Signal analysis device, control circuit, storage medium, and signal analysis method

Abstract

A signal analysis device (100) comprising a frequency / sampling rate adjustment unit (101) which adjusts at least one of the frequency and sampling rate of a first received signal and which outputs the result as a second received signal, a frequency domain conversion unit (102) which converts the second received signal from a time domain signal into a frequency domain signal and which outputs the signal that has been converted into the frequency domain signal as a third received signal, an action value calculation unit (103) which is formed from a neural network and which uses the third received signal as an input to output a plurality of actions and a value for each of the actions, the actions adjusting the frequency or sampling rate of the first received signal in the frequency / sampling rate adjustment unit (101), and an action determination unit (104) which uses the value to determine the action to be taken in the frequency / sampling rate adjustment unit (101) from among the plurality of actions and which outputs the determined action to the frequency / sampling rate adjustment unit (101).

Description

Signal analysis device, control circuit, storage medium, and signal analysis method

[0001] The present disclosure relates to a signal analysis device, a control circuit, a storage medium, and a signal analysis method for analyzing a received signal.

[0002] Conventionally, in general wireless communications, communication parameters such as carrier frequency or center frequency, modulation method, and symbol rate are shared between a transmitting device and a receiving device. On the other hand, in fields such as spectrum monitoring, these communication parameters are unknown, and a receiving device must estimate these communication parameters from a received signal and perform synchronization processing, demodulation processing, and the like. Patent Document 1 discloses a technique in which a modulation method estimation device estimates the center frequency and modulation method of a wireless communication wave whose communication parameters are unknown using machine learning. The modulation method estimation device of Patent Document 1 uses a time-domain IQ (In-Phase / Quadrature-Phase) signal as input to a neural network, and obtains an estimated center frequency and modulation method as output from the neural network.

[0003] Japanese Patent Application Laid-Open No. 2022-39703

[0004] However, according to the above-mentioned conventional technology, the input to the frequency estimation neural network is a time-domain IQ signal. Therefore, when it is desired to support multiple modulation methods, the learning time increases according to the number of supported modulation methods. Furthermore, as the number of supported inputs and outputs increases, the neural network coefficients may not converge well during learning, which may result in a deterioration in estimation accuracy.

[0005] The present disclosure has been made in view of the above, and aims to provide a signal analysis device that can support a large number of modulation methods while suppressing an increase in learning time and a deterioration in estimation accuracy.

[0006] In order to solve the above-mentioned problems and achieve the object, the signal analysis device of the present disclosure is characterized by comprising: an adjustment unit that adjusts at least one of the frequency and sampling rate of a first received signal and outputs it as a second received signal; a frequency domain conversion unit that converts the second received signal from a time domain signal to a frequency domain signal and outputs the converted frequency domain signal as a third received signal; an action value calculation unit that is composed of a neural network, receives the third received signal as input, and outputs a plurality of actions and a value for each action that adjusts the frequency or sampling rate of the first received signal in the adjustment unit; and an action determination unit that determines an action to be taken in the adjustment unit from the plurality of actions based on the value, and outputs the determined action to the adjustment unit.

[0007] The signal analysis device according to the present disclosure has the advantage of being able to support a large number of modulation methods while suppressing an increase in learning time and a deterioration in estimation accuracy.

[0008] FIG. 1 is a diagram showing a configuration example of a signal analyzing device according to embodiment 1. FIG. 1 is a diagram showing a configuration example of a frequency domain transform unit included in the signal analyzing device according to embodiment 1. FIG. 2 is a diagram showing an image of the operation of the frequency domain transform unit included in the signal analyzing device according to embodiment 1. Flowchart showing the operation of the signal analyzing device according to embodiment 1. FIG. 2 is a diagram showing a configuration example of a processing circuit that realizes the signal analyzing device according to embodiment 1 when the processing circuit is configured with a processor and a memory. FIG. 3 is a diagram showing an example of a processing circuit that realizes the signal analyzing device according to embodiment 1 when the processing circuit is configured with dedicated hardware. FIG. 3 is a diagram showing a configuration example of a signal analyzing device according to embodiment 2. FIG. 4 is a diagram showing a configuration example of an input signal generating unit included in the signal analyzing device according to embodiment 2. FIG. 4 is a diagram showing a configuration example of a signal analyzing device according to embodiment 3.

[0009] A signal analysis device, a control circuit, a storage medium, and a signal analysis method according to embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0010] 1 is a diagram showing an example of the configuration of a signal analyzing device 100 according to a first embodiment. The signal analyzing device 100 is mounted on a receiving device (not shown) that receives a received signal with unknown communication parameters, and analyzes the received signal with unknown communication parameters. The signal analyzing device 100 includes a frequency sampling rate adjusting unit 101, a frequency domain transforming unit 102, an action value calculating unit 103, an action determining unit 104, a modulation scheme estimating unit 105, and a demodulating unit 106.

[0011] The frequency sampling rate adjustment unit 101 is an adjustment unit that adjusts at least one of the frequency and the sampling rate of a received signal input to the signal analyzing device 100 in accordance with the action determined by the action determination unit 104. The frequency sampling rate adjustment unit 101 outputs the adjusted signal to the frequency domain conversion unit 102 and the modulation scheme estimation unit 105. The received signal input to the signal analyzing device 100 is, for example, an IQ signal as described in the background art. In the following description, the received signal input to the signal analyzing device 100, i.e., the received signal to be adjusted by the frequency sampling rate adjustment unit 101, may be referred to as a first received signal, and the signal adjusted by the frequency sampling rate adjustment unit 101 may be referred to as a second received signal.

[0012] For example, immediately after the signal analyzing device 100 is started up and the action determining unit 104 has not yet determined an action for adjusting at least one of the frequency and the sampling rate, the frequency sampling rate adjusting unit 101 outputs the received signal input to the signal analyzing device 100 as is. Note that the frequency sampling rate adjusting unit 101 may adjust at least one of the frequency and the sampling rate a specified number of times in accordance with the action determined by the action determining unit 104, and then output the signal to the modulation scheme estimating unit 105. In other words, if the frequency sampling rate adjusting unit 101 has not adjusted at least one of the frequency and the sampling rate a specified number of times in accordance with the action determined by the action determining unit 104, the frequency sampling rate adjusting unit 101 may not output the signal to the modulation scheme estimating unit 105.

[0013] The frequency domain transform unit 102 transforms the signal adjusted by the frequency sampling rate adjustment unit 101 from a time domain signal to a frequency domain signal and outputs the converted frequency domain signal to the activity value calculation unit 103. The frequency domain transform unit 102 outputs a frequency axis value and a power value or amplitude value corresponding to the frequency axis value as the converted frequency domain signal. In the following description, the signal converted to a frequency domain signal by the frequency domain transform unit 102 may be referred to as a third received signal. The configuration of the frequency domain transform unit 102 will be described in detail. FIG. 2 is a diagram showing an example configuration of the frequency domain transform unit 102 included in the signal analyzing apparatus 100 according to the first embodiment. The frequency domain transform unit 102 includes an FFT (Fast Fourier Transform) 201, a power conversion unit 202, and a normalization unit 203. Here, an example will be described in which the frequency domain transform unit 102 outputs a frequency axis value and a power value corresponding to the frequency axis value as the converted frequency domain signal.

[0014] The FFT 201 converts the signal adjusted by the frequency sampling rate adjustment unit 101 from a time-domain signal to a frequency-domain signal. The FFT 201 outputs a frequency axis value and an FFT result corresponding to the frequency axis value as a frequency-domain signal. To improve the accuracy of the spectrum, the FFT 201 may store the input signal adjusted by the frequency sampling rate adjustment unit 101 in a memory (not shown) and average it using the Welch method or the like before performing the conversion process. After the conversion process, the FFT 201 may also average it in the frequency domain by applying a moving average filter or the like. The FFT 201 may also perform the conversion process by using a neural network separate from the behavioral value calculation unit 103 (described later) to train the neural network to improve the accuracy of the spectrum in the frequency domain. The frequency domain conversion unit 102 may use a DFT (Discrete Fourier Transform) as a method for converting the signal adjusted by the frequency sampling rate adjustment unit 101 from a time-domain signal to a frequency-domain signal.

[0015] The power conversion unit 202 converts the FFT result acquired from the FFT 201 into a power value. As described above, when a frequency axis value and an amplitude value corresponding to the frequency axis value are output as a signal converted into a frequency domain signal, the frequency domain conversion unit 102 has a configuration for converting the FFT result into an amplitude value instead of the power conversion unit 202.

[0016] The normalization unit 203 normalizes the frequency axis values ​​obtained from the FFT 201 and the power values ​​obtained from the power conversion unit 202. The normalization unit 203 may perform normalization processing to stabilize the learning coefficient in the downstream action value calculation unit 103, such as setting the average of various values ​​to 0 and the variance to 1, or fitting the values ​​within a range from a maximum value of 1 to a minimum value of 0. Note that the frequency domain transformation unit 102 may normalize only one of the frequency axis values ​​from the FFT 201 and the power values ​​from the power conversion unit 202.

[0017] FIG. 3 is a diagram illustrating an example of the operation of the frequency domain transform unit 102 included in the signal analyzing apparatus 100 according to the first embodiment. In the frequency domain transform unit 102, the FFT 201 can convert time-domain signals into frequency-domain signals for modulation schemes such as binary phase shift keying (BPSK), quadrature phase shift keying (QPSK), quadrature amplitude modulation (16QAM), 32QAM, and 64QAM, among a plurality of modulation schemes, into signals with similar signal shapes. Therefore, the signal analyzing apparatus 100 can reduce the number of input patterns in the configuration subsequent to the frequency domain transform unit 102, i.e., the action value calculation unit 103, thereby reducing the number of inputs and outputs handled by the action value calculation unit 103. As shown in FIG. 3 , the frequency domain transform unit 102 can also be applied to modulation schemes such as minimum shift keying (MSK) and frequency shift keying (FSK) in addition to the above modulation schemes.

[0018] The action value calculation unit 103 is configured with a neural network that receives as input the signal converted into a frequency domain signal from the frequency domain conversion unit 102. The output of the neural network is a plurality of actions that are candidates for adjustment in the frequency sampling rate adjustment unit 101, and a value for each action. That is, the action value calculation unit 103 receives as input the signal converted into a frequency domain signal from the frequency domain conversion unit 102, and outputs a plurality of actions for which the frequency or sampling rate of the received signal is adjusted in the frequency sampling rate adjustment unit 101, and a value for each action. The plurality of actions may include both the frequency adjustment amount and the sampling rate adjustment amount as shown in Example 1, or may include only one of the frequency adjustment amount and the sampling rate adjustment amount as shown in Examples 2 and 3, or may include a simple method such as finding the center of gravity of a spectrum, such as a rough frequency estimation method, as shown in Example 4.

[0019] (Example 1) "Action 1: Frequency +x%", "Action 2: Frequency -x%", "Action 3: Sampling rate +y%", "Action 4: Sampling rate -y%"

[0020] (Example 2) "Action 1: Frequency + x%", "Action 2: Frequency - x%", "Action 3: Frequency + 10x%", "Action 4: Frequency - 10x%"

[0021] (Example 3) "Action 1: Sampling rate +y%", "Action 2: Sampling rate -y%", "Action 3: Sampling rate +10y%, "Action 4: Sampling rate -10y%"

[0022] (Example 4) "Action 1: Frequency + x%", "Action 2: Frequency - x%", "Action 3: Frequency coarse estimation method 1"

[0023] In the above example, the adjustment amounts are the same in the positive and negative directions, such as +x% or -x% for the frequency and +y% or -y% for the sampling rate, but are not limited to this. １ %”, “Action 2: Frequency-x ２The amount of adjustment may be different in the positive and negative directions, such as "action 1: frequency x %". １ %”, “Action 2: Frequency x ２ "%" and "x" without showing + or -. １ " and "x ２ " may include instructions in both + and - directions. In addition, if the action decision unit 104 selected "Action 1: Frequency +x%" in the previous adjustment, the action value calculation unit 103 may select the same "Action 1: Frequency +x%" in the + direction in the next adjustment, but may select "Action 2: Frequency -x%" in the - direction. ２ The magnitude of the adjustment amount may be changed, for example, "%". In this way, the action value calculation unit 103 outputs to the action determination unit 104, as the multiple actions, at least two of the one or more actions for adjusting the frequency in the frequency sampling rate adjustment unit 101 and the one or more actions for adjusting the sampling rate.

[0024] The action value calculation unit 103 calculates the coefficients of the neural network in advance by simulation or the like. The neural network of the action value calculation unit 103 is trained in advance so that the subsequent action decision unit 104 will receive the information necessary to decide on an action, such as the larger the value value output to the action decision unit 104, the higher the value of taking the corresponding action. The action value calculation unit 103 also determines the amount of adjustment for each action in advance, i.e., when training the neural network.

[0025] The action determination unit 104 acquires multiple actions and values ​​from the action value calculation unit 103. The action determination unit 104 determines an action to be performed by the frequency sampling rate adjustment unit 101 from the multiple actions based on the values, i.e., determines an action for the received signal input to the signal analysis device 100, and outputs the determined action to the frequency sampling rate adjustment unit 101. Regarding the method of determining an action, the action determination unit 104 may select one action indicated as having the highest value as shown in Example 1, may select one action with high value for each frequency and sampling rate as shown in Example 2, or may select probabilistically based on the ratio of the output values ​​as shown in Example 3.

[0026] (Example 1) "Action 1: Frequency + x% Value = 1.0", "Action 2: Frequency - x% Value = 0.2", "Action 3: Sampling rate + y% Value = 0.3", "Action 4: Sampling rate - y% Value = 0.9" In this case, the action decision unit 104 selects action 1.

[0027] (Example 2) "Action 1: Frequency + x% Value = 1.0", "Action 2: Frequency - x% Value = 0.2", "Action 3: Sampling rate + y% Value = 0.3", "Action 4: Sampling rate - y% Value = 0.9" In this case, the action decision unit 104 selects actions 1 and 4.

[0028] (Example 3) "Action 1: Frequency + x% Value = 1.0", "Action 2: Frequency - x% Value = 0.2", "Action 3: Sampling rate + y% Value = 0.3", "Action 4: Sampling rate - y% Value = 0.9" In this case, the action decision unit 104 selects action 1 with a probability of 1.0 / (1.0 + 0.2 + 0.3 + 0.9), action 2 with a probability of 0.2 / (1.0 + 0.2 + 0.3 + 0.9), action 3 with a probability of 0.3 / (1.0 + 0.2 + 0.3 + 0.9), and action 4 with a probability of 0.9 / (1.0 + 0.2 + 0.3 + 0.9). Note that if the action decision unit 104 selected an option with a low probability in the previous selection, it may select an option with a high probability in the next selection.

[0029] In this way, the behavior determining unit 104 determines one behavior from a plurality of behaviors based on the value, or determines one behavior from one or more behaviors for adjusting the frequency and one behavior from one or more behaviors for adjusting the sampling rate based on the value.

[0030] The modulation scheme estimation unit 105 estimates the modulation scheme of a signal whose frequency and / or sampling rate have been adjusted by the frequency sampling rate adjustment unit 101. The modulation scheme estimation method includes, but is not limited to, a method using a neural network as described in Japanese Patent No. 7130179. The modulation scheme estimation unit 105 is designed in advance with certain assumptions in mind, such as the frequency range of the received signal and the ratio of the sampling rate to the symbol rate of the received signal, i.e., the oversampling rate. Therefore, the frequency sampling rate adjustment unit 101, the frequency domain conversion unit 102, the action value calculation unit 103, and the action decision unit 104 perform processing multiple times to adjust the frequency and sampling rate of the received signal so that they fall within the ranges expected by the modulation scheme estimation unit 105, thereby enabling the modulation scheme to be correctly estimated.

[0031] The demodulation unit 106 demodulates the received signal input to the signal analyzing device 100 based on the signal whose frequency and / or sampling rate have been adjusted by the frequency sampling rate adjustment unit 101 and the estimation result of the modulation scheme estimated by the modulation scheme estimation unit 105. Generally, in wireless communication, demodulation is possible if the frequency, symbol rate, modulation scheme, etc. of the target signal are known. Therefore, by using the estimation results from the previous stages, the demodulation unit 106 can demodulate the received signal whose communication parameters are unknown.

[0032] In the signal analyzing device 100, the adjustment of the frequency and sampling rate may not be complete immediately after receiving a received signal, and it is possible that the estimation accuracy of the modulation scheme and the demodulation accuracy may be poor. Therefore, in the signal analyzing device 100, the frequency sampling rate adjusting unit 101 may manage an adjustment completion flag to control the operation of the modulation scheme estimating unit 105 and the demodulating unit 106. For example, when the processes of the frequency sampling rate adjusting unit 101, the frequency domain transforming unit 102, the action value calculating unit 103, and the action determining unit 104 are repeated multiple times and a predetermined number of adjustments are performed, the frequency sampling rate adjusting unit 101 determines that the frequency and sampling rate have become approximately constant and sets the adjustment completion flag. While the adjustment completion flag is not set, the frequency sampling rate adjusting unit 101 does not output the signal after the adjustment of the frequency and sampling rate to the modulation scheme estimating unit 105. However, when the adjustment completion flag is set, the frequency sampling rate adjusting unit 101 outputs the signal after the adjustment of the frequency and sampling rate to the modulation scheme estimating unit 105.

[0033] 4 is a flowchart showing the operation of the signal analyzing device 100 according to the first embodiment. In the signal analyzing device 100, the frequency sampling rate adjusting unit 101 adjusts at least one of the frequency and the sampling rate of the received signal in accordance with the action determined by the action determining unit 104 (step S1). Note that, as described above, the frequency sampling rate adjusting unit 101 may omit step S1 if the action determining unit 104 has not yet determined an action, such as immediately after the signal analyzing device 100 is started up. The frequency domain transforming unit 102 converts the received signal adjusted by the frequency sampling rate adjusting unit 101 from a time domain signal to a frequency domain signal (step S2). The action value calculating unit 103 calculates multiple actions that are candidates for adjustment by the frequency sampling rate adjusting unit 101 and the value of each action, based on the converted frequency domain signal obtained from the frequency domain transforming unit 102 (step S3).

[0034] The action determination unit 104 determines an action based on the multiple actions and values ​​acquired from the action value calculation unit 103 (step S4). The action determination unit 104 selects from the multiple actions, i.e., outputs the determined action to the frequency sampling rate adjustment unit 101. The modulation scheme estimation unit 105 estimates the modulation scheme of the received signal acquired from the frequency sampling rate adjustment unit 101 and adjusted by the frequency sampling rate adjustment unit 101 (step S5). The demodulation unit 106 demodulates the received signal based on the received signal whose frequency and sampling rate have been adjusted and the estimated modulation scheme (step S6). Note that the signal analysis device 100 may use an adjustment completion flag as described above so as not to perform the operations of steps S5 and S6 while the adjustment completion flag is not set. The signal analysis device 100 repeatedly performs the operations from step S1 to step S4 while the adjustment completion flag is not set.

[0035] Next, the hardware configuration of the signal analyzing device 100 will be described. In the signal analyzing device 100, the frequency sampling rate adjusting unit 101, the frequency domain transforming unit 102, the action value calculating unit 103, the action deciding unit 104, the modulation scheme estimating unit 105, and the demodulating unit 106 are realized by processing circuits. The processing circuits may be a processor and memory that executes a program stored in a memory, or may be dedicated hardware. The processing circuits are also called control circuits.

[0036] FIG. 5 is a diagram illustrating an example of the configuration of a processing circuit 900 that implements the signal analyzing device 100 according to the first embodiment, where the processing circuit is configured with a processor 901 and a memory 902. The processing circuit 900 illustrated in FIG. 5 is a control circuit and includes a processor 901 and a memory 902. When the processing circuit 900 is configured with the processor 901 and the memory 902, each function of the processing circuit 900 is implemented by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 902. The processor 901 reads and executes the program stored in the memory 902 to implement each function of the processing circuit 900. That is, the processing circuit 900 includes the memory 902 for storing a program that results in the processing of the signal analyzing device 100 being executed. This program can also be said to be a program that causes the signal analyzing device 100 to execute each function implemented by the processing circuit 900. This program may be provided by a storage medium on which the program is stored, or by other means such as a communication medium.

[0037] The above program can also be said to be a program executed by the signal analysis device 100: an adjustment step in which the frequency sampling rate adjustment unit 101 adjusts at least one of the frequency and the sampling rate of the first received signal and outputs it as a second received signal; a frequency domain conversion step in which the frequency domain conversion unit 102 converts the second received signal from a time domain signal to a frequency domain signal and outputs the frequency domain signal as a third received signal; an action value calculation step in which the action value calculation unit 103, which is composed of a neural network, receives the third received signal as input and outputs multiple actions for which the frequency sampling rate adjustment unit 101 adjusts the frequency or sampling rate of the first received signal and a value for each action; and an action determination step in which the action determination unit 104 determines an action to be performed in the frequency sampling rate adjustment unit 101 from the multiple actions based on the value and outputs the determined action to the frequency sampling rate adjustment unit 101.

[0038] Here, the processor 901 is, for example, a CPU (Central Processing Unit), a processing device, an arithmetic unit, a microprocessor, a microcomputer, or a DSP (Digital Signal Processor), etc. Furthermore, the memory 92 is, for example, a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable ROM), or an EEPROM (Electrically EPROM), a magnetic disk, a flexible disk, an optical disk, a compact disk, a minidisk, or a DVD (Digital Versatile Disc).

[0039] FIG. 6 is a diagram illustrating an example of a processing circuit 903 that implements the signal analyzing device 100 according to the first embodiment and is configured with dedicated hardware. The processing circuit 903 illustrated in FIG. 6 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof. The processing circuit may be partially implemented with dedicated hardware and partially implemented with software or firmware. In this way, the processing circuit can implement each of the above-described functions using dedicated hardware, software, firmware, or a combination thereof.

[0040] As described above, according to this embodiment, in the signal analyzing device 100, the frequency sampling rate adjustment unit 101 adjusts at least one of the frequency and sampling rate of the received signal in the time domain in accordance with the action determined by the action determining unit 104. The frequency domain conversion unit 102 converts the signal adjusted by the frequency sampling rate adjustment unit 101 from a time domain signal to a frequency domain signal. The action value calculation unit 103, which is configured as a neural network, receives the signal converted to the frequency domain by the frequency domain conversion unit 102 and outputs multiple actions that are candidates for adjustment by the frequency sampling rate adjustment unit 101 and the value of each action. The action determining unit 104 determines the action to be taken by the frequency sampling rate adjustment unit 101 based on the value. In this way, by repeating a series of operations from the frequency sampling rate adjustment unit 101 to the action determining unit 104, the signal analyzing device 100 can estimate the modulation method of the received signal in the modulation method estimation unit 105 and adjust the frequency and sampling rate of the received signal to an extent that the received signal can be demodulated by the demodulation unit 106. The signal analyzing device 100 can support a large number of modulation methods while suppressing an increase in learning time and a deterioration in estimation accuracy.

[0041] Generally, when supporting multiple modulation schemes, the learning time increases according to the number of modulation schemes, and the large number of input / output correspondences means that the neural network coefficients do not converge well during learning, which can result in a deterioration in estimation accuracy. In contrast, the signal analysis device 100 of this embodiment converts the received signal into the frequency domain and inputs it into the neural network, which often results in different modulation schemes being represented by the same frequency spectrum. This reduces the number of input / output correspondences, thereby reducing the learning time and improving accuracy. Furthermore, the signal analysis device 100 of this embodiment performs frequency adjustment in the time domain before converting to the frequency domain, so it can support fine frequency shifts without relying on the resolution of the conversion to the frequency domain.

[0042] Furthermore, in the case of estimating and decoding the modulation scheme of a received signal, if a frequency estimation value is directly output as in the prior art disclosed in Patent Document 1, when the output value deviates significantly from the true value, for example, when the SNR (Signal to Noise Ratio) drops momentarily, additional determination processing, correction processing, etc. is required. In contrast, the signal analysis device 100 of the present embodiment treats the output of the neural network as "action" and "value," and places restrictions on the adjustment of communication parameters, such as "adjusting the frequency or symbol rate by a specified value," thereby making it possible to suppress performance degradation due to momentary deterioration of the communication environment.

[0043] Furthermore, the signal analyzing device 100 of this embodiment can adjust the sampling rate of the received signal to convert it into an oversampling rate, which is the ratio between the sampling rate and the symbol rate assumed during learning, thereby enabling operation in the modulation scheme estimation unit 105 and demodulation unit 106 at the subsequent stages.

[0044] Second Embodiment In a second embodiment, a case will be described in which the estimation result of the modulation scheme estimation unit 105 and the demodulation result of the demodulation unit 106 are input to the neural network of the action value calculation unit 103.

[0045] 7 is a diagram showing an example of the configuration of a signal analyzing device 100a according to embodiment 2. The signal analyzing device 100a is a device that is installed in a receiving device (not shown) that receives a received signal whose communication parameters are unknown, and analyzes the received signal whose communication parameters are unknown. The signal analyzing device 100a of embodiment 2 shown in FIG. 7 is obtained by adding an input signal generating unit 107 to the signal analyzing device 100 of embodiment 1 shown in FIG.

[0046] In the second embodiment, the modulation scheme estimation unit 105 outputs an estimation result of the modulation scheme to the input signal generation unit 107. The modulation scheme estimation unit 105 may output the modulation scheme with the highest probability and its probability value, as shown in Example 1, or may output a plurality of modulation schemes and the accuracy for each modulation scheme, as shown in Example 2. Note that the modulation scheme estimation unit 105 may use the same method of estimating the modulation scheme as in the first embodiment, with only the number of output destinations for the estimation result increasing.

[0047] (Example 1) "QPSK 80%"

[0048] (Example 2) "QPSK 70%", "BPSK 20%", "FSK 10%"

[0049] In the second embodiment, the demodulation unit 106 outputs the demodulation result to the input signal generation unit 107. The demodulation unit 106 outputs, for example, a constellation, an eye pattern, an EVM (Error Vector Magnitude), a bit pattern, etc. as the demodulation result. Note that the demodulation method of the demodulation unit 106 may be the same as that of the first embodiment, except that the number of output destinations of the demodulation result has increased.

[0050] The input signal generation unit 107 generates a signal to be output to the action value calculation unit 103, using the estimation result acquired from the modulation scheme estimation unit 105 and the demodulation result acquired from the demodulation unit 106. Fig. 8 is a diagram showing an example configuration of the input signal generation unit 107 included in the signal analysis device 100a according to Embodiment 2. The input signal generation unit 107 includes a maximum probability selection unit 301, a probability difference calculation unit 302, a constellation generation unit 303, an eye pattern generation unit 304, an EVM calculation unit 305, a bit pattern analysis unit 306, and a normalization unit 307.

[0051] The maximum probability selection unit 301 outputs "QPSK 80%" when the estimation result acquired from the modulation scheme estimation unit 105 is like Example 1 described in the description of the modulation scheme estimation unit 105 of the second embodiment, and selects and outputs "QPSK 70%" when the estimation result is like Example 2 described in the description of the modulation scheme estimation unit 105 of the second embodiment.

[0052] When the estimation result acquired from the modulation scheme estimation unit 105 is like Example 2 described in the description of the modulation scheme estimation unit 105 of the second embodiment, the probability difference calculation unit 302 uses "QPSK 70%" and "FSK 10%" to calculate and output a variance value or the like that indicates how far the probabilities of a plurality of modulation schemes differ from each other.

[0053] If the demodulation result acquired from the demodulation unit 106 is a constellation, the constellation generation unit 303 generates and outputs an image of the constellation.

[0054] If the demodulation result obtained from the demodulation unit 106 is an eye pattern, the eye pattern generation unit 304 generates and outputs an image of the eye pattern.

[0055] If the demodulation result obtained from the demodulation unit 106 is an EVM, the EVM calculation unit 305 may output it as is, or if the demodulation result obtained from the demodulation unit 106 is a constellation, the EVM calculation unit 305 may calculate the EVM from the constellation and output it.

[0056] If the demodulation result obtained from the demodulation unit 106 is a bit pattern, the bit pattern analysis unit 306 calculates and outputs the average value, variance value, etc., which represent the randomness of the bit pattern.

[0057] The normalization unit 307 normalizes the outputs from the maximum probability selection unit 301 to the bit pattern analysis unit 306. The normalization unit 307 may perform normalization processing to stabilize the learning coefficient in the downstream action value calculation unit 103, such as setting the average of various values ​​to 0 and the variance to 1, or keeping the values ​​within a range from a maximum value of 1 to a minimum value of 0. Note that the normalization unit 307 may normalize only a portion of the outputs from the maximum probability selection unit 301 to the bit pattern analysis unit 306.

[0058] In this way, the input signal generation unit 107 outputs at least one of the probability of the most probable modulation scheme based on the estimation result of the modulation scheme estimation unit 105, the difference between the probabilities of the multiple modulation schemes based on the estimation result of the modulation scheme estimation unit 105, the constellation based on the demodulation result of the demodulation unit 106, the eye pattern based on the demodulation result of the demodulation unit 106, the EVM based on the demodulation result of the demodulation unit 106, and the bit pattern based on the demodulation result of the demodulation unit 106 to the action value calculation unit 103. The action value calculation unit 103 further receives information from the input signal generation unit 107 as input, and outputs multiple actions and values ​​to the action decision unit 104.

[0059] The hardware configuration of the signal analyzing device 100a will be described. In the signal analyzing device 100a, the input signal generating unit 107 is also realized by a processing circuit. The processing circuit may be a processor and memory that executes a program stored in a memory, or may be dedicated hardware.

[0060] As described above, according to this embodiment, in the signal analyzing device 100a, the input signal generating unit 107 uses the estimation result obtained from the modulation scheme estimating unit 105 and the demodulation result obtained from the demodulating unit 106 to generate a signal to be output to the action value calculating unit 103. The action value calculating unit 103, which is configured as a neural network, receives as input the signal converted into the frequency domain by the frequency domain converting unit 102 and the signal generated by the input signal generating unit 107, and outputs a plurality of actions that are candidates for adjustment by the frequency sampling rate adjusting unit 101 and the value of each action. The action determining unit 104 determines an action to be taken by the frequency sampling rate adjusting unit 101 based on the value.

[0061] As a result, by repeating a series of operations from the frequency sampling rate adjustment unit 101 to the demodulation unit 106, the signal analyzing device 100a can handle not only the received signal but also the estimation result from the subsequent modulation scheme estimation unit 105 and the demodulation result from the demodulation unit 106, thereby improving the accuracy of the action and value calculated by the action value calculation unit 103. Since the action is intended to improve the performance of the subsequent processing, the results of this subsequent processing, i.e., the estimation result from the modulation scheme estimation unit 105 and the demodulation result from the demodulation unit 106, are fed back. By also using the signal generated by the input signal generation unit 107, the signal analyzing device 100a can improve the accuracy of the action and value calculated by the action value calculation unit 103 compared to the signal analyzing device 100 of the first embodiment.

[0062] Third Embodiment In a third embodiment, a case will be described in which a reward based on the estimation result of the modulation scheme estimation unit 105 and the demodulation result of the demodulation unit 106 is input to the neural network of the action value calculation unit 103.

[0063] 9 is a diagram showing an example of the configuration of a signal analyzing device 100b according to embodiment 3. The signal analyzing device 100b is a device that is mounted on a receiving device (not shown) that receives a received signal with unknown communication parameters and analyzes the received signal with unknown communication parameters. The signal analyzing device 100b of embodiment 3 shown in FIG. 9 is configured by adding a reward calculation unit 108 to the signal analyzing device 100a of embodiment 2 shown in FIG. 7.

[0064] In the third embodiment, the modulation scheme estimation unit 105 outputs the estimation result of the modulation scheme to the remuneration calculation unit 108. The modulation scheme estimation unit 105 only has more output destinations for the estimation result, and the modulation scheme estimation method itself and the content of the estimation result may be the same as in the second embodiment.

[0065] In the third embodiment, the demodulation unit 106 outputs the demodulation result to the reward calculation unit 108. The demodulation unit 106 only has more output destinations for the demodulation result, and the demodulation method itself and the contents of the demodulation result may be the same as in the second embodiment.

[0066] The reward calculation unit 108 calculates a reward for updating the coefficients of the neural network that constitutes the action value calculation unit 103 based on the action selection result obtained from the action decision unit 104, the modulation method estimation result obtained from the modulation method estimation unit 105, and the demodulation result obtained from the demodulation unit 106.

[0067] For example, the reward calculation unit 108 outputs a high reward, such as a reward "1," if, as a result of taking a certain action to adjust the frequency or sampling rate based on the action selection result obtained from the action determination unit 104, the frequency becomes equal to or less than the desired frequency error, or the sampling rate becomes within the desired rate range. Conversely, if, as a result of taking a certain action to adjust the frequency or sampling rate, the result significantly deviates from the true value, the reward calculation unit 108 outputs a low reward, such as a reward "-1." If, as a result of taking a certain action to adjust the frequency or sampling rate, the frequency or sampling rate does not become the desired state or significantly deviate from the true value, the reward calculation unit 108 outputs a reward "0" as a result of nothing. While this method is difficult when the specifications of the received signal are unknown, it is effective when learning is performed by inputting a signal with known specifications, such as a test signal.

[0068] Furthermore, the reward calculation unit 108 outputs a high reward if the probability value of the most probable modulation scheme exceeds a certain threshold, based on the estimation result of the modulation scheme acquired from the modulation scheme estimation unit 105, and outputs a low reward if the probability value of the most probable modulation scheme is equal to or less than a certain threshold. Based on the estimation result of the modulation scheme acquired from the modulation scheme estimation unit 105, the reward calculation unit 108 outputs a high reward if a variance value representing the degree of difference in probabilities between multiple modulation schemes exceeds a certain threshold, and outputs a low reward if the variance value representing the degree of difference in probabilities between multiple modulation schemes is equal to or less than a certain threshold.

[0069] Furthermore, the reward calculation unit 108 outputs a high reward if the variation in the center of the eye pattern is within a specified range, i.e., small, based on the demodulation result acquired from the demodulation unit 106, and outputs a low reward if the variation in the center of the eye pattern is outside the specified range, i.e., large, based on the demodulation result acquired from the demodulation unit 106. The reward calculation unit 108 outputs a high reward if the EVM is equal to or less than a certain threshold, and outputs a low reward if the EVM exceeds the certain threshold.

[0070] As described above, the reward calculation unit 108 may determine the reward based on each of the results of the action selection result obtained from the action determination unit 104, the modulation method estimation result obtained from the modulation method estimation unit 105, and the demodulation result obtained from the demodulation unit 106, or may determine the reward based on two or more results.

[0071] Compared to the action value calculation unit 103 of the second embodiment, the action value calculation unit 103 further acquires a reward output from the reward calculation unit 108. The action value calculation unit 103 updates, i.e., learns, each coefficient of the neural network that calculates the action value based on the reward calculated by the reward calculation unit 108 and its own output results. For example, the action value calculation unit 103 determines that the action and value are correct if the obtained reward is high, and that the action and value are incorrect if the obtained reward is low, and learns to output an action and value that will obtain a higher reward. This type of learning is generally a framework called reinforcement learning, and can be performed using a general-purpose algorithm. In the first and second embodiments, the action value calculation unit 103 was assumed to calculate the neural network coefficients in advance using simulations or the like. However, by adding the reward calculation unit 108 as in the third embodiment, it is also possible to learn from received signals.

[0072] In this way, the reward calculation unit 108 calculates a reward for an action based on the modulation method estimation result obtained from the modulation method estimation unit 105 and the demodulation result obtained from the demodulation unit 106, and outputs the calculated reward to the action value calculation unit 103. The action value calculation unit 103 performs learning by updating each coefficient of the neural network based on the multiple actions and values ​​output to the action decision unit 104 and the reward obtained from the reward calculation unit 108, so that a high reward can be obtained from the reward calculation unit 108.

[0073] The hardware configuration of the signal analyzing device 100b will be described. In the signal analyzing device 100b, the reward calculation unit 108 is also realized by a processing circuit. The processing circuit may be a processor and memory that executes a program stored in a memory, or may be dedicated hardware.

[0074] As described above, according to this embodiment, in the signal analyzing device 100b, the reward calculation unit 108 calculates a reward to be given to the action value calculation unit 103 based on the action selection result acquired from the action determination unit 104, the modulation method estimation result acquired from the modulation method estimation unit 105, and the demodulation result acquired from the demodulation unit 106. The action value calculation unit 103 performs learning so as to obtain a high reward from the reward calculation unit 108. This enables the signal analyzing device 100b to perform not only offline learning using prior simulations, but also online learning from received signals, and to estimate a modulation method adapted to the environment and demodulate the received signal even when the characteristics of the target received signal change.

[0075] The configurations shown in the above embodiments are merely examples, and may be combined with other known technologies, or different embodiments may be combined with each other. It is also possible to omit or modify parts of the configurations as long as they do not deviate from the gist of the invention.

[0076] 100, 100a, 100b Signal analysis device, 101 Frequency sampling rate adjustment unit, 102 Frequency domain conversion unit, 103 Action value calculation unit, 104 Action decision unit, 105 Modulation method estimation unit, 106 Demodulation unit, 107 Input signal generation unit, 108 Reward calculation unit, 201 FFT, 202 Power conversion unit, 203, 307 Normalization unit, 301 Maximum probability selection unit, 302 Probability difference calculation unit, 303 Constellation generation unit, 304 Eye pattern generation unit, 305 EVM calculation unit, 306 Bit pattern analysis unit, 900, 903 Processing circuit, 901 Processor, 902 Memory.

Claims

1. an adjustment unit that adjusts at least one of a frequency and a sampling rate of the first received signal and outputs the adjusted signal as a second received signal; a frequency domain transform unit that transforms the second received signal from a time domain signal to a frequency domain signal and outputs the frequency domain signal as a third received signal; an action value calculation unit that is configured with a neural network, receives the third received signal as an input, and outputs a plurality of actions for adjusting the frequency or the sampling rate of the first received signal in the adjustment unit and a value for each action; a behavior determining unit that determines an action to be performed by the adjustment unit from the plurality of actions based on the value and outputs the determined action to the adjustment unit; A signal analysis device comprising:

2. the action value calculation unit outputs to the action determination unit, as the plurality of actions, at least two of the one or more actions of adjusting the frequency and the one or more actions of adjusting the sampling rate; 2. The signal analyzing device according to claim 1 .

3. The action determination unit determines one action from the plurality of actions based on the value.

3. The signal analyzing device according to claim 2.

4. The action determination unit determines one action from one or more actions to adjust the frequency and determines one action from one or more actions to adjust the sampling rate based on the value.

3. The signal analyzing device according to claim 2.

5. the frequency domain transform unit outputs, as the third received signal, a value on a frequency axis and a power value or an amplitude value corresponding to the value on the frequency axis.

5. The signal analyzing device according to claim 1, wherein the signal analyzing device comprises: a first input section;

6. a modulation scheme estimation unit that estimates a modulation scheme of the second received signal; a demodulation unit that demodulates the first received signal based on the second received signal and an estimation result of the modulation scheme estimated by the modulation scheme estimation unit; 5. The signal analyzing device according to claim 1, further comprising:

7. an input signal generation unit which outputs to said action value calculation unit at least one of the probability of the most probable modulation scheme based on said estimation result, the difference between the probabilities of each of the plurality of modulation schemes based on said estimation result, a constellation based on the demodulation result of said demodulation unit, an eye pattern based on said demodulation result, an Error Vector Magnitude based on said demodulation result, and a bit pattern based on said demodulation result; Equipped with the action value calculation unit further receives information from the input signal generation unit and outputs the plurality of actions and the values ​​to the action determination unit; 7. The signal analyzing device according to claim 6.

8. a reward calculation unit that calculates a reward for the action based on the estimation result and the demodulation result of the demodulation unit, and outputs the reward to the action value calculation unit; Equipped with the action value calculation unit performs learning by updating each coefficient of the neural network based on the plurality of actions, the values, and the rewards output to the action determination unit so as to obtain a high reward from the reward calculation unit; 7. The signal analyzing device according to claim 6.

9. A control circuit for controlling a signal analyzing device, adjusting at least one of a frequency and a sampling rate of the first received signal and outputting the adjusted signal as a second received signal; converting the second received signal from a time domain signal to a frequency domain signal, and outputting the signal after the conversion to the frequency domain signal as a third received signal; receiving the third received signal as an input, and outputting a plurality of actions for adjusting the frequency or the sampling rate of the first received signal and a value for each action; determining an action for the first received signal from the plurality of actions based on the value, and outputting the determined action; A control circuit that causes the signal analyzer to execute the above.

10. A storage medium storing a program for controlling a signal analyzing device, The program is adjusting at least one of a frequency and a sampling rate of the first received signal and outputting the adjusted signal as a second received signal; converting the second received signal from a time domain signal to a frequency domain signal, and outputting the signal after the conversion to the frequency domain signal as a third received signal; receiving the third received signal as an input, and outputting a plurality of actions for adjusting the frequency or the sampling rate of the first received signal and a value for each action; determining an action for the first received signal from the plurality of actions based on the value, and outputting the determined action; A storage medium for causing the signal analysis device to carry out the above-mentioned.

11. an adjustment step in which an adjustment unit adjusts at least one of a frequency and a sampling rate of the first received signal and outputs the adjusted signal as a second received signal; a frequency domain transform step in which a frequency domain transform unit transforms the second received signal from a time domain signal to a frequency domain signal and outputs the signal after the transformation into the frequency domain signal as a third received signal; an action value calculation step in which an action value calculation unit configured with a neural network receives the third received signal as an input, and outputs a plurality of actions for which the frequency or the sampling rate of the first received signal is adjusted in the adjustment unit, and a value for each action; an action determination step in which an action determination unit determines an action to be performed by the adjustment unit from the plurality of actions based on the value, and outputs the determined action to the adjustment unit; A signal analysis method comprising: