Electromagnetic wave detection device

Abstract

This technology allows for easier detection of the electromagnetic waves of a specific modulated wireless signal compared to conventional techniques. [Solution] The electromagnetic wave detection device 10 of the present invention comprises a measurement unit 3 and a calculation processing unit 5. The measurement unit 3 receives electromagnetic waves and outputs waveform data of the received electromagnetic waves. The calculation processing unit 5 calculates the spectral entropy of the electromagnetic waves based on the waveform data, and detects the presence of an electromagnetic wave generator that generates a predetermined electromagnetic wave or the arrival of such electromagnetic waves when the calculated spectral entropy is the spectral entropy measured in advance for an electromagnetic wave generator that generates modulated electromagnetic waves or a value close thereto. Furthermore, the calculation processing unit 5 compares the calculated spectral entropy of the electromagnetic waves with the spectral entropy of floor noise measured in advance, and detects the presence or arrival of electromagnetic waves when the calculated spectral entropy of the electromagnetic waves becomes smaller than the spectral entropy of floor noise.

Description

Technical Field

[0001] The present invention relates to an electromagnetic wave detection device that detects electromagnetic waves of a modulated wireless signal such as an analog wireless signal or a digital wireless signal, or a predetermined electromagnetic noise. Here, the analog wireless signal refers to a wireless signal modulated by a predetermined analog modulation method, and the digital wireless signal refers to a wireless signal modulated by a predetermined digital modulation method.

Background Art

[0002] Conventionally, various studies have been conducted on the evaluation of radio wave interference with a wireless communication system in a drone or the like from electromagnetic noise including electromagnetic waves from another wireless communication system or noise from a power supply circuit. On the other hand, in a diversifying electromagnetic environment, various EMC noise processing methods have been proposed (see, for example, Non-Patent Document 1).

[0003] In addition, in Non-Patent Document 2, in cognitive radio, due to the necessity of accurately recognizing wireless environment information, identification methods for modulation methods for each symbol, such as identification of BPSK signals and QPSK signals, and identification of QPSK signals and QAM signals, have been studied. In particular, a method for identifying PSK signals and QAM signals using the likelihood of digital modulation signals under an unknown SNR state has been proposed.

[0004] Furthermore, in Patent Document 1, in the "noise source identification method and information system", after storing information including the frequency, modulation frequency, and intensity of electromagnetic noise generated by a device different from the target device in a noise source DB, information is compared with the information in the noise source DB from the measurement data of the electromagnetic noise of the target device to identify the noise source.

[0005] In addition, in Patent Document 2, based on the characteristics of the signal spectrum envelope from the inertial sensor of a mobile device located at the same location as the user, the position of the mobile device with respect to the user is detected by classifying the position state of the mobile device with respect to the user. In particular, the "spectral entropy" is measured to infer the position of the mobile device with respect to the user. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2020-204561 [Patent Document 2] Japanese Patent Publication No. 2016-39999 [Non-patent literature]

[0007] [Non-Patent Document 1] Masao Masugi, "EMC Noise Analysis Processing Methods in Diverse Electromagnetic Environments," IEICE Transactions on Electronics, Information and Communication Engineers, Vol. J96-B, No. 4, pp. 476-485, April 2013. [Non-Patent Document 2] Shinji Nishijima et al., “PSK and QAM Classification by Likelihood under Unknown SNR Condition,” Journal of Signal Processing, vol.20, no.4,pp183-187,July 2016 [Non-Patent Document 3] 3GPP TS 36.141, V11.6.1, Nov. 2013 [Overview of the project] [Problems that the invention aims to solve]

[0008] These conventional technologies require the pre-storement of detailed electromagnetic noise parameters and complex signal processing. However, if the type of electromagnetic wave in a modulated radio signal could be easily identified, it would lead to the suppression of unwanted radio interference according to that type of electromagnetic wave. These conventional technologies have the problem that they cannot easily detect the electromagnetic wave of a specific modulated radio signal or specific electromagnetic noise.

[0009] The object of the present invention is to provide an electromagnetic wave detection device that can easily detect electromagnetic waves of a modulated specific radio signal, or specific electromagnetic noise, compared to the prior art. [Means for solving the problem]

[0010] An electromagnetic wave detection device according to one aspect of the present invention is: A measuring means that receives electromagnetic waves and outputs waveform data of the received electromagnetic waves, A computational processing means that calculates the spectral entropy of the received electromagnetic wave based on the waveform data of the electromagnetic wave, and detects the presence or arrival of a modulated electromagnetic wave of a specific radio signal, or the presence or arrival of a specific electromagnetic noise, based on the calculated spectral entropy of the electromagnetic wave, It is equipped with.

[0011] Here, the arithmetic processing means is (A) When the calculated spectral entropy of the electromagnetic wave is equal to or near the spectral entropy previously measured for an electromagnetic wave generator that generates a modulated predetermined electromagnetic wave or predetermined electromagnetic noise, the presence or arrival of the electromagnetic wave or electromagnetic noise is detected, or (B) The spectral entropy of the calculated electromagnetic wave is compared with the spectral entropy of the floor noise measured in advance, and the presence or arrival of the electromagnetic wave or electromagnetic noise is detected when the calculated spectral entropy of the electromagnetic wave becomes smaller than the spectral entropy of the floor noise. [Effects of the Invention]

[0012] Therefore, according to the electromagnetic wave detection device of the present invention, compared with the prior art, it is possible to simply detect the presence or arrival of the electromagnetic wave of a specific modulated radio signal or the presence or arrival of a specific electromagnetic noise based on the calculated spectral entropy of the electromagnetic wave. For example, when the calculated spectral entropy of the electromagnetic wave becomes the spectral entropy measured in advance for an electromagnetic wave generator that generates a predetermined modulated electromagnetic wave or a predetermined electromagnetic noise or a value in the vicinity thereof, or when the calculated spectral entropy of the electromagnetic wave is compared with the spectral entropy of the floor noise measured in advance and the calculated spectral entropy of the electromagnetic wave becomes smaller than the spectral entropy of the floor noise, the presence or arrival of the predetermined electromagnetic wave or the predetermined electromagnetic noise can be detected.

Brief Description of the Drawings

[0013] [Figure 1] It is a block diagram showing a configuration example of the electromagnetic wave detection device 10 according to the embodiment. [Figure 2] It is a flowchart showing the first pre-measurement process executed by the arithmetic processing unit 5 in FIG. 1. [Figure 3] It is a flowchart showing the second pre-measurement process executed by the arithmetic processing unit 5 in FIG. 1. [Figure 4] It is a flowchart showing the measurement arithmetic process executed by the arithmetic processing unit 5. [Figure 5] It is a block diagram showing a configuration example of the neural network 11 according to the first modification of the search determination in step S25 of FIG. 4. [Figure 6] It is a block diagram showing a configuration example of the neural network 12 according to the second modification of the search determination in step S25 of FIG. 4. [Figure 7] It is a block diagram showing a configuration example of the neural network 13 according to the third modification of the search determination in step S25 of FIG. 4. [Figure 8]It is a graph showing the temporal change of the spectral entropy related to each signal, which is the simulation result of the electromagnetic wave detection device 10 in FIG. 1. [Figure 9] It is a graph showing the characteristics of the time-averaged spectral entropy with respect to the SNR related to each signal without a band-pass filter, which is the simulation result of the electromagnetic wave detection device 10 in FIG. 1. [Figure 10] It is a graph showing the characteristics of the time-averaged spectral entropy with respect to the SNR related to each LTE signal without a band-pass filter, which is the simulation result of the electromagnetic wave detection device 10 in FIG. 1. [Figure 11] It is a graph showing the characteristics of the time-averaged spectral entropy with respect to the SNR related to the QPSK signal with band-pass filters of various bandwidths, which is the simulation result of the electromagnetic wave detection device 10 in FIG. 1. [Figure 12] It is a graph showing the characteristics of the time-averaged spectral entropy with respect to the SNR related to the 64QAM signal with band-pass filters of various bandwidths, which is the simulation result of the electromagnetic wave detection device 10 in FIG. 1. [Figure 13] It is a graph showing the result of clustering the spectral entropy values of each signal using the k-means method when SNR = -5 dB, which is the simulation result of the electromagnetic wave detection device 10 in FIG. 1. [Figure 14] It is a graph showing the result of clustering the spectral entropy values of each signal using the k-means method when SNR = -4 dB, which is the simulation result of the electromagnetic wave detection device 10 in FIG. 1. [Figure 15] It is a block diagram showing an application example of the electromagnetic wave detection device 10 in FIG. 1. [Figure 16] It is a graph showing the received power with respect to the distance between the eavesdropping device 30 and the electromagnetic wave detection device 10 when measuring the received power by the electromagnetic wave detection device 10 in FIG. 1 when the eavesdropping device 30 as an electromagnetic wave generator is placed in the room 20 which is a predetermined free space, which is the simulation result related to the eavesdropping device in the application example of FIG. 15. [Figure 17]Figure 15 shows the simulation results for the listening device 30 in the application example, and is a graph showing the spectral entropy as a function of the distance between the listening device 30 and the electromagnetic wave detection device 10, as measured by the electromagnetic wave detection device 10 in Figure 1. [Modes for carrying out the invention]

[0014] Embodiments and modified examples of the present invention will be described below with reference to the drawings. The same or similar components are denoted by the same reference numerals.

[0015] (Definition of spectral entropy) In Shannon's information theory, information is quantified by its information content, and the nature of the information source is represented by "information entropy." In the entropy function H(p) that represents information entropy, when the probability p is varied from 0 to 1, the value of the function H(p) is maximized when p=q=0.5. That is, when symbols s1 and s2 occur with equal probability and are completely unpredictable, the probability of the information content of the symbols is maximized, which means that "the more disordered the information source, the larger the average information content." In other words, the average information content can represent the disorder of the information source, and can therefore be called "information entropy."

[0016] In this embodiment, we focus on "spectral entropy," which is information entropy that measures the spectral distribution of a signal, and in particular, we demonstrate that the type of digital radio signal can be identified using the spectral entropy characteristics of a digital radio signal. Spectral entropy is calculated by processing the normalized power distribution of a signal in the frequency domain as a probability distribution and calculating the information entropy. The equation for spectral entropy can be obtained from the power spectrum and probability distribution equations of the signal as follows.

[0017] When the length of a time-series signal x(n) is N, that is, when n=0,1,...,N-1, the discrete Fourier transform X(m) of the time-series signal x(n) is expressed by the following equation.

[0018]

number

[0019] Here, m = 0, 1, ..., M-1, j is the imaginary unit, and 2π / N corresponds to each frequency of each frequency component.

[0020] Next, the power spectrum S(m) of the discrete Fourier transform X(m) of x(n) is expressed by the following equation using the discrete Fourier transform X(m) of x(n).

[0021]

number

[0022] Here, the variable n in the time-series signal x(n) represents the sampling point in the time domain, and the variable m in the discrete Fourier transform X(m) represents the discrete point (index) in the frequency domain. Therefore, the probability distribution P(m) is expressed by the following equation.

[0023]

number

[0024] Here, the spectral entropy H is expressed by the following equation.

[0025]

number

[0026] Next, let's consider the time-frequency power spectrum S(t,m). The time-frequency power spectrum S(t,m) is obtained by the short-time Fourier transform (STFT) to find the frequency power spectrum at a specific time t, and is defined as follows.

[0027]

number

[0028] Here, X(t,m) is the Short-Time Fourier Transform (STFT) of the time-series signal x(n), representing the Discrete Fourier Transform of the signal divided in the time domain by time t. When the frequency domain index m is converted to a physical frequency f, the power spectrum S(t,f) at time t and frequency f is expressed by the following equation.

[0029]

number

[0030] In this case, if the time-frequency power spectrogram S(t,f) is known, in order to find the probability distribution P(m) of the frequency power spectrum, first, the power is integrated (summed) over the entire time t for each frequency, and the frequency power spectrum Σ is obtained. t After calculating S(t,m), the sum of the powers at all frequencies is calculated as Σ. f Σ t The value is normalized by S(t,f). Therefore, the probability distribution P(m) of the frequency power spectrum in this case is expressed by the following equation using the time-frequency power spectrum S(t,m) from equation (3).

[0031]

number

[0032] Here, the spectral entropy H is given by equation (4) and is expressed by the following equation.

[0033]

number

[0034] Here, when calculating the instantaneous spectral entropy for a given time-frequency power and spectrogram S(t,f), the probability distribution P(t,m) at time t is expressed by the following equation from equation (7).

[0035]

number

[0036] Here, from equation (9), the spectral entropy at time t is expressed by the following equation.

[0037]

number

[0038] In this embodiment, the spectral entropy of a sinusoidal signal, a digitally modulated signal, and a digital radio signal is calculated using equation (10). In this embodiment, the spectral entropy of the signal spectrum, which is part of the information entropy, can be calculated using, for example, Matlab®.

[0039] The spectral entropy function value succinctly represents the spectral characteristics of a signal, with a value of 1 when the characteristics are most ambiguous and 0 when there is no ambiguity. Applying this to the spectral entropy of a signal, for example, it is thought to be 1 when it is random noise such as floor noise, and approaches 0 when there is no influence from random noise, the spectral characteristics of the signal are clear, and there is no randomness.

[0040] (Embodiment) Figure 1 is a block diagram showing an example configuration of an electromagnetic wave detection device 10 according to an embodiment. In Figure 1, the electromagnetic wave detection device 10 comprises an antenna 1, an input switching switch (hereinafter referred to as a switch) 2, a measurement unit 3, an AD conversion unit 4, an arithmetic processing unit 5, a storage unit 6, a display unit 7, a termination resistor 8, and a signal generation unit 9. Here, the storage unit 6 is connected to the arithmetic processing unit 5 and includes a floor noise memory 6A and an SNR vs. spectral entropy (SE) table memory 6B for each type of electromagnetic wave.

[0041] Here, "floor noise" refers to white noise, which is so-called random noise such as thermal noise generated in the termination resistor 8 and each processing unit 3, 4, and 5. "Electromagnetic waves" include electromagnetic noise such as digital radio signals modulated with digital modulation signals, analog radio signals modulated with analog modulation signals, unmodulated sine wave signals, and switching noise generated in power supply circuits. Furthermore, "SNR" is the signal-to-noise power ratio or signal-to-noise level ratio, where level refers to the level of an electrical signal such as voltage or current.

[0042] In Figure 1, Antenna 1 receives and detects electromagnetic waves and outputs them to the measurement unit 3 via Switch 2. Antenna 1 can be any element that detects electromagnetic waves, such as a probe. Switch 2 is controlled by the arithmetic processing unit 5 to selectively switch between Antenna 1, Termination resistor 8, and Signal generation unit 9, and connects the selected element to the measurement unit 3. The measurement unit 3 converts the electromagnetic waves (including electromagnetic noise) input via Switch 2 into analog electrical signals and outputs them to the AD conversion unit 4. The AD conversion unit 4 samples the input analog electrical signals at a predetermined sampling frequency, performs AD conversion to convert them into digital waveform data, and outputs it to the arithmetic processing unit 5. The arithmetic processing unit 5 also writes and reads data to and from the storage unit 6, and controls the operation of Switch 2 and Signal generation unit 9. Specifically, the arithmetic processing unit 5 performs the first pre-measurement process in Figure 2, the second pre-measurement process in Figure 3, and the measurement calculation process in Figure 4 to detect the arrival or presence of electromagnetic waves different from floor noise, detect their type, and display the detection result on the display unit 7.

[0043] Furthermore, the measurement unit 3 can calculate and measure the received power (or received intensity) based on the received amplitude of the input electromagnetic wave, and output it to the arithmetic processing unit 5.

[0044] Figure 2 is a flowchart showing the first pre-measurement process performed by the arithmetic processing unit 5 in Figure 1.

[0045] In step S1 of Figure 2, switch 2 is switched to the termination resistor 8 side, terminating the input terminal of the measurement unit 3 with the termination resistor 8. Next, in step S2, based on the waveform data of the digital value measured by the measurement unit 3 and converted by the AD conversion unit 4, the floor noise level N and the spectral entropy SEnoise of the floor noise are calculated and stored in the floor noise memory 6A of the storage unit 6, thus ending the first pre-measurement process.

[0046] Figure 3 is a flowchart showing the second pre-measurement process performed by the arithmetic processing unit 5 in Figure 1.

[0047] In step S11 of Figure 3, switch 2 is switched to the signal generation unit 9 side, and the input terminal of the measurement unit 3 is connected to the signal generation unit 9. Next, in step S12, a predetermined electromagnetic wave is generated from the signal generation unit 9, and while the signal level S is changed, the signal level S of the electromagnetic wave is calculated based on the waveform data of the digital value measured by the measurement unit 3 and converted by the AD conversion unit 4. After calculating the SNR based on the floor noise level N of the floor noise memory 6A and the calculated electromagnetic wave signal level S, the spectral entropy (pre-measured value) SEprem of the electromagnetic wave is calculated based on the waveform data, and the spectral entropy (pre-measured value) SEprem data for SNR is stored in the SNR vs. SE table memory 6B of the storage unit 6. Furthermore, in step S13, it is determined whether or not pre-measurement processing has been performed for all assumed types of electromagnetic waves. If NO, the process returns to step S12, while if YES, the second pre-measurement processing is terminated.

[0048] Figure 4 is a flowchart showing the measurement calculation process performed by the arithmetic processing unit 5.

[0049] In step S21 of FIG. 4, switch 2 is switched to the antenna 1 side to connect the input terminal of measurement unit 3 to antenna 1. Next, in step S22, the electromagnetic wave to be measured is received by antenna 1, and based on the waveform data of the digital value measured by measurement unit 3 and AD-converted by AD conversion unit 4, the signal level S of the electromagnetic wave is calculated. After calculating the SNR based on the level N of the floor noise in floor noise memory 6A and the calculated signal level S of the electromagnetic wave, the spectral entropy (measured value) SEmeasure of the electromagnetic wave is calculated based on the waveform data. In step S23, it is determined whether SEmes < SEnoise. If YES, the process proceeds to step S24, while if NO, the process returns to step S22.

[0050] In step S24, it is determined that "an electromagnetic wave has arrived or exists" other than the floor noise is detected at antenna 1, and "Electromagnetic wave detected" is displayed on display unit 7. Next, in step S25, for the detected electromagnetic wave, based on the spectral entropy (measured value) SEmeasure with respect to the SNR, the spectral entropy (pre-measured value) SEprem in the SNR-SE table memory 6B of storage unit 6 is searched to retrieve the type of the corresponding electromagnetic wave, and the retrieved "type of electromagnetic wave" is displayed on display unit 7.

[0051] (Modified example) In the measurement calculation process shown in Figure 4 above, in steps S2, S12, S22-S23, the floor noise N and spectral entropy SEnoise, SEprem, and SEmeasure may be measured across the entire measurement band of the measurement unit 3. However, the present invention is not limited to this, and the floor noise N and spectral entropy SEnoise, SEprem, and SEmeasure may be measured while sweeping the frequency using a bandpass filter of a predetermined bandwidth within the measurement band of the measurement unit 3. Instead of the bandpass filter, a filter such as a low-pass filter or high-pass filter of a predetermined bandwidth may be used. The filter may be provided in the measurement unit 3, or the filtering process may be performed by calculation processing in the calculation processing unit 5. Here, spectral entropy SEnoise, SEprem, and SEmeasure are, for example, time-averaged values ​​over a predetermined time, and spectral entropy (measured value) SEmeasure may be a real-time value.

[0052] In the embodiments described above, "floor noise" is defined as white noise, which is so-called random noise such as thermal noise generated in the termination resistor 8 and each of the processing units 3, 4, and 5. However, the present invention is not limited to this, and the switch 2 may be switched to the antenna 1 side to detect floor noise including electromagnetic noise in the free space on which the electromagnetic wave detection device 10 is placed and noise from each of the processing units 3, 4, and 5 within the device.

[0053] In step S25 (search process) in Figure 4, the search for spectral entropy involves determining, for example, whether the spectral entropy (measured value) SEmeasure (time-averaged or real-time value) matches the spectral entropy (pre-measured value) SEprem in the SNR vs. SE table memory 6B within a predetermined margin range, with the median value. Alternatively, the search process may be performed using the neural networks described in the modified examples 1 to 3 in Figures 5 to 7.

[0054] The data in the SNR vs. SE table memory 6B for each type of electromagnetic wave in Figure 1 is, as described above, spectral entropy (pre-measured value) SEprem for SNR, which is stored according to the type of modulated electromagnetic wave or electromagnetic noise. However, the present invention is not limited to this, and may also include frequency or frequency band (calculated in the arithmetic processing unit 5) and spectral entropy (pre-measured value) SEprem for SNR. By adding SNR, or SNR and frequency or frequency band data to the search, the accuracy of electromagnetic wave detection and detection of its type can be increased.

[0055] The neural networks 11-13 in the modified examples 1-3 of Figures 5-7 are configured to include an input layer, one or more hidden layers, and an output layer. Here, the neural networks 11-13 are pre-trained using the following input data (with known output signals) and output signals as training data. (1) The neural network 11 in Figure 5 takes spectral entropy as input and outputs a decision signal indicating the detection of one or more types of electromagnetic waves. In the modified example 1 of Figure 5, the type of electromagnetic wave is detected based solely on spectral entropy. That is, in the embodiment of Figure 1, the SNR vs. SE table memory 6B may not store SNR, and the arithmetic processing unit 5 may detect the type of electromagnetic wave based solely on spectral entropy. (2) The neural network 12 in Figure 6 takes spectral entropy and SNR as input and outputs a decision signal indicating the detection of one or more types of electromagnetic waves. By adding SNR as input data, the accuracy of electromagnetic wave detection and detection of its type can be increased. (3) The neural network 13 in Figure 7 takes spectral entropy, SNR, and frequency or frequency band as input and outputs a decision signal indicating the detection of one or more types of electromagnetic waves. By adding SNR and frequency or frequency band as input data, the accuracy of electromagnetic wave detection and detection of its type can be increased.

[0056] Here, the decision signals, which are the output signals of neural networks 11-13, may be set to, for example, 1 when detected and 0 when not detected, or alternatively, they may output detection likelihood values ​​ranging from 0 to 1.

[0057] (Effects of the embodiment) As described above, according to this embodiment, even if the electromagnetic wave is below the floor noise level, the presence or arrival of the electromagnetic wave can be detected by comparing the calculated spectral entropy (measured value) SEmeasure of the electromagnetic wave with the spectral entropy SEnoise of the floor noise, using calculations that are extremely simple compared to the prior art. Furthermore, the type of electromagnetic wave can be detected by searching the SNR vs. SE table memory 6B of the storage unit 6 based on the calculated spectral entropy (measured value) SEmeasure of the electromagnetic wave. [Examples]

[0058] (Computer simulation) The computing system for the computer simulation was constructed using personal computers that run Matlab® code. Here, a digital radio signal was generated by digitally modulating it according to a pseudo-random data signal with a predetermined bitrate, and then upconverted to a high-frequency digital radio signal with a predetermined carrier frequency. Random noise corresponding to floor noise with a predetermined SNR was added to the high-frequency digital radio signal, and then the spectral entropy in the computer simulation was calculated. Here, the SNR was changed by varying the applied level of random noise. The results of the computer simulation for the spectral entropy of each digital radio signal in the high-frequency band are shown below.

[0059] (Simulation conditions) In computer simulations in the high-frequency band, spectral entropy calculations were performed using LTE signals (with a carrier frequency of 865 MHz, which is actually used) and QPSK and 64QAM signals with a 10 MHz carrier, as well as a 10 MHz sine wave (SIN) signal, as examples, used in the LTE mobile phone system of 3GPP® (see Non-Patent Literature 3). Here, in order to sufficiently cover the bandwidth of each signal, the sampling frequency was set to four times the carrier frequency, and the recording settings of the vector signal analyzer (VSA) and the signal settings of the signal analyzer were performed as follows.

[0060] (1) Center frequency = 865MHz (LTE), 10MHz (QPSK, 64QAM, SIN). (2) Bandwidth = 10MHz (LTE), 2MHz (QPSK, 64QAM, SIN). (3) Recording time = 300 ms. However, the spectral entropy calculation was performed for the intermediate period of 100 ms during which the signal stabilized.

[0061] Furthermore, to demonstrate the correspondence with the experimental results, we also show the simulation results when a bandpass filter is inserted using the bandpass function in Matlab® to limit the bandwidth immediately before calculating the spectral entropy.

[0062] For the LTE signal, we used data generated using the MATLAB® application, which produced test mode signals as defined by 3GPP (TMN3.1 (64QAM), TMN3.2 (16QAM), TMN3.3 (QPSK); where ( ) is the format of the primary modulation signal, and for secondary modulation, the LTE signal (OFDM) was generated using a pseudo-random pattern).

[0063] Figure 8 shows the simulation results of the electromagnetic wave detection device 10 in Figure 1, and is a graph showing the temporal change in spectral entropy for each signal. In Figure 8, the temporal change in the time-series spectral entropy data values ​​is shown for each digital radio signal and sinusoidal signal in a state where there is almost no white noise (floor noise) (SNR = 30 dB).

[0064] As is clear from Figure 8, each digital radio signal is modulated according to pseudo-random data, so its spectral entropy changes slightly, but each has its own unique spectral entropy that is different from the others. In Figure 8, it can be seen that the spectral entropy of the LTE signal is close to 1, and approaches the value of the spectral entropy of a sinusoidal signal depending on the modulation order of the PSK or QAM signal. The ideal value of the spectral entropy of white noise is 1, and it is thought that the spectral entropy of each digital radio signal takes a value between 1 and the spectral entropy of a sinusoidal signal. Since the LTE signal is randomized by second-order modulation, it is close to 1, and as the modulation order of the PSK or QAM signal increases, its spectrum approaches that of a sinusoidal signal and approaches the spectral entropy of a sinusoidal signal.

[0065] Figure 9 shows the simulation results of the electromagnetic wave detection device 10 in Figure 1, and is a graph showing the characteristics of time-averaged spectral entropy with respect to SNR for each signal without a bandpass filter. That is, Figure 9 shows the time-averaged spectral entropy for each digital radio signal and sinusoidal signal when the SNR is changed. In this simulation, "time-averaged spectral entropy" refers to the value (average value) obtained by time-averaging the time-series spectral entropy over the measurement time.

[0066] As is clear from Figure 9, even when the SNR is changed, the spectral entropy for each digital radio signal has the same magnitude relationship as shown in Figure 8. However, the spectral entropy of the LTE signal is particularly small when the SNR is below -10 dB, and is smaller than that of the QPSK and 64QAM signals. This is thought to be because its eigenvalue hardly changes with the two modulations, primary and secondary.

[0067] In Figure 9, a notable feature is that even if the SNR is less than 0 dB, that is, even if each digital radio signal is hidden by floor noise, as long as the SNR is at least -6 dB or higher, the spectral entropy values ​​of each digital radio signal remain different from one another, making it possible to distinguish the type of each digital radio signal. In other words, by measuring the spectral entropy value when there is no digital radio signal, it is possible to detect the presence of some digital radio signal when that measured value decreases.

[0068] Furthermore, in Figure 9, only the case where the primary modulation scheme is 64QAM is shown for the LTE signal. The reason for this is that the change in SNR is small even when the primary modulation scheme is changed. Figure 10 shows the spectral entropy characteristics with respect to SNR for other primary modulation schemes. As is clear from Figure 10, even when the primary modulation scheme is changed, the spectral entropy value changes within approximately 0.985 to 0.96, and the range of change is small.

[0069] Figure 11 is a simulation result of the electromagnetic wave detection device 10 in Figure 1, and is a graph showing the time-averaged spectral entropy characteristics with respect to SNR for QPSK signals with bandpass filters of various bandwidths. Specifically, Figure 11 shows the characteristics of the spectral entropy characteristics with respect to SNR in a QPSK signal when the signal components for which spectral entropy is calculated are limited to a predetermined bandwidth using a bandpass filter. As is clear from Figure 11, reducing the limited bandwidth reduces the white noise component and thus the spectral entropy value decreases.

[0070] Figure 12 shows the simulation results of the electromagnetic wave detection device 10 in Figure 1, and is a graph showing the time-averaged spectral entropy characteristics with respect to SNR for 64QAM signals with bandpass filters of various bandwidths. Specifically, Figure 12 shows the spectral entropy characteristics with respect to SNR for a 64QAM signal when the signal components used to calculate spectral entropy are limited to a predetermined bandwidth using a bandpass filter.

[0071] As is clear from Figure 12, reducing the bandwidth limit reduces the white noise component and thus the spectral entropy value. However, compared to the QPSK signal in Figure 11, the effect of bandwidth limiting is small. This is thought to be because the spectral entropy value of the 64QAM signal asymptotically approaches that of the sinusoidal signal, and as mentioned above, in a given electromagnetic environment, it cannot be smaller than the spectral entropy of the sinusoidal signal, resulting in this characteristic.

[0072] (Cluster analysis based on simulation results) Below, we verified whether the types of digital radio signals can be identified using the k-means method, a cluster analysis method, based on the results of computer simulations of the spectral entropy of three different types of digital radio signals. Here, the MATLAB® function kmeans was used as the verification tool for the k-means method.

[0073] Here, among the spectral entropy data values ​​obtained from the above simulation, (1) 512 data values ​​(data index = 1-512) relating to the spectral entropy of the LTE signal, (2) 512 data values ​​(data index = 513-1024) relating to the spectral entropy of the QPSK signal, (3) 512 data values ​​(data index = 1025-1536) relating to the spectral entropy of the 64QAM signal, For a sequence of 1536 data values ​​including [the specified value], cluster analysis was performed assuming a known number of clusters (NCL) of 3.

[0074] Figures 13 and 14 are simulation results of the electromagnetic wave detection device 10 in Figure 1, respectively, and are graphs showing the results of clustering the spectral entropy values ​​of each signal using the k-averaging method when SNR = -5dB and -4dB. In Figure 13, at SNR = -5dB, the clustering method does not adequately distinguish between the LTE signal and the QPSK signal, but in Figure 14, at SNR = -4dB, it is shown that the clustering method can distinguish between the LTE signal, the QPSK signal and the 64QAM signal.

[0075] (Specific effects and benefits) As explained above, since the type of digital radio signal can be identified when the SNR is less than 0 dB, it may be possible to detect digital radio signals hidden by floor noise while a moving object such as a drone is in motion. Specifically, by measuring the spectral entropy value when there is no digital radio signal, it is possible to detect the presence or arrival of some kind of digital radio signal when the measured value decreases, and using this type identification method as a trigger, it is thought that the type of digital radio signal can be identified.

[0076] Conventional technologies require pre-storing electromagnetic noise parameters and complex signal processing. In this embodiment, spectral entropy, which is the information entropy related to the signal spectrum, is used as a parameter for identifying the type of digital radio signal. In particular, computer simulations and high-frequency experiments have shown that by using spectral entropy averaged over a predetermined time period, the type of digital radio signal can be identified in the high-frequency band even when the SNR is less than 0 dB.

[0077] Furthermore, by measuring the spectral entropy value when no digital radio signals are present, it is possible to detect the presence or arrival of some kind of digital radio signal when this value decreases. Using this as a trigger, the type of digital radio signal can be identified using this type identification method. Here, easily identifying the type of digital radio signal, which is electromagnetic noise for wireless communication devices mounted on mobile devices, could lead to interference suppression.

[0078] (Examples of application) Figure 15 is a block diagram showing an application example of the electromagnetic wave detection device 10 shown in Figure 1. Here, Figure 15 shows the situation when an example of an electromagnetic wave generating device, such as a listening device 30, is not placed in a room 20 in free space, and when the listening device 30 is placed in the room. In the application example of Figure 15, it is assumed that there are no electromagnetic wave generating devices other than the listening device 30 in the room 20, and switch 2 is switched to the antenna 1 side.

[0079] (A) When there is no listening device 30 In Figure 15, when the listening device 30 is not placed, the electromagnetic wave detection device 10 detects floor noise, which includes noise in the room 20 and noise from each of the processing units 3, 4, and 5 within the electromagnetic wave detection device 10, and calculates the spectral entropy. The entropy at this time corresponds to SNR ≤ -30 dB in Figures 9 to 12 and is a value close to 1 or less.

[0080] (B) When there is a listening device 30 In Figure 15, when the listening device 30 is placed, the electromagnetic wave detection device 10 detects the electromagnetic waves from the listening device 30 in the room 20 and calculates its spectral entropy. At this time, the detected electromagnetic waves include noise from each processing unit 3, 4, and 5 in the electromagnetic wave detection device 10, but the electromagnetic waves in the room 20 are essentially electromagnetic waves from the listening device 30. The entropy at this time is the spectral entropy value of the wireless signal of the electromagnetic waves from the listening device 30.

[0081] Next, the simulation results for the case where the listening device 30 generates an analog radio signal, for example, by analog FM modulation of a 399MHz carrier wave (transmission power of 10mW (10dBm)) according to the audio signal in room 20 are shown below.

[0082] Figure 16 is a simulation result relating to the listening device 30 in the application example of Figure 15, and is a graph showing the received power as a function of distance between the listening device 30 and the electromagnetic wave detection device 10 when the listening device 30, which is an electromagnetic wave generator, is placed in a predetermined free space, a room 20, and the received power is measured by the electromagnetic wave detection device 10 of Figure 1. Figure 17 is also a simulation result, and is a graph showing the received power as a function of distance between the listening device 30 and the electromagnetic wave detection device 10, as measured by the electromagnetic wave detection device 10 of Figure 1.

[0083] As is clear from Figure 16, the received power of the analog radio signal from the eavesdropping device 30 decreases exponentially with distance, but as is clear from Figure 17, the spectral entropy of the analog radio signal from the eavesdropping device 30 remains constant regardless of distance. In other words, when the spectral entropy of the electromagnetic waves from a predetermined electromagnetic wave generator is a known value that has been measured in advance, it is possible to detect the presence or arrival of electromagnetic waves from that electromagnetic wave generator, indicating the existence of that particular electromagnetic wave generator.

[0084] The inventors conducted an experiment of the application example shown in Figure 15 in a designated darkroom. The experimental results confirmed that the spectral entropy of the analog radio signal from the eavesdropping device 30 changes slightly with distance, but remains within a predetermined range (the previously measured spectral entropy or a nearby value). Therefore, when a spectral entropy within the predetermined range is detected, the presence of a predetermined electromagnetic wave generator, or the absence of an electromagnetic wave generator, can be detected.

[0085] In the above application examples, we have described the case where the electromagnetic wave generator is a listening device 30, but it may also be an electromagnetic wave generator other than a listening device 30 that generates electromagnetic waves having a predetermined spectral entropy. In other words, the electromagnetic wave detection device 10 according to the present invention may be a dedicated device for detecting a specific electromagnetic wave generator.

[0086] In the above embodiments or applications, when the free space differs depending on the presence or absence of an electromagnetic wave generator, the presence or absence of an electromagnetic wave generator may be detected based on the spectral entropy measured in each of the different free spaces.

[0087] In the embodiments and application examples described above, a modulated electromagnetic wave is detected. However, the present invention is not limited to this, and may also detect electromagnetic noise having a predetermined spectrum (for example, different spectra of noise such as white noise, pink noise, brown noise, and other colored noises will result in different spectral entropies) or its type, generated from, for example, a drone or a digital circuit.

[0088] In the embodiments and application examples described above, a modulated electromagnetic wave is detected. However, the present invention is not limited to this, and may be configured to detect the presence or arrival of an unmodulated radio signal (e.g., a sine wave) electromagnetic wave (see Figures 8 and 9). [Industrial applicability]

[0089] As described in detail above, the electromagnetic wave detection device according to the present invention can more easily detect the presence or arrival of a modulated electromagnetic wave of a specific radio signal, or the presence or arrival of a specific electromagnetic noise, based on the calculated spectral entropy of the electromagnetic wave, compared to the prior art. For example, the calculation processing means can detect the presence or arrival of the predetermined electromagnetic wave or the predetermined electromagnetic noise when the calculated spectral entropy of the electromagnetic wave is the spectral entropy measured in advance for a predetermined modulated electromagnetic wave or an electromagnetic wave generator that generates a predetermined electromagnetic noise, or when the calculated spectral entropy of the electromagnetic wave is compared with the spectral entropy of a predetermined floor noise, and the calculated spectral entropy of the electromagnetic wave becomes smaller than the spectral entropy of the floor noise. [Explanation of Symbols]

[0090] 1 Antenna 2 switches 3 Measuring part 4 AD Conversion Unit 5. Arithmetic Processing Unit 6 Memory section 6A Floor Noise Memory 6B SNR vs. SE Table Memory for Each Type of Electromagnetic Wave 7 Display section 8 Termination resistors 9. Signal generation unit 11,12,13 Neural Networks 20 rooms 30 Listening devices

Claims

1. A measuring means that receives electromagnetic waves and outputs waveform data of the received electromagnetic waves, A computational processing means that calculates the spectral entropy of the received electromagnetic wave based on the waveform data of the electromagnetic wave, and detects the presence or arrival of a modulated electromagnetic wave of a specific radio signal, or the presence or arrival of a specific electromagnetic noise, based on the calculated spectral entropy of the electromagnetic wave, An electromagnetic wave detection device equipped with the following features.

2. The aforementioned arithmetic processing means is (A) When the calculated spectral entropy of the electromagnetic wave is equal to or near the spectral entropy previously measured for an electromagnetic wave generator that generates a modulated predetermined electromagnetic wave or predetermined electromagnetic noise, the presence or arrival of the electromagnetic wave or electromagnetic noise is detected, or (B) The spectral entropy of the calculated electromagnetic wave is compared with the spectral entropy of the floor noise measured in advance, and when the spectral entropy of the calculated electromagnetic wave becomes smaller than the spectral entropy of the floor noise, the presence or arrival of the electromagnetic wave or the electromagnetic noise is detected. The electromagnetic wave detection device according to claim 1.

3. The calculation processing means detects the presence of the electromagnetic wave in the free space where the electromagnetic wave generator exists, based on the measured spectral entropy in each free space which differs depending on the presence or absence of the electromagnetic wave generator. The electromagnetic wave detection device according to claim 1.

4. The electromagnetic wave detection device further includes a storage means for storing at least one of the spectral entropy of an electromagnetic wave generator that generates a modulated predetermined electromagnetic wave or predetermined electromagnetic noise, and the spectral entropy of the floor noise, which has been measured in advance. An electromagnetic wave detection device according to any one of claims 1 to 3.

5. The calculation processing means calculates the spectral entropy of the electromagnetic wave, which is time-averaged over a predetermined measurement time, based on the waveform data of the electromagnetic wave. An electromagnetic wave detection device according to any one of claims 1 to 3.

6. (1) The calculation processing means calculates the spectral entropy of the electromagnetic wave after filtering using a filter of a predetermined bandwidth based on the waveform data of the electromagnetic wave, or (2) The measurement means outputs waveform data of the electromagnetic wave after filtering the received electromagnetic wave using a filter of a predetermined bandwidth, and the calculation processing means calculates the spectral entropy of the electromagnetic wave based on the waveform data of the electromagnetic wave. An electromagnetic wave detection device according to any one of claims 1 to 3.

7. The storage means stores the spectral entropy of at least one type of electromagnetic wave corresponding to the type, The calculation processing means detects the type of electromagnetic wave by searching for the spectral entropy of the calculated electromagnetic wave by referring to the spectral entropy stored in the storage means. The electromagnetic wave detection device according to claim 4.

8. The storage means stores the spectral entropy of at least one type of electromagnetic wave, which has been calculated in advance, as a table corresponding to the type in relation to the signal-to-noise ratio. The calculation processing means detects the type of electromagnetic wave calculated by searching for the spectral entropy of the calculated electromagnetic wave by referring to the table. The electromagnetic wave detection device according to claim 4.

9. The storage means stores the spectral entropy of at least one type of electromagnetic wave, which has been calculated in advance, as a table for frequency or frequency band and signal-to-noise ratio, and stores it as a table corresponding to the type. The calculation processing means calculates the frequency or frequency band of the received electromagnetic wave based on the waveform data of the electromagnetic wave, and detects the type of the calculated electromagnetic wave by searching for the spectral entropy of the calculated electromagnetic wave by referring to the table based on the calculated frequency or frequency band. The electromagnetic wave detection device according to claim 4.

10. The received electromagnetic wave is a digital radio signal, an analog radio signal, or electromagnetic noise. An electromagnetic wave detection device according to any one of claims 1 to 3.

Images