 5 dB, the threshold values of the first to fifth layers are about 0.15, 0.25, 0.42, 0.68 and 1.05, respectively.

[0077] The determined adaptive threshold is used to perform soft thresholding on the wavelet coefficients. The soft thresholding method is: when the absolute value of the wavelet coefficient is less than the threshold value, the coefficient is set to zero; when the absolute value of the wavelet coefficient is greater than or equal to the threshold value, the absolute value of the coefficient is reduced by the threshold value, and the original sign is maintained. After processing, the signal is reconstructed by wavelet inverse transform to obtain the denoised signal. After processing, the signal-to-noise ratio of the signal is improved by about 8 dB, and the background noise is significantly reduced.

[0078] The denoised signal is divided into time windows for time-frequency analysis. The window length is set to 1024 sampling points, and the window overlap rate is 75%. The Hanning window function is used to window the signal in each time window to reduce spectral leakage. The 1024-point fast Fourier transform is performed on the signal in each windowed time window to obtain the frequency spectrum. The frequency spectra of all time windows are arranged in time order to form a time-frequency spectrum, with a resolution of sampling rate / 1024, about 9.77 kHz.

[0079] The mean square error of each time-frequency point in the time-frequency spectrum is calculated to estimate the noise power spectrum. The specific method is: select the region of the time-frequency spectrum where the signal is not active (usually the high-frequency region, such as above 8 MHz) as the pure noise region, and calculate the average value of the power spectrum in this region as the initial noise power estimation value. For each time-frequency point, the power difference with the surrounding 8 time-frequency points is calculated, and if the difference is less than a preset threshold (set to 3 dB), it is considered that the point is a noise point and is included in the noise statistics; otherwise, it is considered that the point contains signal components and is not included in the statistics. After multiple iterations (usually 5 times), a stable noise power spectrum estimation value is obtained.

[0080] The Wiener filter coefficient is calculated according to the estimated noise power spectrum. For each time-frequency point with a frequency of f and a time of t, if the ratio of the signal power to the noise power of the point is greater than a preset threshold (2.5), the point is retained; otherwise, the point is attenuated according to a function relationship of the signal-to-noise ratio. The specific attenuation coefficient calculation method is: the signal power is subtracted from the noise power, and then divided by the signal power to obtain the Wiener filter coefficient of the time-frequency point.

[0081] The calculated Wiener filter coefficient is applied to the time-frequency spectrum of the de-noised signal to realize frequency domain filtering. The filtered time-frequency spectrum is subjected to inverse Fourier transform and window effect removal to reconstruct a time domain signal. Finally, the signals of the time windows are added to obtain a complete pre-processed electromagnetic signal.

[0082] Based on the above technical solution, high-precision and low-distortion pre-processing effect of the electromagnetic signal in the target region can be realized. On the one hand, the automatic gain control stabilizes the amplitude of the collected signal within the dynamic range, effectively coping with the change of signal intensity in different environments and improving the signal quality; on the other hand, the combination of multi-layer wavelet decomposition and local variance adaptive threshold can accurately distinguish the signal and the noise, realize flexible de-noising processing in multiple scales, and retain the key characteristics of the signal. Further combining time-frequency spectrum analysis and Wiener filtering technology, through noise power spectrum estimation and filter coefficient fine adjustment, residual noise can be further suppressed in the frequency domain, the signal-to-noise ratio can be improved, and spectral leakage can be reduced, so that a cleaner, more stable, and more suitable pre-processed electromagnetic signal for subsequent feature extraction and positioning analysis can be obtained.

[0083] In an optional implementation, the pre-processed electromagnetic signal is subjected to fast Fourier transform to obtain a signal spectrum, and extracting mobile phone signal feature frequency band information from the signal spectrum includes:

[0084] The pre-processed electromagnetic signal is subjected to wavelet decomposition by using an orthogonal wavelet basis function to obtain a plurality of frequency band coefficients, energy values of the frequency bands are calculated according to the frequency band coefficients, the energy values are compared with a preset frequency band threshold to determine a target frequency band, the frequency band coefficients of the target frequency band are subjected to phase compensation, and the compensated frequency band coefficients are subjected to Fourier transform to obtain a target spectrum signal.

[0085] The target spectrum signal is divided into an uplink sub-band and a downlink sub-band, a probability density function is constructed for the uplink sub-band and the downlink sub-band respectively, an adaptive threshold is calculated based on the probability density functions to obtain an optimal segmentation point, and the adaptive threshold is used as a signal peak value detection threshold.

[0086] Performing morphological filtering on the spectrum signals of the uplink sub-band and the downlink sub-band, performing peak search on the filtered spectrum signals using the signal peak detection threshold to obtain a signal peak position, calculating a signal bandwidth based on the signal peak position, and setting an integer multiple of the signal bandwidth as an energy concentration interval;

[0087] The variational method is used to optimize the integration interval boundary within the energy concentration interval, and the energy density integration is performed on the spectrum within the optimized integration interval to obtain the uplink frequency band signal strength value and the downlink frequency band signal strength value.

[0088] Exemplarily, the preprocessed electromagnetic signal is first subjected to wavelet decomposition using orthogonal wavelet basis functions. In this embodiment, the DB4 wavelet basis function is selected, and the electromagnetic signal is decomposed into five levels to obtain six frequency band coefficients, each corresponding to a different frequency range. For example, for a signal with a sampling rate of 100 MHz, after decomposition, six frequency bands can be obtained: [0-1.56 MHz], [1.56-3.13 MHz], [3.13-6.25 MHz], [6.25-12.5 MHz], [12.5-25 MHz], and [25-50 MHz]. The energy value is calculated for each frequency band coefficient by summing the squares of the coefficients. For example, the energy of the first frequency band is 10.5, the energy of the second frequency band is 15.2, the energy of the third frequency band is 120.6, the energy of the fourth frequency band is 325.8, the energy of the fifth frequency band is 95.3, and the energy of the sixth frequency band is 12.7. The energy value of each frequency band is compared with a preset frequency band threshold value, which is 50. Frequency bands with energy values ​​greater than the threshold value are determined as target frequency bands, which are the third, fourth, and fifth frequency bands in this example.

[0089] Phase compensation is performed on the target frequency band's coefficients, using a Hilbert transform to eliminate phase distortion. The third frequency band coefficients are multiplied by a compensation factor of 0.85; the fourth frequency band coefficients are multiplied by a compensation factor of 0.92; and the fifth frequency band coefficients are multiplied by a compensation factor of 0.88. A Fast Fourier Transform (FFT) is performed on the compensated frequency band coefficients to obtain the target spectrum signal. For example, after performing an FFT on the compensated third frequency band, a sequence of spectrum values ​​within the frequency range of [3.13-6.25 MHz] is obtained. Similarly, the sequence of spectrum values ​​for the other target frequency bands is obtained, and these sequences are combined to obtain the complete target spectrum signal.

[0090] The target spectrum signal is divided into uplink sub-band and downlink sub-band. In this embodiment, [890-915MHz] is divided into uplink sub-band and [935-960MHz] is divided into downlink sub-band according to the communication standard. The probability density function is constructed for the uplink sub-band and the downlink sub-band respectively, and in the construction process, the distribution of the spectrum amplitude is counted, and the histogram of the spectrum amplitude is generated, taking the amplitude as the abscissa and the normalized frequency as the ordinate. For example, the spectrum amplitude distribution of the uplink sub-band is: [0-0.1] interval accounts for 0.72, [0.1-0.2] interval accounts for 0.15, [0.2-0.3] interval accounts for 0.08, and [0.3 above] interval accounts for 0.05. The optimal segmentation point is calculated based on the probability density function, and the OTSU maximum between-class variance method is used to iteratively calculate the between-class variance, and the amplitude at the maximum variance is selected as the adaptive threshold. In this example, the adaptive threshold of the uplink sub-band is 0.18, and the adaptive threshold of the downlink sub-band is 0.23, which are used as the signal peak detection threshold.

[0091] The spectrum signals of the uplink sub-band and the downlink sub-band are morphologically filtered, and the open operation and the closed operation are combined to remove noise and smooth the spectrum. In the open operation, a structure element with a length of 5 is used for erosion and dilation operation; in the closed operation, a structure element with a length of 7 is used for dilation and erosion operation. The signal peak detection threshold is used to search for the peak value of the filtered spectrum signal, and the local maximum value detection method is used to find the point greater than the threshold and the local maximum in the sliding window. For example, the peak value positions detected in the uplink sub-band are 897MHz, 905MHz and 912MHz; the peak value positions detected in the downlink sub-band are 942MHz, 950MHz and 957MHz.

[0092] The signal bandwidth is calculated according to the signal peak position. For each detected peak value, search for the points on both sides whose amplitude drops to half of the peak amplitude, and the frequency difference between the two points is the signal bandwidth corresponding to the peak value. For example, the bandwidth of the 897MHz peak value is 2.1MHz, the bandwidth of the 905MHz peak value is 1.8MHz, and the bandwidth of the 912MHz peak value is 2.3MHz. Take the average bandwidth 2.07MHz, and set its integer multiple 2MHz as the energy aggregation interval.

[0093] The integral interval boundary is optimized by using the variational method in the energy accumulation interval. For each peak, the initial integral interval is [peak-1 MHz, peak+1 MHz], and the boundary is iteratively optimized by the variational method to maximize the signal-to-noise ratio in the interval. For example, for the 897 MHz peak, the optimized integral interval is [895.8 MHz, 898.3 MHz]. The energy density integration is performed on the spectrum in the optimized integral interval, and the trapezoidal integration method is used, and linear interpolation is used between discrete frequency points. The uplink frequency band signal strength value and the downlink frequency band signal strength value are calculated. For example, the signal strength values of the three peaks in the uplink are 0.76, 0.83 and 0.92, and the total intensity value is 2.51; the signal strength values of the three peaks in the downlink are 0.81, 0.95 and 0.89, and the total intensity value is 2.65.

[0094] Based on the above technical solutions, the characteristic information of the mobile phone signal in different frequency bands can be accurately extracted, and the accuracy and robustness of signal recognition are improved. Through orthogonal wavelet decomposition and frequency band energy analysis, the preliminary identification of the signal frequency band and the positioning of the target frequency band are effectively realized; combined with phase compensation and Fourier transform, the time-frequency focusing of the frequency spectrum can be enhanced, and the frequency spectrum distortion can be reduced. By constructing a probability density function and calculating an adaptive peak detection threshold, dynamic segmentation and adaptive identification of the uplink and downlink signals are realized, and false negatives or false positives caused by fixed thresholds are avoided. Morphological filtering combined with peak search can extract real signal feature points from complex backgrounds, and accurately measure the bandwidth and energy distribution area. Finally, by optimizing the integral interval by the variational method and performing energy density integration, the resolution and stability of the uplink and downlink signal strength estimation are effectively improved, providing reliable data support for communication state judgment and subsequent positioning analysis.

[0095] In an optional implementation, judging the mobile phone communication state according to the ratio of the uplink frequency band signal strength value and the downlink frequency band signal strength value includes:

[0096] The uplink frequency band signal strength value and the downlink frequency band signal strength value are subjected to time domain sliding processing, a decreasing weight coefficient is set according to the order of sampling time, the decreasing weight coefficient is multiplied by the signal strength value at the corresponding time and accumulated to obtain a smoothed uplink signal strength value and a smoothed downlink signal strength value;

[0097] The ratio of the smoothed uplink signal strength value and the smoothed downlink signal strength value is calculated, the fluctuation amplitude and the change rate of the ratio are extracted, a feature matrix is constructed according to the fluctuation amplitude and the change rate, and a main feature component is obtained by singular value decomposition of the feature matrix;

[0098] A decision function is established based on the principal eigenvectors, and the optimal segmentation point of the decision function is calculated by using the maximum inter-class variance method, and the optimal segmentation point is taken as the discrimination threshold of the communication state;

[0099] The value range of the signal strength ratio is divided into a normal communication interval and an abnormal communication interval according to the discrimination threshold, the ratio of the smoothed uplink signal strength value to the smoothed downlink signal strength value is calculated in real time, and when the current ratio falls into the normal communication interval, it is determined as a normal communication state, and when the current ratio falls into the abnormal communication interval, the link fault type is determined according to the relative positional relationship between the current ratio and the normal communication interval, and the corresponding communication state determination result is output.

[0100] In the embodiment, first, the uplink frequency band signal strength value and the downlink frequency band signal strength value of the mobile phone terminal are collected. The uplink frequency band signal strength value represents the strength of the signal sent by the mobile phone terminal to the base station, and the downlink frequency band signal strength value represents the strength of the signal sent by the base station to the mobile phone terminal. These signal strength values can be obtained through the radio frequency signal monitoring module integrated on the mobile phone terminal, and the sampling period can be set to 100 milliseconds to ensure the real-time and accuracy of data collection.

[0101] Figure 2 A signal strength ratio distribution and a communication state determination diagram are shown in FIG. 1. In order to eliminate the influence of instantaneous signal fluctuations on the determination result, the uplink frequency band signal strength value and the downlink frequency band signal strength value are subjected to time domain sliding processing. Specifically, a sliding window with a length of N is set, and N can be 20, that is, the latest 2 seconds of sampling data are retained. According to the order of sampling time, a decreasing weight coefficient is set, the latest sampling point is given the highest weight, and the weight gradually decreases over time. The decreasing weight coefficient can be calculated according to a decay rate of 0.95, that is, the weight of the i-th sampling point is 0.95^(N-i), where i represents the serial number of the sampling point, from 1 to N. The decreasing weight coefficient is multiplied by the signal strength value at the corresponding time and accumulated to obtain the smoothed uplink signal strength value and the smoothed downlink signal strength value.

[0102] Suppose that the original uplink signal strength value sequence collected at a certain time is [-75, -76, -74, -73, -77, -76, -75, -74, -75, -76] dBm, and the original downlink signal strength value sequence is [-65, -66, -64, -65, -67, -66, -65, -63, -64, -65] dBm. After applying the decreasing weight coefficient, the smoothed uplink signal strength value is calculated to be -74.8 dBm, and the smoothed downlink signal strength value is calculated to be -64.7 dBm.

[0103] The ratio of the smoothed uplink signal strength value to the smoothed downlink signal strength value is calculated. In this example, the signal strength ratio is -74.8 / (-64.7) = 1.156. Since signal strength is expressed in negative decibel-milliwatts (dBm), a larger value indicates a weaker signal, so this ratio actually reflects the degree of attenuation of the uplink signal relative to the downlink signal.

[0104] The change in the ratio is monitored over successive time periods, and the fluctuation amplitude and the rate of change of the ratio are extracted. The fluctuation amplitude can be characterized by calculating the standard deviation of the signal strength ratio, and the rate of change can be characterized by calculating the average of the absolute values of the difference between adjacent sample points of the signal strength ratio. With a 5-minute statistical period, the signal strength ratio is recorded every second, and 300 data points are obtained. The calculated fluctuation amplitude is 0.05, and the rate of change is 0.01 / second.

[0105] A feature matrix is constructed based on the fluctuation amplitude and the rate of change. The feature matrix contains the fluctuation amplitude and the rate of change data in multiple time windows. For example, 5 minutes of data are divided into 5 one-minute time windows, and the fluctuation amplitude and the rate of change are calculated for each window to form a 5x2 feature matrix. In this example, the feature matrix is [[0.04, 0.008], [0.05, 0.01], [0.06, 0.012], [0.04, 0.009], [0.05, 0.011]].

[0106] The feature matrix is subjected to singular value decomposition to obtain the main feature components. Singular value decomposition can decompose the feature matrix into the product of three matrices, in which the singular values in the diagonal matrix are arranged in descending order, and the larger the singular value, the more important the corresponding feature component. The feature vector corresponding to the largest singular value is taken as the main feature vector. In this example, the main feature vector obtained by singular value decomposition is [0.7, 0.3], indicating that the weights of the fluctuation amplitude and the rate of change in judging the communication state are 0.7 and 0.3, respectively.

[0107] A decision function is established based on the main feature vector. The decision function is expressed as the weighted sum of the fluctuation amplitude and the rate of change, i.e., decision value = 0.7x fluctuation amplitude + 0.3x rate of change. The optimal split point of the decision function is calculated using the maximum inter-class variance method. This method selects the point that maximizes the inter-class variance as the optimal split point by traversing all possible split points. In this example, the calculated optimal split point is 0.046, which is taken as the discrimination threshold of the communication state.

[0108] The value range of the signal strength ratio is divided into a normal communication interval and an abnormal communication interval according to a discrimination threshold. The normal communication interval is defined as [1.0, 1.2], and the abnormal communication interval is divided into two parts: [0.8, 1.0) below the normal interval and (1.2, 1.4] above the normal interval. The ratio of the smoothed uplink signal strength value to the smoothed downlink signal strength value is calculated in real time. When the current ratio falls within the normal communication interval, it is determined to be a normal communication state. When the current ratio falls within the abnormal communication interval, the type of link failure is determined according to the relative position relationship between the current ratio and the normal communication interval.

[0109] Specifically, when the ratio is less than 1.0, it indicates that the uplink signal is too strong relative to the downlink signal, and there may be an abnormal amplification of the uplink signal, so it is determined to be an uplink abnormality. When the ratio is greater than 1.2, it indicates that the uplink signal is too weak relative to the downlink signal, and there may be an abnormal amplification of the downlink signal or suppression of the uplink signal, so it is determined to be a downlink abnormality.

[0110] In order to improve the accuracy of the judgment, a continuous judgment mechanism is set. Only when the judgment results of 5 consecutive samples are consistent, the final communication state judgment result is output. This can effectively avoid misjudgment caused by signal instantaneous fluctuation.

[0111] In addition, considering the difference in signal strength ratio under different communication environments, the present embodiment also provides an adaptive threshold adjustment mechanism. The system will record the signal strength ratio of the user in the normal communication state at different positions, and establish a mapping relationship between the position and the signal strength ratio. When the user's position changes, the system will automatically adjust the range of the normal communication interval according to the current position. The position information can be obtained by GPS or base station positioning, and the accuracy is set to 100 meters.

[0112] In actual application, the system will issue an alarm prompt to the user when detecting an abnormal communication state. The alarm methods include screen prompt, sound prompt and vibration prompt. The user can take appropriate measures according to the alarm prompt, such as switching network mode, changing position or dialing customer service phone.

[0113] In the prior art, the judgment of the mobile phone communication state mostly depends on the signal strength judgment at a single point in time, usually adopts a fixed threshold or a static ratio analysis method, and it is difficult to accurately reflect the dynamic change of the communication state, and it is sensitive to signal fluctuations and interference, resulting in unstable judgment results and low accuracy. The application introduces a sliding weighting mechanism to perform time domain smoothing processing on the uplink and downlink signal strength values, which can weaken the interference of short-time mutations on the judgment results, extract more representative smoothed signal features, and improve the time sequence continuity and robustness of signal processing. On this basis, a feature matrix is constructed by the fluctuation amplitude and the change rate of the ratio, and the principal component is extracted by singular value decomposition, further eliminating redundant information and noise interference, so that the feature recognition of the communication state is more accurate. At the same time, the maximum inter-class variance method is used to construct a decision function and determine the optimal segmentation point, and the discrimination threshold of the communication state is adaptively generated, breaking through the limitations of the artificial setting threshold in the traditional method. Through the above technical improvements, not only the dynamic recognition of normal communication and abnormal communication state is realized, but also the link fault type can be further judged, and the state judgment accuracy and stability of the system in complex environment are improved, providing a more reliable basis for subsequent positioning and alarm.

[0114] In an optional implementation, the signal strength data in the target area is acquired, the real-time position coordinates of the mobile phone signal source are calculated by using a signal strength difference positioning algorithm based on the mobile phone communication state and the signal strength data, and the movement trajectory data is obtained by recording the change of the position coordinates with time, comprising:

[0115] A plurality of signal collection nodes are arranged in the target area to collect signal strength data in multiple directions;

[0116] The ratio of the short-time standard deviation to the long-time standard deviation of the signal strength data is calculated according to the pre-acquired mobile phone communication state, and the effective signal detector is screened according to the ratio;

[0117] The signal strength change trend collected by the effective signal detector is calculated, the smoothing window length is determined according to the change trend, the spatial weight is determined based on the spatial distribution position of the signal detector, and the spatial weight is weighted and smoothed with the signal strength data to obtain the smoothed signal strength value;

[0118] Any two effective signal detectors are selected to form a detector pair, the difference value of the smoothed signal strength values between the detector pair is calculated, the path loss index is dynamically adjusted according to the mobile phone communication state, and the distance difference value between the detector pair is calculated;

[0119] The signal correlation and the stability of the signal strength difference of the detector pair are calculated to obtain a credibility score, a hyperbolic equation is constructed according to the distance difference value, the credibility score is taken as the initial weight of the hyperbolic equation, and the hyperbolic equation set is solved by iteration to obtain the real-time position of the signal source.

[0120] According to the real-time position of the signal source within the continuous time window, the moving speed and direction are calculated, and the abnormal position is corrected by combining the mobile phone communication state, to obtain a smooth moving trajectory.

[0121] Exemplarily, first, a plurality of signal collection nodes are arranged in the target area, each node is equipped with a directional antenna array, and signal strength data in different directions can be collected simultaneously. The signal collection nodes are usually arranged at the edges or high points of the target area to form a network covering the entire area. For example, in a 100m x 100m indoor scene, one signal collection node can be arranged at each of the four corners and the center, and each node is equipped with 4 directional antennas pointing in different directions. The signal collection nodes transmit the collected data to the central processing server in real time through wired or wireless means.

[0122] The mobile phone communication state is analyzed, including the call state, data transmission state, signal type, etc. For the signal strength data collected by each signal collection node, the standard deviation in a short time window (such as 1 second) and the standard deviation in a long time window (such as 10 seconds) are calculated, and then the ratio of the two is calculated. When the ratio is greater than a preset threshold (such as 0.8), it indicates that the detector can effectively capture signal changes, and it is marked as an effective signal detector. For example, for detector A, if the short-time standard deviation is 2.3 dBm and the long-time standard deviation is 2.5 dBm, the ratio is 0.92, which is greater than the threshold 0.8, and it is determined as an effective detector.

[0123] For each effective signal detector, the signal strength trend collected is analyzed. The first-order difference of signal strength at consecutive time points is calculated to judge the degree of signal change. If the signal change is gentle (such as the absolute value of the difference is less than 1 dBm), a longer smoothing window (such as 5 seconds) is used; if the signal change is severe (such as the absolute value of the difference is greater than 3 dBm), a shorter smoothing window (such as 1 second) is used. At the same time, based on the spatial distribution position of the signal detector, the spatial weight is calculated, and the closer the detector is to the center of the target area, the higher the weight. For example, the weight of the detector located at the center can be set to 1.0, and the weight of the detector located at the edge can be set to 0.7. The spatial weight and signal strength data are weighted and averaged to obtain the smoothed signal strength value. For example, the original signal strength of a certain detector at 5 consecutive time points is -65 dBm, -67 dBm, -63 dBm, -64 dBm, and -66 dBm, and the spatial weight is 0.9, then the smoothed signal strength is -65.0 dBm.

[0124] Select any two from the effective signal detectors to form a detector pair, and a total of n x (n-1) / 2 detector pairs are formed. Calculate the smoothed signal strength difference between each detector pair. Dynamically adjust the path loss index according to the communication state of the mobile phone, such as setting the path loss index to 3.5 when the mobile phone is in a call state, 3.2 when in a data transmission state, and 3.0 when in a standby state. Based on the signal strength difference and the path loss index, calculate the distance difference between the two detectors to the signal source. For example, the smoothed signal strengths of detectors A and B are -65 dBm and -70 dBm, respectively, with a difference of 5 dBm, and under the condition of a path loss index of 3.2, the distance difference can be calculated to be about 8.2 meters.

[0125] Calculate the signal correlation and signal strength difference stability for each detector pair to obtain a credibility score. The signal correlation is obtained by calculating the correlation coefficient of the signal strength changes of the two detectors, and the stability is obtained by calculating the standard deviation of the difference. For example, the signal correlation coefficient of detector pair A-B is 0.85, and the difference standard deviation is 0.6 dBm, so the credibility score can be calculated as 0.82. Based on the distance difference of each detector pair, construct a hyperbolic equation, and use the credibility score as the initial weight of the equation. Use the weighted least squares method to iteratively solve the intersection of all hyperbolic equation sets to obtain the real-time position coordinates of the signal source. For example, in a certain calculation, the system selects 4 detector pairs, and through 5 iterations of calculation, the final signal source position coordinates are determined to be (45.3 meters, 67.8 meters).

[0126] Record the signal source position coordinates in a continuous time window (such as 5 seconds), calculate the distance and direction angle between adjacent positions, and obtain the moving speed and direction. If the calculated speed exceeds a reasonable range (such as more than 2 meters / second in an indoor scene), or the direction change is too large (such as more than 90 degrees between adjacent positions), then perform constraint correction according to the communication state of the mobile phone and the historical trajectory. For example, if the mobile phone is in a stationary call state, then the abnormal position is moved closer to the historical stable position; if the mobile phone is in a moving state, then perform smoothing according to the historical moving trend. The final moving trajectory data contains information such as time stamp, position coordinates, speed and direction, which can be used for subsequent behavior analysis or location services.

[0127] In this embodiment, by introducing multi-node signal acquisition and dynamic path loss modeling, high-precision positioning and trajectory tracking of mobile phone signal source location are realized. By fusing communication state information and dynamically adjusting path loss index, the positioning adaptability and accuracy in complex electromagnetic environment are improved. By screening effective detectors through short-term and long-term standard deviation ratio, the robustness of the system to abnormal data and interference sources is enhanced. At the same time, by introducing spatial weight and smoothing mechanism, the influence of local fluctuations on the positioning result is suppressed, and the continuity and spatial consistency of signal strength data are improved. In terms of positioning algorithm, the iterative solution based on hyperbolic equation set is adopted, and the reliability score is set as the weight combining signal correlation and difference stability, which strengthens the accuracy control and error suppression ability in multi-node fusion. Finally, by analyzing the speed and direction of the position information in the continuous time window, and combining the communication state to realize the dynamic constraint correction of the sudden trajectory, the smoothness and actual reflection ability of the trajectory output are further improved. Overall, this scheme realizes the goal of extracting reliable position data from unstable and multi-interference signals, and provides accurate and continuous trajectory support for subsequent behavior recognition and alarm decision.

[0128] In an alternative embodiment, the signal correlation and signal strength difference stability of the detector pair are calculated to obtain a reliability score, a hyperbolic equation is constructed according to the distance difference value, the reliability score is used as the initial weight of the hyperbolic equation, and the hyperbolic equation set is solved iteratively to obtain the real-time position of the signal source, including:

[0129] The signal strength sequence of the detector pair is obtained, the cross-correlation coefficient of the signal strength sequence and the standard deviation of the signal strength difference sequence are calculated, and the reciprocal of the cross-correlation coefficient and the standard deviation is adaptively weighted and combined to obtain the initial reliability score of the detector pair.

[0130] The direction angle of the line between the detector pair is calculated, the spatial distribution weight is determined according to the distribution of the direction angle, and the modified reliability score of the detector pair is obtained by multiplying the spatial distribution weight and the initial reliability score.

[0131] The distance difference value of the detector pair is weighted using the modified reliability score, a hyperbolic equation set is constructed according to the weighted distance difference value, and a weighted least squares method with constraints is used to solve the hyperbolic equation set to obtain an initial position estimate.

[0132] The position of the signal source at the previous time is obtained, the position constraint region at the current time is determined combining the maximum moving speed constraint, and the adaptive step iterative search is performed in the position constraint region with the initial position estimate as the starting point to obtain the current iterative position.

[0133] The theoretical distance difference of the detector pair is calculated based on the current iteration position, a deviation between the theoretical distance difference and a measured distance difference is taken as a positioning residual, a real-time confidence score is obtained by dynamically attenuating and updating a correction confidence score according to the positioning residual, and the weight of the hyperbolic equation set is updated by using the real-time confidence score, and the iteration is continued until the position estimation value converges to obtain the real-time position of the signal source.

[0134] In the implementation of the present application, first, a plurality of signal detectors are deployed to form a detection network, each detector can detect multiple frequency bands of mobile phone signals at the same time. These detectors are distributed at different locations, and their coordinate positions are known and fixed. Each detector is equipped with a multi-band receiving antenna and can receive and record mobile phone signals of different systems such as GSM, CDMA, WCDMA, LTE, and 5G. The detector collects signal strength data according to a preset sampling period, and the sampling period is set to 100 milliseconds to meet the real-time requirement.

[0135] The detectors are paired two by two to form detector pairs. For example, in a system consisting of 4 detectors, 6 detector pairs can be formed: (1, 2), (1, 3), (1, 4), (2, 3), (2, 4), (3, 4). Each pair of detectors receives signals from the same signal source and records the time of arrival and signal strength. Since the speed of electromagnetic wave propagation is the speed of light, the distance difference between the signal source and the two detectors can be calculated according to the time difference of the signal arrival at the two detectors. For each pair of detectors, a signal strength sequence within a time window is obtained. The length of the time window is set to 5 seconds, and 50 signal strength samples are collected by each detector during this period. The cross-correlation coefficient of the two signal strength sequences is calculated, which reflects the consistency of the change trend of the two signal sequences. The cross-correlation coefficient ranges from -1 to 1, and the closer the value is to 1, the more consistent the change trend of the two sequences, and the higher the data quality. In practical applications, for good quality signals, the cross-correlation coefficient is usually greater than 0.8.

[0136] The standard deviation of the signal strength difference sequence of the detector pair is calculated. The signal strength difference sequence refers to the sequence of the difference between the signal strengths measured by the two detectors at the same time. The standard deviation reflects the fluctuation degree of the difference sequence, and the smaller the standard deviation, the more stable the difference and the higher the positioning reliability. The initial confidence score of the detector pair is obtained by adaptively weighting and combining the cross-correlation coefficient and the reciprocal of the standard deviation. The weight of adaptive weighting is dynamically adjusted according to the actual signal environment. In an environment with large signal fluctuations, the weight of the cross-correlation coefficient is set to 0.7, and the weight of the reciprocal of the standard deviation is set to 0.3; in a relatively stable signal environment, the weights of both are set to 0.5.

[0137] Suppose the signal strength sequences of detector 1 and detector 2 are [-60, -62, -61, -63, -60] dBm and [-65, -67, -66, -68, -65] dBm respectively, the calculated cross-correlation coefficient is 0.95, the signal strength difference sequence is [5, 5, 5, 5, 5] dBm, the standard deviation is 0, and the reciprocal of the standard deviation is a very large value. In order to calculate the stability, it is limited to 10. Take 10. In the environment with large signal fluctuation, the initial credibility score is 0.7*0.95+0.3*10=3.665.

[0138] The direction angle of the connection between the detector pairs is calculated. The direction angle refers to the angle between the detector pair and the reference direction (such as the north direction). In order to ensure the uniformity of the spatial coverage, the distribution of the direction angle of the detector pair should be as uniform as possible. According to the distribution of the direction angle, the spatial distribution weight is determined. The calculation method of the spatial distribution weight is: divide the 360-degree space into 8 sectors, count the number of detector pairs in each sector, and give higher weight to the detector pairs in the sector with less number; give lower weight to the detector pairs in the sector with more number. In this way, the positioning accuracy in different directions can be balanced. For example, if there is 1 pair of detectors in the 0-45 degree sector and 3 pairs of detectors in the 45-90 degree sector, the spatial distribution weight of the detector pairs in the 0-45 degree sector is 1.5, and the spatial distribution weight of the detector pairs in the 45-90 degree sector is 0.5. Multiply the spatial distribution weight by the initial credibility score to obtain the modified credibility score of the detector pair. Suppose the detector pair (1, 2) is located in the 0-45 degree sector, and the spatial distribution weight is 1.5, then the modified credibility score is 3.665*1.5=5.4975.

[0139] The distance difference of the detector pair is weighted using the modified credibility score. The distance difference refers to the difference between the distances of the signal source to the two detectors, which can be obtained by multiplying the time difference of the signal arrival by the speed of light. According to the weighted distance difference, a hyperbolic equation set is constructed. In a two-dimensional plane, each pair of detector pairs determines a hyperbola, and the signal source is located on the hyperbola; multiple pairs of detector pairs determine multiple hyperbolas, and the signal source is located near the intersection of these hyperbolas.

[0140] The weighted least squares method with constraints is used to solve the hyperbolic equation set to obtain the initial position estimate. In the solving process, considering the influence of measurement error and noise, multiple hyperbolas may not exactly intersect at a point. The goal of the weighted least squares method is to find a point that minimizes the weighted distance square sum of each hyperbola. The weight is the modified credibility score, and the detector pair with high credibility has greater influence in the positioning calculation.

[0141] The position constraint region of the current time is determined by combining the maximum moving speed constraint and the position of the signal source at the previous time. Assuming that the maximum moving speed of the mobile phone user is 10 meters per second, the position at the previous time is (100, 200) meters, and the sampling interval is 0.1 second, the position constraint region of the current time is a circular region with (100, 200) as the center and a radius of 1 meter. The adaptive step iterative search is started from the initial position estimate value in the position constraint region to obtain the current iteration position.

[0142] The specific method of adaptive step iterative search is as follows: set the initial step size to 0.5 meters, move one step in the gradient direction, and calculate the target function value (i.e. the weighted residual sum of squares) of the new position. If the target function value decreases, accept this step of movement and increase the step size (e.g. multiply by 1.2); if the target function value increases, reject this step of movement, decrease the step size (e.g. multiply by 0.5), and try a new direction. Stop the iteration when the step size is less than a preset threshold (e.g. 0.01 meters) or the number of iterations reaches an upper limit (e.g. 50 times).

[0143] The theoretical distance difference value of the detector pair is calculated based on the current iteration position. The theoretical distance difference value refers to the difference between the calculated distances of the signal source to the two detectors assuming that the signal source is located at the current iteration position. The deviation of the theoretical distance difference value from the measured distance difference value is used as the positioning residual. For example, the measured distance difference value of detector pair (1, 2) is 10 meters, and the theoretical distance difference value is 9.8 meters, so the positioning residual is 0.2 meters.

[0144] The real-time confidence score is updated by dynamically attenuating the correction confidence score according to the positioning residual. The specific method is as follows: set the residual threshold to 1 meter, keep the confidence score unchanged when the positioning residual is less than the threshold, and decrease the confidence score according to the exponential attenuation law when the positioning residual is greater than the threshold, and the attenuation coefficient is proportional to the residual size. For example, if the positioning residual is 2 meters, which exceeds the threshold of 1 meter, and the attenuation coefficient is set to 0.8, the real-time confidence score is the correction confidence score multiplied by 0.8 raised to the power of 1, i.e. 5.4975 x 0.8 = 4.398.

[0145] The weight of the hyperbolic equation set is updated using the real-time confidence score, and the iteration is continued until the position estimate value converges to obtain the real-time position of the signal source. The convergence condition is that the position change of two consecutive iterations is less than 0.05 meters, or the number of iterations reaches 20 times. The final position obtained is the real-time position of the signal source.

[0146] Figure 3For multi-dimensional reliability evaluation and iterative optimization positioning technology performance comparison analysis schematic diagram, in the positioning accuracy aspect, the present application realizes 1.0 meter high-precision positioning, compared with 4.0 meters of traditional method and 3.0 meters of basic reliability score method, has obvious promotion, the accuracy increases by 75%. On the convergence time, the present application only needs 2.0 seconds to complete position calculation, compared with traditional method, saves 50% time, embodies the great improvement of algorithm efficiency.

[0147] In the anti-interference ability score, the present application reaches 8.0 points, far higher than 3.0 points of traditional method and 5.0 points of basic reliability score method, indicating that the method has excellent anti-noise performance in complex electromagnetic environment. System robustness test shows that the present application scores 7.0 points, 5.0 points higher than traditional method, proving the stability of the algorithm in various scenarios.

[0148] These performance improvements are mainly due to multi-dimensional reliability evaluation mechanism, spatial distribution weight adjustment, position constraint region limitation and real-time positioning residual feedback and other technical innovations, effectively solve the problem of unstable precision and poor anti-interference ability of traditional positioning method in complex environment.

[0149] In the embodiment, by introducing multi-dimensional reliability evaluation and iterative optimization mechanism, high-precision estimation of signal source position in complex environment is realized. Traditional positioning method often uses fixed weight or does not consider the weighted method of signal dynamic characteristics, which is easily affected by signal noise, path obstruction and uneven detector layout, resulting in unstable positioning accuracy. The present application first uses the standard deviation of cross-correlation coefficient and signal strength difference to construct initial reliability score, which considers signal synchronization and suppresses the uncertainty caused by violent fluctuation, improves the rationality of score. Further introduce the directional angle distribution between detectors, construct spatial distribution weight, effectively alleviate the solution deviation problem of traditional hyperbolic positioning when the geometric distribution is uneven. Combined with weighted least squares method and position constraint region limitation, the search range is controlled by using signal source historical position and speed boundary, which enhances the continuity and physical rationality of positioning result. In the iterative process, introduce real-time positioning residual feedback mechanism, dynamically update the reliability score, make the weight more fit the current channel state, improve the convergence speed and solution accuracy. Overall, the present application significantly enhances the robustness and adaptive ability of signal positioning system through multi-layer weighting and feedback optimization, improves the real-time positioning accuracy in complex scenes.

[0150] In an optional embodiment, according to the time sequence relationship between the mobile trajectory data and the mobile phone communication state, the current mobile phone use scene is judged, the mobile phone use scene is evaluated according to the pre-set scene threat level parameter, and the alarm information is sent to the monitoring terminal according to the threat level evaluation result, including:

[0151] acquire mobile trajectory data and communication state data of the mobile phone, extract motion features and spatial distribution features in the mobile trajectory data, extract communication behavior features and signal quality features in the communication state data, and generate time sequence feature data;

[0152] perform time window segmentation on the time sequence feature data, calculate correlation degrees between features in each time window, construct a feature correlation matrix, and match the feature correlation matrix with a preset mobile phone use scenario model to obtain a current mobile phone use scenario;

[0153] perform comparative analysis on the time sequence feature data and preset normal scenario features corresponding to the current mobile phone use scenario, generate a feature deviation sequence, perform time sequence clustering on the feature deviation sequence to obtain an abnormal event cluster, and generate abnormal event features based on a time span and a number of occurrences of the abnormal event cluster;

[0154] calculate a threat score according to the abnormal event features, combine a preset threat level parameter corresponding to the current mobile phone use scenario for weighting, obtain a real-time threat level, and when the real-time threat level exceeds an early warning threshold, trigger an abnormal event feature of the early warning as a monitoring template, count a number of matched abnormal events in an observation period, and when the number of matched abnormal events exceeds a set threshold and the threat level continues to rise, send an alarm information to a monitoring terminal.

[0155] Exemplarily, first, mobile trajectory data and communication state data of the mobile phone are acquired. The mobile trajectory data includes GPS position coordinates, acceleration sensor data, gyroscope data, etc., and motion features and spatial distribution features can be extracted through the data. The motion features include average speed, speed change rate, acceleration fluctuation, and travel direction change frequency, etc.; the spatial distribution features include position concentration, stay point density, activity radius, and conventional path deviation degree, etc. The communication state data includes call records, short message records, network connection states, signal strengths, etc., and communication behavior features and signal quality features can be extracted through the data. The communication behavior features include call frequency, call duration distribution, short message sending frequency, and traffic usage mode, etc.; the signal quality features include signal strength fluctuation, network switching frequency, and connection stability, etc. For example, the mobile trajectory data of a user shows that the average speed is 5.2 km / h, and the position is mainly concentrated in three areas; the communication state data shows that the call frequency on weekdays is 8 times per day, the average call duration is 3.5 minutes, and the signal strength fluctuates between -85 dBm and -70 dBm.

[0156] The time window segmentation is performed on the time series feature data, and time windows with a length varying from 5 minutes to 60 minutes are adopted, and the window size is dynamically adjusted according to different scene characteristics. The correlation degree between features in each time window is calculated, and a feature correlation matrix is constructed. The correlation degree calculation considers the consistency of the trend of feature values, the time lag of the change of feature values, and the mutual influence degree of feature values. For example, by calculating the consistency of the trend of the change of the moving speed and the call frequency in the same time window, a correlation coefficient of 0.78 is obtained, indicating a strong positive correlation. The constructed feature correlation matrix contains the correlation coefficients between all extracted features, forming an n x n matrix (n is the number of features). The feature correlation matrix is matched with a preset mobile phone use scene model, and the preset scene model includes an office scene, a home scene, a traffic travel scene, a social scene, etc., and each scene has a specific feature correlation pattern. By calculating the similarity of the current feature correlation matrix and each preset scene model, the scene with the highest similarity is selected as the current mobile phone use scene. For example, a certain feature correlation matrix has a similarity of 0.85 with the office scene model and a similarity of 0.62 with the home scene model, and it is determined that the current scene is the office scene.

[0157] The time series feature data is compared and analyzed with the preset normal scene features corresponding to the current mobile phone use scene. The preset normal scene features are the normal range and change pattern of each feature in this scene based on historical data statistics. By calculating the deviation of the current feature from the normal feature, a feature deviation sequence is generated. For example, in the office scene, if the current location concentration is 0.35 and the normal range is 0.75-0.95, the location concentration deviation is -0.5. The time series clustering is performed on the feature deviation sequence, and the density clustering algorithm is adopted to cluster the data points with similar feature deviation patterns and close in time into a class, forming an abnormal event cluster. The abnormal event features are generated based on the time span and occurrence frequency of the abnormal event cluster. For example, the location concentration is detected to be continuously lower than the normal value by more than 50% for 3 hours, and the call frequency increases by 200% and the network switching frequency increases by 150% during this period, forming an abnormal event cluster with a time span of 3 hours and an occurrence frequency of 1 time.

[0158] The threat score is calculated according to the abnormal event characteristics, and the score considers factors such as abnormal degree, duration, and impact range. The real-time threat level is obtained by combining the preset threat level parameters corresponding to the current mobile phone use scenario. The preset threat level parameters are set according to the sensitivity of safety in different scenarios, for example, the threat level parameter of the office scenario is 1.5, the home scenario is 1.2, and the traffic scenario is 1.8. When the real-time threat level exceeds the warning threshold (for example, set to 75 points), the abnormal event characteristics that trigger the warning are used as the monitoring template, and the number of matching abnormal events is counted within the observation period (such as 24 hours). When the number of matches exceeds the set threshold (such as 3 times) and the threat level continues to rise, an alarm message is sent to the monitoring terminal. The alarm message includes abnormal event characteristics description, threat level score, abnormal duration, and affected function modules. For example, a user is detected in an office scenario with location abnormal dispersion, abnormal call mode, and frequent network switching, and the threat score is calculated to be 65 points. After combining the office scenario threat level parameter 1.5, the real-time threat level is 97.5 points, which exceeds the warning threshold of 75 points. Within the 24-hour observation period, 4 similar abnormal events are detected, and the threat level rises from 97.5 points to 105 points. The system sends an alarm message to the monitoring terminal.

[0159] The above technical solution fuses the time sequence characteristics of mobile phone movement trajectory and communication state, realizes intelligent identification of mobile phone use scenario and dynamic evaluation of threat level. In the prior art, only static position or signal strength is often relied on for abnormal detection, which is difficult to capture the complete evolution process of user behavior and is prone to false positives or false negatives. The present solution extracts multi-dimensional features such as motion, space, communication, and signal quality, and constructs a feature correlation matrix, which can accurately identify the real scenario of mobile phone use from the overall time sequence evolution. Combined with scenario model matching and deviation analysis mechanism, it can effectively identify abnormal events deviating from normal behavior patterns, and construct event clusters through time sequence clustering means, improving the sensitivity to continuous abnormal behavior. In addition, by introducing dynamic threat scoring and weighting of preset threat level parameters, the threat assessment is targeted and flexible, avoiding errors caused by fixed rule judgment. Finally, combined with the warning threshold and matching statistics mechanism within the observation period, the effective tracking and real-time alarm of persistent threats are realized. Overall, the present solution improves the identification ability of potential risks in complex behavior patterns, enhances the scene adaptability and safety response efficiency of the system.

[0160] In a second aspect of the embodiments of the present application, a multi-band mobile phone signal detection and alarm system is provided, which comprises:

[0161] A first unit is configured to collect electromagnetic signals in a target area and pre-process the electromagnetic signals to obtain pre-processed electromagnetic signals.

[0162] The second unit is used for performing fast Fourier transform on the preprocessed electromagnetic signal to obtain a signal spectrum, extracting a mobile phone signal characteristic frequency band information from the signal spectrum, the mobile phone signal characteristic frequency band information including an uplink frequency band signal strength value and a downlink frequency band signal strength value, and judging the mobile phone communication state according to a ratio of the uplink frequency band signal strength value and the downlink frequency band signal strength value.

[0163] The third unit is used for obtaining signal strength data in a target area, calculating real-time position coordinates of a mobile phone signal source by using a signal strength difference positioning algorithm based on the mobile phone communication state and the signal strength data, and recording changes of the position coordinates with time to obtain movement trajectory data.

[0164] The fourth unit is used for judging a current mobile phone use scene according to a time sequence relationship between the movement trajectory data and the mobile phone communication state, performing threat level evaluation on the mobile phone use scene according to a pre-set scene threat level parameter, and sending alarm information to a monitoring terminal according to a threat level evaluation result.

[0165] In a third aspect, an electronic device is provided, including:

[0166] a processor;

[0167] a memory for storing processor-executable instructions;

[0168] The processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0169] In a fourth aspect, a computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0170] The present application can be a method, device, system and / or computer program product. The computer program product can include a computer readable storage medium having computer readable program instructions stored therein, the computer readable program instructions being executable by a processor to perform various aspects of the present application.

[0171] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for multi-band mobile phone signal detection and alarm, characterized in that, The method comprises the following steps: Collecting electromagnetic signals in a target area and pre-processing to obtain pre-processed electromagnetic signals; Performing fast Fourier transform on the pre-processed electromagnetic signals to obtain signal spectrum, extracting mobile phone signal characteristic frequency band information from the signal spectrum, the mobile phone signal characteristic frequency band information including uplink frequency band signal strength value and downlink frequency band signal strength value, and judging the mobile phone communication state according to the ratio of the uplink frequency band signal strength value and the downlink frequency band signal strength value; Obtaining signal strength data in the target area, calculating the real-time position coordinates of the mobile phone signal source based on the mobile phone communication state and the signal strength data by using a signal strength difference positioning algorithm, and recording the change of the position coordinates with time to obtain mobile trajectory data; According to the time sequence relationship between the mobile trajectory data and the mobile phone communication state, judging the current mobile phone use scene, evaluating the threat level of the mobile phone use scene according to the pre-set scene threat level parameter, and sending alarm information to the monitoring terminal according to the threat level evaluation result.

2. The method of claim 1, wherein, Collecting electromagnetic signals in a target area and pre-processing to obtain pre-processed electromagnetic signals comprises: Collecting mobile phone signal full-band electromagnetic signals in the target area, detecting the signal strength of the electromagnetic signals in real time, automatically adjusting the signal gain according to the signal strength, maintaining the amplitude of the electromagnetic signals within a pre-set dynamic range, and obtaining gain-adjusted signals; Performing multi-layer wavelet decomposition on the gain-adjusted signals to obtain wavelet coefficients of different scales, calculating the local variance of the wavelet coefficients, determining an adaptive threshold value according to the local variance, performing soft threshold processing on the wavelet coefficients using the adaptive threshold value, and obtaining denoised signals through wavelet reconstruction; Dividing the denoised signals into a plurality of time windows, performing Fourier transform on the signals in each time window to obtain frequency spectrum, combining the frequency spectrum to obtain time-frequency spectrum, calculating the mean square error of each time-frequency point in the time-frequency spectrum, estimating the noise power spectrum of each time-frequency point based on the mean square error, calculating the corresponding Wiener filter coefficients according to the noise power spectrum, and performing filtering processing on the denoised signals using the Wiener filter coefficients to obtain pre-processed electromagnetic signals.

3. The method according to claim 1, characterized in that Performing fast Fourier transform on the pre-processed electromagnetic signals to obtain signal spectrum, extracting mobile phone signal characteristic frequency band information from the signal spectrum comprises: Performing wavelet decomposition on the pre-processed electromagnetic signals using an orthogonal wavelet basis function to obtain a plurality of frequency band coefficients, calculating the energy values of the frequency bands according to the frequency band coefficients, comparing the energy values with a pre-set frequency band threshold to determine a target frequency band, performing phase compensation on the frequency band coefficients of the target frequency band, and performing Fourier transform on the compensated frequency band coefficients to obtain a target frequency spectrum signal; Dividing the target frequency spectrum signal into uplink sub-bands and downlink sub-bands, constructing probability density functions for the uplink sub-bands and the downlink sub-bands respectively, calculating the optimal segmentation points based on the probability density functions to obtain adaptive threshold values as signal peak detection threshold values; The spectrum signals of the uplink sub-band and the downlink sub-band are morphologically filtered, the filtered spectrum signals are searched for signal peaks using the signal peak detection threshold, the signal peak positions are obtained, the signal bandwidth is calculated according to the signal peak positions, and an integer multiple of the signal bandwidth is set as an energy aggregation interval; The integral interval boundary in the energy aggregation interval is optimized using a variational method, and the energy density of the spectrum in the optimized integral interval is integrated to obtain the uplink frequency band signal strength value and the downlink frequency band signal strength value.

4. The method of claim 1, wherein, The ratio of the uplink frequency band signal strength value to the downlink frequency band signal strength value is used to determine the mobile phone communication state, including: The uplink frequency band signal strength value and the downlink frequency band signal strength value are processed in the time domain, a decreasing weight coefficient is set according to the order of sampling time, the decreasing weight coefficient is multiplied by the signal strength value at the corresponding time and accumulated to obtain the smoothed uplink signal strength value and the smoothed downlink signal strength value; The ratio of the smoothed uplink signal strength value to the smoothed downlink signal strength value is calculated, the fluctuation amplitude and the change rate of the ratio are extracted, a feature matrix is constructed according to the fluctuation amplitude and the change rate, and the main feature components are obtained by singular value decomposition of the feature matrix; A decision function is established based on the main feature vector, the optimal segmentation point of the decision function is calculated using the maximum inter-class variance method, and the optimal segmentation point is used as the discrimination threshold of the communication state; The value range of the signal strength ratio is divided into a normal communication interval and an abnormal communication interval according to the discrimination threshold, the ratio of the smoothed uplink signal strength value to the smoothed downlink signal strength value is calculated in real time, and when the current ratio falls within the normal communication interval, it is determined as a normal communication state, and when the current ratio falls within the abnormal communication interval, the link fault type is determined according to the relative position relationship between the current ratio and the normal communication interval, and the corresponding communication state determination result is output.

5. The method of claim 1, wherein, Signal strength data in a target area is obtained, and based on the mobile phone communication state and the signal strength data, a signal strength difference positioning algorithm is used to calculate the real-time position coordinates of the mobile phone signal source, and the change of the position coordinates with time is recorded to obtain the movement trajectory data, including: A plurality of signal collection nodes are set in the target area to collect signal strength data in multiple directions; The ratio of the short-time standard deviation to the long-time standard deviation of the signal strength data is calculated according to the previously obtained mobile phone communication state, and the effective signal detectors are screened according to the ratio; The signal strength trend collected by the effective signal detectors is calculated, the smoothing window length is determined according to the trend, the spatial weight is determined based on the spatial distribution position of the signal detectors, and the spatial weight is weighted and smoothed with the signal strength data to obtain the smoothed signal strength value; Any two effective signal detectors are selected to form a detector pair, the difference between the smoothed signal strength values of the detector pair is calculated, the path loss index is dynamically adjusted according to the mobile phone communication state, and the distance difference between the detector pair is calculated. The stability of the signal correlation and the signal intensity difference of the detector pair is calculated to obtain a credibility score, a hyperbolic equation is constructed according to the distance difference, the credibility score is taken as an initial weight of the hyperbolic equation, and the hyperbolic equation set is solved in an iterative manner to obtain a real-time position of the signal source; The moving speed and direction are calculated according to the real-time position of the signal source in a continuous time window, and the abnormal position is corrected by combining the mobile phone communication state to obtain a smooth moving track.

6. The method of claim 5, wherein, The stability of the signal correlation and the signal intensity difference of the detector pair is calculated to obtain a credibility score, a hyperbolic equation is constructed according to the distance difference, the credibility score is taken as an initial weight of the hyperbolic equation, and the hyperbolic equation set is solved in an iterative manner to obtain a real-time position of the signal source, including: A signal intensity sequence of the detector pair is obtained, the cross-correlation coefficient of the signal intensity sequence and the standard deviation of the signal intensity difference sequence are calculated, and the cross-correlation coefficient and the reciprocal of the standard deviation are adaptively weighted and combined to obtain an initial credibility score of the detector pair; The direction included angle of the connecting line between the detectors is calculated, the spatial distribution weight is determined according to the distribution of the direction included angle, and the spatial distribution weight and the initial credibility score are multiplied to obtain a modified credibility score of the detector pair; The distance difference of the detector pair is weighted by using the modified credibility score, a hyperbolic equation set is constructed according to the weighted distance difference, and a weighted least square method with constraints is used to solve the hyperbolic equation set to obtain an initial position estimation value; The position of the signal source at the previous moment is obtained, the position constraint region at the current moment is determined by combining the maximum moving speed constraint, and the adaptive step iterative search is performed in the position constraint region with the initial position estimation value as the starting point to obtain the current iterative position; Based on the current iterative position, the theoretical distance difference of the detector pair is calculated, the deviation of the theoretical distance difference and the measured distance difference is taken as a positioning residual error, the modified credibility score is dynamically attenuated and updated according to the positioning residual error to obtain a real-time credibility score, the weight of the hyperbolic equation set is updated by using the real-time credibility score, and the iterative solution is continued until the position estimation value converges to obtain the real-time position of the signal source.

7. The method of claim 1, wherein, According to the time sequence relationship between the moving track data and the mobile phone communication state, the current mobile phone use scene is judged, the threat level of the mobile phone use scene is evaluated according to the pre-set scene threat level parameter, and alarm information is sent to the monitoring terminal according to the threat level evaluation result, including: The moving track data and the communication state data of the mobile phone are obtained, the motion features and the spatial distribution features in the moving track data are extracted, the communication behavior features and the signal quality features in the communication state data are extracted, and time sequence feature data is generated; The time sequence feature data is segmented by time window, the correlation degree between features in each time window is calculated, a feature correlation matrix is constructed, and the current mobile phone use scene is obtained by matching the feature correlation matrix with a pre-set mobile phone use scene model. The time sequence feature data is compared and analyzed with preset normal scene features corresponding to the current mobile phone use scene, a feature deviation sequence is generated, time sequence clustering is performed on the feature deviation sequence to obtain an abnormal event cluster, and an abnormal event feature is generated based on a time span and a number of occurrences of the abnormal event cluster; A threat score is calculated according to the abnormal event feature, a preset threat level parameter corresponding to the current mobile phone use scene is combined for weighting, a real-time threat level is obtained, when the real-time threat level exceeds an early warning threshold, an abnormal event feature triggering the early warning is taken as a monitoring template, a number of matched abnormal events is counted in an observation period, when the number of matched abnormal events exceeds a set threshold and the threat level continuously rises, an alarm information is sent to a monitoring terminal.

8. A multi-band mobile phone signal detection and alarm system for implementing the method of any one of the preceding claims 1-7, characterized by, Comprise: The first unit is used for collecting electromagnetic signals in the target area and pre-processing to obtain pre-processed electromagnetic signals; The second unit is used for performing fast Fourier transform on the pre-processed electromagnetic signals to obtain signal frequency spectrum, extracting mobile phone signal feature band information from the signal frequency spectrum, the mobile phone signal feature band information includes uplink frequency band signal strength value and downlink frequency band signal strength value, and judging the mobile phone communication state according to the ratio of the uplink frequency band signal strength value and the downlink frequency band signal strength value; The third unit is used for obtaining signal strength data in the target area, calculating real-time position coordinates of the mobile phone signal source based on the mobile phone communication state and the signal strength data by using a signal strength difference positioning algorithm, and recording the change of the position coordinates with time to obtain movement trajectory data; The fourth unit is used for judging the current mobile phone use scene according to the time sequence relationship between the movement trajectory data and the mobile phone communication state, evaluating the threat level of the mobile phone use scene according to the preset scene threat level parameter, and sending alarm information to the monitoring terminal according to the threat level evaluation result.

9. An electronic device, comprising: Comprise: A processor; A memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions are executed by the processor to implement the method of any one of claims 1 to 7.

Images