Spectrum sensing system and method for discrete shared frequency band

Abstract

The invention relates to a spectrum sensing system and method for a discrete shared frequency band, and belongs to the technical field of wireless communication. The system comprises a base station, an edge computing device and a plurality of sensing terminals, the plurality of sensing terminals collect spectrum data, perform spectrum sensing and send sensing results to the base station, and the base station forwards the sensing results to the edge computing device for fusion processing of the plurality of sensing results. And the edge computing device fuses the plurality of sensing results, predicts the spectrum occupation condition of a plurality of subsequent time steps through a pre-trained binary prediction model, and sends a spectrum prediction result to the base station, and the base station screens idle spectrum frequency points according to the spectrum prediction result to communicate with the sensing terminal. According to the method, discrete shared frequency band sensing is carried out based on the system, and the method can adapt to the conditions that the signal coverage range of discrete frequency bands is wide, an unknown communication system exists and real-time spectrum analysis is needed.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communications and relates to a spectrum sensing system and method for discrete shared frequency bands. Background Art

[0002] Discrete shared frequency band refers to the 233-235MHz communication frequency band. This frequency band is used for communication networks in various industries, such as electricity, gas, civil air defense, and water services. Currently, LTE 230 and IoT 230 communication systems are mainly used. In 2018, the Ministry of Industry and Information Technology issued the "Notice of the Ministry of Industry and Information Technology on Adjusting the Frequency Use Planning of Wireless Data Transmission Systems in the 223-235MHz Frequency Band", allowing various units to integrate discrete frequency points in the 223-226MHz and 229-233MHz frequency bands to use them in a continuous spectrum, supporting multiple different services to transmit data in a common frequency band in a time / frequency division multiplexing manner.

[0003] As 5G RedCap technology matures and is commercialized, it is expected that this technology and its network will be deployed in the 233-235MHz communication frequency band in the future, further improving the spectrum utilization rate of this frequency band. However, since discrete shared frequency bands are difficult to manage, different communication services may adopt inconsistent communication systems and communication technologies, illegal data transmission stations randomly occupy some frequency points, and low-frequency high-power devices have serious interference. These problems will affect the deployment and implementation of communication networks. To address these problems, it is necessary to apply spectrum sensing technology in the field of cognitive radio. On the one hand, spectrum sensing technology can realize interference detection of discrete frequency points, and taking reasonable avoidance methods can improve the reliability of communication systems. On the other hand, spectrum sensing technology helps to discover idle spectrum, and using it can improve the effectiveness of communication.

[0004] At present, narrowband spectrum sensing technology performs signal detection on a certain frequency band, which can achieve good detection sensitivity and detect user signals within the frequency band under low signal-to-noise ratio conditions. Figure 1 As shown, the principle is to collect data in a certain frequency band, and then calculate a detection statistic that can reflect the characteristics of the user signal through a corresponding algorithm, and then compare it with a detection threshold to determine whether there is a user signal in the frequency band.

[0005] Classic spectrum sensing methods include energy detection, cyclostationary detection, matched filtering, etc. These spectrum sensing methods require some prior information to determine the decision threshold, otherwise the spectrum sensing performance will be reduced. Later, relevant researchers proposed some blind spectrum sensing methods that do not require prior information, such as eigenvalue detection and covariance matrix detection. Blind sensing methods can further improve the system stability of spectrum sensing, but the complexity of its perception will increase accordingly. For discrete shared frequency bands, the existing spectrum sensing technology has the following two problems and cannot perform effective spectrum sensing.

[0006] First: The signal spectrum characteristics of discrete shared frequency bands require appropriate spectrum sensing technology. Existing spectrum sensing technology is suitable for detecting wideband signals, collecting signals with large amounts of data, or sensing known system signals. In discrete shared frequency bands, the frequency points are clearly divided, with a total of 480 frequency points, and each frequency point has a bandwidth of 25kHz. Users in this frequency band use spectrum resources in units of frequency points. User signals have the characteristics of short symbol period, discrete frequency points, and narrow signal bandwidth. It is difficult to accurately collect a large amount of user data when sensing each frequency point in the frequency band, so the existing spectrum sensing technology is not applicable.

[0007] Second: Communication systems that use discrete shared frequency bands require appropriate spectrum sensing technology. Due to the long transmission distance of user signals in this frequency band, the wide coverage area, and the serious interference from low-frequency and high-power devices, the existing spectrum sensing technology needs to further improve the sensitivity of sensing. Currently, 5G communication networks support the use of multi-antenna technology to achieve long-distance, high-reliability transmission, and there is no shortage of communication systems that use multi-antenna technology in discrete shared frequency bands. Applying sensing algorithms based on the signal correlation characteristics between multiple antennas will help improve the detection sensitivity of the sensing system. There is no shortage of sensing algorithms based on multiple antennas in existing spectrum sensing technologies, but they mainly remain at the theoretical level, and there is little research on the actual use of several antennas and their related parameter settings. In addition, discrete shared frequency bands are mainly used in IoT services, and the communication systems used need to consider cost issues.

[0008] In summary, based on the characteristics of discrete frequency distribution in discrete shared frequency bands and the possible random appearance of interfering radio stations, as well as the fact that multiple transmission antennas are allowed to be used in the communication system of this frequency band, it is necessary to redesign the perception system and method. Summary of the invention

[0009] In view of this, an object of the present invention is to provide a spectrum sensing system and method for discrete shared frequency bands.

[0010] In order to achieve the above object, the present invention provides the following technical solutions:

[0011] A spectrum sensing system for a discrete shared frequency band, the system comprising: at least one base station, at least one edge computing device and a plurality of sensing terminals, each sensing terminal being provided with at least two receiving antennas, a dual-channel amplification and filtering circuit, a dual-channel digital-to-analog converter and an internal data processing module, the internal data processing module comprising an in-phase orthogonal signal demodulation module, a secondary mixing module, a filtering sampling module, a fast Fourier transform module, a frequency selection grouping module, a spectrum sensing module and a sensing result sending module connected in sequence, wherein:

[0012] Multiple sensing terminals collect spectrum data and perform spectrum sensing, and send the sensing results to the base station. The base station then forwards the sensing results to the edge computing device for fusion processing of multiple sensing results. The edge computing device fuses multiple sensing results and predicts the spectrum occupancy of subsequent multiple time steps through a pre-trained binary prediction model, and sends the spectrum prediction results to the base station. The base station selects idle spectrum frequency points based on the spectrum prediction results to communicate with the sensing terminal.

[0013] Each sensing terminal collects two channels of RF data. The RF data comes from two antennas receiving RF signals. After amplification and filtering by a dual-channel amplification and filtering circuit, a dual-channel digital-to-analog converter is used to sample the time-domain RF signal. The in-phase orthogonal signal demodulation module performs IQ demodulation on the time-domain RF signal to obtain IQ data. The IQ data is transmitted to the secondary mixing module for secondary mixing to divide the signal spectrum into two sub-bands. The fast Fourier transform module performs fast Fourier transform on the signals of the two sub-bands to obtain spectrum data of the two antennas. The spectrum data is grouped according to frequency points through the spectrum sensing module, and then spectrum sensing is performed by the spectrum sensing module. Finally, the spectrum sensing result and sensing time information are sent to the base station through the sensing result sending module.

[0014] Furthermore, the in-phase orthogonal signal demodulation module in each sensing terminal performs IQ demodulation on the discrete shared frequency band RF signal to obtain IQ data, wherein the in-phase orthogonal signal demodulation module respectively combines the two-way RF signal data of antenna 1 and antenna 2 collected by the dual-channel mode converter with the primary mixing carrier frequency f c1 Multiply the two signals to get the collected IQ data, and mix the carrier frequency f c1 Denoted as local carrier 1, it includes frequency f c The I carrier and Q carrier of the c1 n s / f s1 ) and -sin(2πf c1 n s / f s1 ), where n s =1,2,…,N s , N sIndicates the length of data collected in one sampling cycle of the ADC, f s1 is the ADC sampling frequency; the IQ demodulation module outputs IQ data and generates baseband signal IQ data after filtering and sampling, which is recorded as baseBandSignal[m], where m represents the antenna number and its value is 1,2.

[0015] Furthermore, the secondary mixing module of each sensing terminal performs secondary mixing on the baseband signal in the obtained IQ data baseBandSignal, and the secondary mixing carrier frequency f c21 , f c22 , denoted as local carrier 2, where

[0016] The secondary mixing module mixes the baseBandSignal of antenna 1 with the two carriers f of local carrier 2. c21 , f c22 Multiply them respectively to obtain the mixed IQ data of sub-band 1 and sub-band 2 corresponding to antenna 1, recorded as mixFreqSignal[m1][1] and mixFreqSignal[m1][2], where m1 represents the antenna number and its value is 1;

[0017] The secondary mixing module mixes the baseBandSignal of antenna 2 with the two carriers f of local carrier 2. c21 , f c22 Multiply them respectively to obtain the mixed IQ data of sub-band 1 and sub-band 2 corresponding to antenna 2, recorded as mixFreqSignal[m2][1] and mixFreqSignal[m2][2], where m2 represents the antenna number and its value is 2;

[0018] Among them, local carriers 2 are:

[0019]

[0020] Where n1 = 1, 2, ..., N1, N1 is the baseband signal IQ data length, f s2 is the baseband digital sampling frequency, which is equal to N1 divided by the sampling period.

[0021] Furthermore, the filter sampling module of each sensing terminal includes a filter sampling module 1 and a filter sampling module 2, which are respectively arranged at the output ends of the IQ demodulation module and the secondary mixing module to filter out the frequency signal of the irrelevant frequency band and sample to reduce the IQ data rate, wherein,

[0022] The filtering and sampling module 1 uses a Butterworth low-pass digital filter to filter the IQ data output by the IQ demodulation module, and then uses the sampling module to sample the filtered IQ data. The sampling multiple is equal to the ADC sampling frequency divided by the FFT point number divided by the subcarrier interval and then divided by 2. The IQ data after filtering and sampling is the baseband signal IQ data baseBandSignal[m];

[0023] The filter sampling module 2 uses a Butterworth low-pass digital filter to filter the IQ data output by the secondary mixing module. The IQ data output by the secondary mixing module is expressed as mixFreqSignal[m][n f ], where m represents the antenna number, and its value is 1,2; n f Indicates the sub-band number, the value is 1,2; the sampling module is used to sample the filtered IQ data, and its sampling multiple is 2; the filtered and sampled IQ data is recorded as subBandSignal[m][n f ].

[0024] Furthermore, the fast Fourier transform module of each sensing terminal converts subBandSignal[m][n f ] The IQ data of the two antennas in sub-band 1 and sub-band 2 are respectively transformed by FFT to obtain the spectrum data of antenna 1 and antenna 2, which are recorded as subCarrierData[m][n f ].

[0025] Furthermore, the frequency selection grouping module of each sensing terminal performs frequency selection grouping on the spectrum data corresponding to the two antennas in sub-band 1 and sub-band 2 in subCarrierData based on a single frequency bandwidth B. After grouping, spectrum data in different groups carry spectrum information of different frequency points. The same group contains spectrum data corresponding to the two antennas for one or more sampling periods. The processing process is as follows:

[0026] 1) Calculate the starting position of the spectrum data of the lth frequency point startpos l , l represents the frequency point number;

[0027] 2) In the Tth sampling period, the position is selected from the spectrum data subCarrierData of the two antennas as startpos l ,startpos l +1,…,startpos l +B-1 spectrum data set x (l)1T (1,2,…,B) and x (l)2T (1,2,…,B) is stored as a set of data in a cell of the spectrumDataSet data buffer;

[0028] 3) Store the frequency point data in the vertical direction from small to large according to the frequency point number l, and then store the frequency point data of other signal acquisition periods in the horizontal direction from small to large according to the symbol sampling period number T.

[0029] Furthermore, the spectrum sensing module of each sensing terminal applies the spectrum sensing algorithm to calculate the detection statistics of the spectrum data of each frequency point, and then compares it with the decision threshold to determine whether each frequency point is occupied. The specific processing process is as follows:

[0030] 1) Take out a set of data of one or T sampling periods of the lth frequency point in the horizontal direction of the spectrumDataSet data buffer and directly splice it into a two-row N-column array. The first row of data in the array is the spectrum data of antenna 1, and the second row of data is the spectrum data of antenna 2. Calculate the correlation matrix R of the array x , and its calculation formula is as follows:

[0031]

[0032] Where x1 and x2 are both 1×N arrays, corresponding to the spectrum data of antenna 1 and antenna 2, [·] H represents the conjugate transpose of the matrix, R x is a 2×2 complex matrix;

[0033] 2) According to R x The autocorrelation and cross-correlation values ​​in the matrix are used to calculate the detection statistic Det, as shown in the following formula:

[0034]

[0035] Among them, r 11 ,r 22 YesR x The main diagonal elements of the matrix represent the autocorrelation values ​​of the spectrum data of the two antennas, r 12 ,r 21 YesR x The other elements in the matrix represent the cross-correlation values ​​of the spectrum data of the two antennas, and |·| represents the modulus value of the complex number;

[0036] 3) The false alarm probability is estimated by the ratio of the number of frequency points misjudged as "occupied" in all frequency points of spectrum sensing to the total number of actual idle frequency points. Then, the decision threshold is obtained according to the target false alarm probability and the detection statistic calculation formula. The process is as follows:

[0037] a) According to the detection statistic calculation formula, set the theoretical value of the decision threshold, denoted as threshold_theory;

[0038] b) Perform spectrum sensing based on known idle frequency information, obtain judgment results, and calculate the false alarm probability;

[0039] c) Compare the target false alarm probability with the current false alarm probability, fine-tune the decision threshold, and record the temporary value of the decision threshold, recorded as threshold_tmp;

[0040] d) after repeating steps b) and c) for a preset number of times, calculate the average of a portion of temporary decision threshold values ​​threshold_tmp as the decision threshold used for subsequent spectrum sensing;

[0041] 4) Compare the detection statistic with the current decision threshold. If the detection statistic is greater than the decision threshold, the frequency point is considered to be "occupied" by the user. Otherwise, the frequency point is considered to be "idle".

[0042] 5) Repeat steps 1) to 4) until all frequency points in the frequency band are detected.

[0043] Furthermore, the spectrum sensing module can correct the decision threshold according to the target false alarm probability and the current false alarm probability. The detection statistic Det is only related to the noise correlation coefficient ρ between antennas and has nothing to do with the noise power. The Det value is used as the theoretical decision threshold value, and an approximate value Th of the theoretical decision threshold is set. The statistical probability of Det>Th is fixed and unchanged, which is called the false alarm probability P f , Th value and P f The value is negatively correlated, that is, the larger the Th value, the higher the P f The smaller the value, the smaller the Th value P f The larger the value is, the initial value of the decision threshold Th is set to 1, and then spectrum sensing is performed based on the known idle frequency information, the false alarm probability is calculated, and the decision threshold is corrected based on the target false alarm probability and the current false alarm probability. The process is as follows:

[0044] Step 1: The sensing terminal first obtains idle frequency information, and then caches the spectrum data of the signals currently collected by the two receiving antennas;

[0045] Step 2: The sensing terminal calculates the detection statistic Det according to the above steps, and compares Det with the new decision threshold newThreshod;

[0046] Step 3: Count the idle frequency points. If Det is greater than the new decision threshold, the false alarm counter is incremented by 1. After scanning all idle frequency points of the spectrum, the current false alarm probability is equal to the false alarm counter value divided by the total number of scanned frequency points.

[0047] Step 4, determining a correction factor, wherein the current value of the correction factor is determined according to the correction factor value sensed last time and the relationship between the current false alarm probability and the target false alarm probability;

[0048] Step 5: The new decision threshold newThreshod is equal to the correction factor multiplied by Th.

[0049] Furthermore, the base station manages the system spectrum resources and coordinates the communication services of each terminal, wherein:

[0050] In the spectrum sensing service, the base station receives the spectrum sensing results of each sensing terminal for multiple consecutive sensing times and sends them to the edge computing device;

[0051] In the data communication service, the base station obtains the idle frequency prediction results from the edge computing device and determines the frequency for data communication with the terminal. When the perception terminal requests to obtain the idle frequency information, the base station sends the idle frequency information to the corresponding perception terminal;

[0052] The edge computing device interacts with the base station to receive spectrum sensing results and send idle frequency prediction results.

[0053] The edge computing device groups and fuses the sensing results of each terminal according to its time interval. The fusion rule is that in the same time interval, for a certain frequency point, if the proportion of the number of sensing terminals with the sensing result of "occupied" to the total number of all sensing terminals is greater than or equal to the set proportion V, then the fusion result is "occupied", otherwise the fusion result is "idle";

[0054] The edge computing device uses the long short-term memory neural network model as the basic model to train a binary prediction model based on historical perception data and current spectrum perception data;

[0055] The edge computing device then predicts the frequency occupancy in one or more subsequent time intervals based on the binary prediction model and the latest perception results;

[0056] Finally, the prediction results of idle frequency points within the required time interval are sent to the base station.

[0057] The present invention also proposes a sensing method of the spectrum sensing system based on the aforementioned discrete shared frequency band, the method comprising the following steps:

[0058] Collect spectrum data and perform spectrum sensing through multiple sensing terminals, and send the sensing results to the base station;

[0059] The sensing results are then forwarded to the edge computing device through the base station for fusion processing of multiple sensing results;

[0060] The edge computing device is used to fuse multiple sensing results and predict the spectrum occupancy of subsequent multiple time steps through a pre-trained binary prediction model, and the spectrum prediction results are sent to the base station;

[0061] Finally, the base station selects idle spectrum frequency points according to the spectrum prediction results to communicate with the sensing terminal;

[0062] Among them, each sensing terminal collects two channels of RF data internally. The RF data comes from two antennas receiving RF signals. After amplification and filtering by a dual-channel amplification and filtering circuit, a dual-channel digital-to-analog converter is used to sample the time-domain RF signal. The in-phase orthogonal signal demodulation module performs IQ demodulation on the time-domain RF signal to obtain IQ data. The IQ data is transmitted to the secondary mixing module for secondary mixing to divide the signal spectrum into two sub-bands. The fast Fourier transform module performs fast Fourier transform on the signals of the two sub-bands respectively to obtain spectrum data of the two antennas. The spectrum data is grouped according to frequency point selection by the spectrum sensing module, and then the spectrum sensing module performs spectrum sensing. Finally, the spectrum sensing result and sensing time information are sent to the base station through the sensing result sending module.

[0063] The beneficial effects of the present invention are:

[0064] The present invention uses a blind spectrum sensing algorithm to collect enough data once or multiple times within a sensing period, convert it into the frequency domain, extract the data of each frequency point, and apply the sensing algorithm. The algorithm can stably perform spectrum sensing under low signal-to-noise ratio even when the amount of data at a single frequency point is small.

[0065] The sensing terminal of the present invention uses multi-antenna sensing technology, which can improve spectrum sensing sensitivity while being compatible with existing communication systems.

[0066] The perception system of the present invention uses multi-terminal perception technology to fuse the perception results of different terminals in the edge computing device, and then performs regularity analysis and idle spectrum prediction, which can improve the accuracy of spectrum perception while meeting real-time requirements.

[0067] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0069] Figure 1 A schematic diagram of the spectrum sensing process of the existing narrowband spectrum sensing technology;

[0070] Figure 2 A schematic diagram of the structure of a spectrum sensing system for discrete shared frequency bands of the present invention;

[0071] Figure 3 It is a structural diagram of the internal data processing module of the perception terminal of the present invention;

[0072] Figure 4 A schematic diagram of frequency selection grouping of a perception terminal of the present invention;

[0073] Figure 5 A schematic diagram of the data processing flow of the perception terminal of the present invention;

[0074] Figure 6 A schematic diagram of the data processing flow of the edge computing device of the present invention;

[0075] Figure 7 A schematic diagram of a wireless signal spectrum sensing process of a spectrum sensing system for discrete shared frequency bands of the present invention;

[0076] Figure 8 A schematic diagram of a signal mixing process of a perception terminal of the present invention;

[0077] Fig. 9 A schematic diagram of the process of frequency selection grouping of a perception terminal of the present invention;

[0078] Fig.10 A schematic diagram of a threshold correction process of a sensing terminal of the present invention;

[0079] Fig.11 A schematic diagram of the group fusion process of the edge computing device of the present invention;

[0080] Fig.12 A schematic diagram of the fusion decision process of the edge computing device of the present invention;

[0081] Fig.13 A schematic diagram of the training process of a binary prediction model of an edge computing device of the present invention;

[0082] Fig.14 The training process model evaluation loss value and the model evaluation accuracy change curve of the binary prediction model of the edge computing device of the present invention;

[0083] Fig.15 A schematic diagram of a spectrum prediction process of an edge computing device of the present invention;

[0084] Fig.16 It is a schematic diagram of the perception time frequency prediction result of the edge computing device of the present invention and the terminal perception time frequency point occupancy result;

[0085] Fig.17 Schematic diagram of the 40-step continuous spectrum prediction accuracy of the edge computing device of the present invention. DETAILED DESCRIPTION

[0086] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0087] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0088] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0089] See also Figure 2 to Figure 17 , which is a spectrum sensing system and method for discrete shared frequency bands.

[0090] Example

[0091] This embodiment first provides a detailed structure of a spectrum sensing system and method for discrete shared frequency bands, such as Figure 2 As shown in the figure, the system mainly includes at least one base station, at least one edge computing device and several sensing terminals, which are mainly used for spectrum sensing in discrete shared frequency bands, namely, the 223-235MHz frequency band. The system completes spectrum sensing in discrete shared frequency bands based on compatible communication systems. The sensing method is based on the algorithm of multi-antenna spectrum sensing, and performs narrowband signal sensing on single frequency points one by one to achieve accurate and rapid detection.

[0092] The discrete shared frequency band contains a total of 480 frequency points, and each frequency point has a bandwidth of 25kHz. The perception system uses a subcarrier spacing of 2.5kHz, each frequency point contains 10 subcarriers, the number of FFT points is fixed to 4096, and the ADC sampling period is 0.4ms. The perception terminal is responsible for collecting spectrum data and completing spectrum perception, and then sending the perception results to the base station. The base station sends the received perception results to the edge computing device for processing. The edge computing device fuses the perception results of multiple terminals, then trains a binary prediction model and predicts the spectrum occupancy of subsequent multiple time steps to achieve idle spectrum prediction, and finally sends the spectrum prediction results to the base station. The base station can decide which idle frequencies to use to communicate with the terminal based on the spectrum prediction results.

[0093] The main configuration of the perception terminal is: two receiving antennas (i.e., antenna 1 and antenna 2), dual-channel amplification and filtering circuit, dual-channel analog-to-digital converter (ADC) and internal data processing module. The internal data processing module mainly includes: in-phase orthogonal signal (IQ) demodulation module, secondary mixing module, filter sampling module, fast Fourier transform (FFT) module, frequency selection grouping module, spectrum perception module and perception result sending module. The module connection block diagram is as follows: Figure 3 shown.

[0094] The corresponding description of the data processing module inside the perception terminal is as follows:

[0095] IQ demodulation module: The main function of the IQ demodulation module is to perform IQ demodulation on the RF signal data in the 223-235MHz frequency band to obtain IQ data. The spectrum shifting process is equivalent to a mixing. The mixing carrier frequency f c1 , that is, local carrier 1, is 229MHz.

[0096] This module multiplies the two RF signal data collected by the dual-channel ADC with local carrier 1 to obtain the IQ data of the signals collected by antenna 1 and antenna 2. Local carrier 1 contains the frequency f c The I carrier and Q carrier of the c1 n s / f s1 ), -sin(2πf c1 n s / f s1 ), where n s =1,2,…,N s , N s Indicates the length of data collected within one sampling period (i.e. 0.4ms) of the ADC, f s1 is the ADC sampling frequency, f s1The minimum is 491.52MHz. The IQ demodulation module outputs IQ data and generates baseband signal IQ data after filtering and sampling, which is recorded as baseBandSignal[m], where m represents the antenna number and the value is 1,2.

[0097] Secondary mixing module: The main function of the secondary mixing module is to perform secondary mixing on the baseband signal in baseBandSignal. The mixing carrier frequency f c21 , f c22 , that is, local carrier 2, are 3MHz and -3MHz respectively.

[0098] This module multiplies the IQ data of antenna 1 in baseBandSignal by the two carriers in local carrier 2 respectively to obtain the mixed IQ data of two sub-bands, namely the mixed IQ data of sub-band 1 and sub-band 2, which are recorded as mixFreqSignal[m1][1] and mixFreqSignal[m1][2], where m1 represents the antenna number and its value is 1. Local carrier 2 is n1=1,2,…,N1, N1 is the baseband signal IQ data length, f c21 , f c22 The values ​​are 3×10 6 and -3×10 6 , f s2 is the baseband digital sampling frequency, which is equal to N1 divided by the sampling period (i.e. 0.4ms), f s2 The minimum is 20.48MHz. The IQ data of sub-band 1 mainly retains the frequency band information from -6MHz to 0Hz in the baseband signal, and the IQ data of sub-band 2 mainly retains the frequency band information from 0Hz to 6MHz in the baseband signal.

[0099] Similarly, the IQ data of antenna 2 in baseBandSignal is multiplied by the two carriers in local carrier 2 to obtain the IQ data of subband 1 and subband 2 corresponding to antenna 2, which are recorded as mixFreqSignal[m2][1] and mixFreqSignal[m2][2], where m2 represents the antenna number and its value is 2.

[0100] Filter sampling module 1 and filter sampling module 2: Filter sampling module 1 and filter sampling module 2 are used to filter out frequency signals of irrelevant frequency bands at the output ends of the IQ demodulation module and the secondary mixing module respectively, and to sample and reduce the IQ data rate.

[0101] The filtering and sampling module 1 uses a Butterworth low-pass digital filter to filter the IQ data output by the IQ demodulation module, and its passband cutoff frequency range is greater than 6MHz and less than 8MHz; the sampling module is used to sample the filtered IQ data, and its sampling multiple is equal to the ADC sampling frequency divided by the FFT point number (i.e. 4096) divided by the subcarrier interval (i.e. 2.5kHz) and then divided by 2. The IQ data after filtering and sampling is the baseband signal IQ data baseBandSignal[m].

[0102] The filter sampling module 2 uses a Butterworth low-pass digital filter to output IQ data of the secondary mixing module, that is, mixFreqSignal[m][n f ], filtering is performed, where m represents the antenna number and its value is 1, 2; n f Indicates the sub-band number, the value is 1,2. Its passband cutoff frequency range is greater than 3MHz and less than 4MHz; the sampling module is used to sample the filtered IQ data, and its sampling multiple is 2. The filtered and sampled IQ data is recorded as subBandSignal[m][n f ].

[0103] FFT module: This module is used to convert subBandSignal[m][n f ] The IQ data of the two antennas in sub-band 1 and sub-band 2 are respectively subjected to 4096-point FFT transformation, with a subcarrier spacing of 2.5KHz, to obtain the spectrum data of antenna 1 and antenna 2, recorded as subCarrierData[m][n f ], where m represents the antenna number, and its value is 1, 2; n f Indicates the sub-band number, the value is 1,2.

[0104] Frequency selection grouping module: The frequency selection grouping uses a single frequency bandwidth B as the grouping basis. In this embodiment, 10 subcarriers are selected as the single frequency bandwidth, and the spectrum data corresponding to the two antennas in subband 1 and subband 2 in subCarrierData are frequency selected and grouped. After grouping, the spectrum data of different groups carry spectrum information of different frequencies, and the same group contains spectrum data corresponding to the two antennas for one or more sampling periods. The schematic diagram of frequency selection grouping is shown in the figure below. Figure 4 As shown, the processing process is as follows:

[0105] Step 1: Calculate the starting position of the spectrum data of the lth frequency point startpos l , l represents the frequency point number. In sub-band 1, the frequency point l = 0, 2, ..., 239. In sub-band 2, the frequency point l = 240, 241, ..., 479.

[0106] Step 2: In the Tth sampling period, simultaneously select the position startpos from the spectrum data subCarrierData of the two antennas l ,startpos l +1,…,startpos l +9 spectrum data set x (l)1T (1,2,…,10) and x (l)2T (1,2,…,10) is stored as a group of data in a unit of the spectrumDataSet data buffer. Every 10 consecutive 2.5KHz subcarriers correspond to a 25KHz frequency point.

[0107] Step 3: Store the frequency point data in the vertical expansion direction from small to large according to the frequency point number l, with a maximum of 480 units, and then store the frequency point data of other signal acquisition periods in the horizontal expansion direction from small to large according to the symbol sampling period number T.

[0108] Spectrum sensing module: This module uses the spectrum sensing algorithm to calculate the detection statistics of the spectrum data of each frequency point, and then compares it with the decision threshold to determine whether each frequency point is occupied. It includes the following processing steps:

[0109] Step 1: Figure 4 As shown, a set of data of one or T sampling periods of the lth frequency point is taken out horizontally from the spectrumDataSet data buffer and directly spliced ​​into an array of two rows and N columns, N = 10T. The first row of data in the array is the spectrum data of antenna 1, and the second row of data is the spectrum data of antenna 2. The correlation matrix R of the array is calculated. x , and its calculation formula is as follows:

[0110]

[0111] Where x1 and x2 are both 1×N arrays, corresponding to the spectrum data of antenna 1 and antenna 2, [·] H represents the conjugate transpose of the matrix, R x is a 2×2 complex matrix.

[0112] Step 2: According to R x The autocorrelation and cross-correlation values ​​in the matrix are used to calculate the detection statistic Det, as shown in the following formula:

[0113]

[0114] Among them, r 11 ,r 22 YesR x The main diagonal elements of the matrix represent the autocorrelation values ​​of the spectrum data of the two antennas, r 12 ,r 21YesR x The other elements in the matrix represent the cross-correlation values ​​of the spectrum data of the two antennas, and |·| represents the modulus value of the complex number.

[0115] Step 3: The decision threshold is closely related to the false alarm probability. In the present invention, the false alarm probability is approximately estimated by the ratio of the number of frequencies misjudged as "occupied" to the total number of actual idle frequencies among the 480 frequencies sensed in a spectrum. The decision threshold is obtained based on the target false alarm probability and the detection statistic calculation formula. The method steps are as follows:

[0116] The first step is to set the theoretical value of the decision threshold according to the detection statistic calculation formula, denoted as threshold_theory.

[0117] The second step is to perform spectrum sensing based on known idle frequency point information, obtain a judgment result as described in step 4, and calculate the false alarm probability.

[0118] The third step is to compare the target false alarm probability with the current false alarm probability, fine-tune the decision threshold, and record the temporary value of the decision threshold, recorded as threshold_tmp.

[0119] Step 4: After repeating steps 2 and 3 for a certain number of times, the average of a portion of temporary decision threshold values ​​threshold_tmp is calculated as the decision threshold used for subsequent spectrum sensing.

[0120] Step 4: Compare the detection statistic with the current decision threshold. If the detection statistic is greater than the decision threshold, it is determined that the frequency point is "occupied" by the user. Otherwise, it is determined that the frequency point is "idle".

[0121] Step 5: Repeat steps 1 to 4 until all frequency points in the frequency band are detected.

[0122] Perception result sending module: This module adds the perception results of 480 frequency points in the frequency band together with the perception time node information for storage, and then sends the stored perception results and time node information to the base station.

[0123] Perception terminal internal data processing flow: such as Figure 5 As shown:

[0124] Step 1: The sensing terminal collects two channels of RF data. The RF data comes from two antennas receiving RF signals. After being amplified and filtered by the dual-channel amplifier and filter circuit, the ADC is used to sample the time-domain RF signal, which is recorded as a_rx_waveform. The minimum sampling frequency of the ADC is 491.52MHz. Figure 5 Middle 1 step.

[0125] Step 2: Multiply the digital signal a_rx_waveform collected by the ADC by the local carrier 1 and perform IQ demodulation. The 223-235MHz frequency band signal can be moved to zero frequency at one time to obtain the baseBandSignal baseband signal IQ data, and then filter and sample to reduce the IQ data rate. Figure 5 Middle 2 steps.

[0126] Step 3: multiply the baseband signal IQ data of baseBandSignal by the local carrier 2 and perform secondary mixing. The secondary mixing divides the signal spectrum into two sub-bands, and low-pass filters and samples the mixed IQ signal to adapt to the FFT data rate. Figure 5 Middle 3 steps.

[0127] Step 4: Perform 4096-point FFT transformation on the subBandSignal signals of the two sub-bands respectively to obtain the spectrum data subCarrierData of the two antennas, and store the spectrum data. Figure 5 Middle 4 steps.

[0128] Step 5: The spectrum data of the two antennas with a total of 480 frequency points are grouped according to the frequency point selection, and the grouped data is stored. The same group of data after grouping only contains the spectrum data of the two antennas at the same frequency point. Figure 5 5 steps in.

[0129] Step 6: Apply the spectrum sensing method to perform spectrum sensing, and send the spectrum sensing result and sensing time information together to the base station. Figure 5 Middle 6 steps.

[0130] Base station: In the spectrum sensing system of the present invention, the main functions of the base station include system spectrum resource management and coordination of communication services of various terminals.

[0131] In the spectrum sensing service, the base station receives the spectrum sensing results of each sensing terminal for multiple consecutive sensing times and sends them to the edge computing device.

[0132] In data communication services, the base station obtains idle frequency prediction results from the edge computing device and decides which frequencies to use for data communication with the terminal. If the sensing terminal requests to obtain idle frequency information, the idle frequency information is sent to the corresponding terminal.

[0133] Edge computing devices: Edge computing devices interact with base stations to receive spectrum sensing results and send idle frequency prediction results. They have four core functions: sensing result fusion, spectrum law analysis, spectrum prediction, and sending idle spectrum.

[0134] The main functions of edge computing devices and their corresponding descriptions are as follows:

[0135] Perception result fusion: The perception results of each terminal are grouped and fused according to their time intervals. The fusion rule is that in the same time interval, for a certain frequency point, if the proportion of the number of terminals with the perception result of "occupied" to the total number of all perception terminals is greater than or equal to the set proportion V, the fusion result is "occupied", otherwise the fusion result is "idle".

[0136] Spectrum pattern analysis: Use the long short-term memory (LSTM) neural network model to train a binary prediction model based on historical perception data and current spectrum perception data.

[0137] Spectrum prediction: Based on the latest perception results, the frequency occupancy in one or more subsequent time intervals is predicted and stored.

[0138] Send idle spectrum: According to the command of the base station, the idle frequency prediction results within the required time interval are sent to the base station.

[0139] Edge computing device data flow: Figure 6 As shown:

[0140] Step 1: Receive spectrum sensing results from each sensing terminal from the base station, and store them in groups according to sensing time. Different groups represent different sensing times, and the same group only contains the sensing results of different terminals in the same sensing time interval. Figure 6 Middle 1 step.

[0141] Step 2: Fuse the perception results of different perception terminals in the same group to obtain the fusion results of 480 frequency points, and then fuse all the stored groups in turn to store the fusion results. Figure 6 Middle 2 steps.

[0142] Step 3: Load the historical fusion perception result dataset and use it to train a binary spectrum prediction model. Figure 6 Middle 3 steps.

[0143] Step 4: Use the most recent fusion perception results and the trained spectrum prediction model to perform spectrum prediction and output idle frequency prediction results. Figure 6 Middle 4 steps.

[0144] Step 5: Send the idle frequency prediction result to the base station, and the base station schedules idle spectrum resources for communication based on the result. Figure 6 5 steps in.

[0145] This embodiment describes the present invention in more detail based on actual wireless scenarios. In a discrete shared frequency band, there is a situation where unknown communication devices randomly occupy some frequency points, that is, different users use some frequency points according to different rules.

[0146] Spectrum sensing of wireless signals is essentially a process of performing a series of data processing on wireless signals and finally obtaining a sensing result. In this embodiment, the spectrum sensing process includes the sensing terminal executing the spectrum sensing process, the base station interacting with the sensing terminal and the edge computing device, and the edge computing device processing the data. The sensing terminal is mainly responsible for spectrum sensing and sending the sensing results to the base station, which is then forwarded to the edge computing device. The edge computing device receives the spectrum sensing results, analyzes the spectrum occupancy rules, predicts the idle spectrum, and sends the spectrum prediction results to the base station. The wireless signal spectrum sensing processing block diagram is shown in the figure. Figure 7 shown.

[0147] In this embodiment, the perception terminal is set according to a fixed 4096-point FFT length, 2.5kHz subcarrier spacing, and a sampling period of 0.4ms. The IQ demodulation module and the filter sampling module convert the 223-235MHz band RF signal data into baseband signal IQ data, recorded as baseBandSignal, with a bandwidth of 12MHz. The 4096-point FFT cannot directly process the baseband signal data, and a secondary mixing module is required to convert the baseband signal IQ data into two sub-band IQ data after filtering and sampling, and the sub-band IQ data is processed separately. The sub-band IQ data is recorded as subBandSignal. The perception system processes the 223-235MHz band perception result data in multiple perception cycles, continuously completes spectrum perception, and then performs spectrum law analysis to achieve spectrum prediction, and finally outputs the frequency point occupancy time diagram.

[0148] Perception terminal: The perception terminal is responsible for collecting signal data from two receiving antennas, then performing a series of time domain and frequency domain data processing to complete spectrum perception and send the perception results to the base station.

[0149] Key parts of data processing inside the perception terminal:

[0150] IQ demodulation and filtering sampling: The IQ demodulation module first performs IQ demodulation on the 223-235MHz frequency band RF time domain signal data collected by the ADC, which is equivalent to a mixing. In this embodiment, the ADC sampling frequency is 491.52MHz, and the sampling period is 0.4ms. Since the ADC collects real signals, the RF signal is conjugate symmetric about zero frequency. The IQ demodulation module can move the 223-235MHz frequency band signal spectrum to zero frequency at one time. In order to reduce the data rate, low-pass filtering and sampling are required to obtain the baseband signal IQ data.

[0151] The specific implementation process is as follows:

[0152] Step 1: Multiply the two antenna RF real signal data with local carrier 1 respectively, perform IQ demodulation, and obtain IQ demodulated data. Local carrier 1 contains cos(2πf c1 ns / f s1 ), -sin(2πf c1 n s / f s1 ), where n s =1,2,…,N s , N s Indicates the length of data collected within one sampling period (i.e. 0.4ms) of the ADC, f s1 is the ADC sampling frequency, f s1 is 491.52MHz. In this embodiment, N s is 196608. c1 It is the center frequency of the 223-235MHz band, which is 229MHz.

[0153] Step 2: In this embodiment, a 6th-order Butterworth low-pass digital filter is used to perform low-pass filtering on the IQ demodulated data respectively, and the passband cutoff frequency used by the filter is 7 MHz.

[0154] Step 3: In this embodiment, a 24-fold digital sampler is used to sample the filtered IQ data. The data rate after sampling is 8192 / sampling period, which can represent the signal with a subcarrier spacing of 2.5kHz and a bandwidth of 12MHz without distortion, and is exactly twice the FFT data rate. The sampled data is the baseband signal IQ data baseBandSignal.

[0155] Secondary mixing and filtering sampling: The bandwidth of the baseband signal baseBandSignal is 12MHz, the subcarrier spacing is 2.5kHz, and the required FFT points are too large. It is necessary to use a secondary mixing module to divide the baseband signal spectrum into two sub-bands for separate processing, namely sub-band 1 and sub-band 2. Sub-band 1 mainly retains the -6MHz to 0Hz frequency band information in baseBandSignal, and sub-band 2 mainly retains the 0Hz to 6MHz frequency band information in baseBandSignal. The secondary mixing module mixes the baseband signal, and after filtering and sampling, its data rate matches the FFT points. The signal mixing process is as follows: Figure 8 shown.

[0156] The specific implementation process is as follows:

[0157] Step 1: Multiply the baseband signal IQ data of baseBandSignal by the two complex carriers in local carrier 2 to obtain the mixed IQ data mixFreqSignal[m][n] of sub-band 1 and sub-band 2 respectively. f ], where m represents the antenna number, and its value is 1,2; n f Indicates the sub-band number, the value is 1,2. Local carrier 2 is n1=1,2,…,N1, N1 is the baseband signal IQ data length, f c21 , f c22 The values ​​are 3×10 6 and -3×10 6 , f s2 is the baseband digital sampling frequency. In this embodiment, f s2 It is 20.48MHz.

[0158] Step 2: The signal after secondary mixing contains irrelevant frequency band signals, and the filtered IQ data is not easily interfered by irrelevant frequency band signals due to aliasing during subsequent sampling. In this embodiment, a 6th-order Butterworth low-pass digital filter is used to low-pass filter the mixed IQ data mixFreqSignal of sub-band 1 and sub-band 2 respectively, and the passband cutoff frequency used by the filter is 3.5 MHz.

[0159] Step 3: Use a 2x digital sampler to sample the IQ data filtered in step 2 to reduce the data rate. The sampled IQ data is recorded as subBandSignal[m][n f ], the data length matches the number of FFT points of the system.

[0160] FFT transformation: Perform 4096-point FFT transformation on the IQ data corresponding to the two antennas in sub-band 1 and sub-band 2 in subBandSignal to obtain the spectrum data of antenna 1 and antenna 2, recorded as subCarrierData[m][n f ], similarly, m represents the antenna number, and its value is 1,2; n f Indicates the sub-band number, the value is 1,2.

[0161] Frequency selection grouping: Frequency selection grouping uses a single frequency bandwidth as the grouping basis, that is, 10 subcarriers, and performs frequency selection grouping on the spectrum data of two antennas in subband 1 and subband 2 in subCarrierData. After grouping, spectrum data in different groups carry spectrum information of different frequency points, and the same group contains spectrum data corresponding to two antennas for one or more sampling periods. The frequency selection grouping process is as follows: Fig. 9 As shown:

[0162] The specific implementation process is as follows:

[0163] Step 1: Obtain the two-antenna spectrum data subCarrierData of sub-band 1 and sub-band 2.

[0164] Step 2: Calculate the starting position startpos of the spectrum data of the lth frequency point in the two sub-bands l , the calculation method is as follows:

[0165] startposl =N fft / 2-(N valid / 2+mod(l,240))×C (3)

[0166] Among them, N fft is the number of FFT points, N valid Indicates the effective data length in sub-band 1 or sub-band 2. There are 2400 effective subcarriers. C indicates the number of single frequency data points, which is 10. l is the frequency number. The value of mod(l, 240) is the remainder of l divided by 240. The frequency points of sub-band 1 are l = 0, 2, ..., 239. The frequency points of sub-band 2 are l = 240, 241, ..., 479.

[0167] Step 3: Prepare a two-dimensional data buffer spectrumDataSet, which is expanded horizontally from small to large according to the number of signal acquisition cycles, and vertically from small to large according to the frequency point number. Each unit can store an array of two rows and ten columns, and the array elements contain the spectrum data of the two antennas at the same frequency point.

[0168] Step 4: In the Tth sampling period, simultaneously select the position from the spectrum data subCarrierData of the two antennas as startpos l ,startpos l +1,…,startpos l The spectrum data of +9 is organized into an array of two rows and ten columns and stored in the (l, T) unit in the spectrumDataSet data buffer.

[0169] Step 5: Store the spectrum data of each frequency point in order according to the expansion direction of the frequency point number from small to large, with a maximum of 480 units. Then store the spectrum data of other sampling periods in the expansion direction of the signal sampling period number from small to large.

[0170] Spectrum sensing: The spectrum sensing algorithm is used to calculate the detection statistics of the spectrum data of each frequency point, and then compared with the decision threshold to determine whether each frequency point is occupied.

[0171] The perception algorithm implementation process is as follows:

[0172] Step 1: Take out a set of data of one or T sampling periods of frequency point l from the spectrumDataSet data buffer in the horizontal direction and directly splice it into an array of two rows and N columns, N=10T, and T is 1 in this embodiment. The first row of data in the array is the spectrum data of antenna 1, and the second row of data is the spectrum data of antenna 2. Calculate the correlation matrix R of the array x , R xIt is a 2×2 complex matrix, in which each element is the autocorrelation value or cross-correlation value of the spectrum data of two antennas.

[0173] Step 2: According to R x The autocorrelation values ​​and cross-correlation values ​​in the matrix are used to calculate the detection statistic Det, and the formula is as follows:

[0174]

[0175] Among them, r 11 ,r 22 YesR x The main diagonal elements of the matrix represent the autocorrelation values ​​of the spectrum data of the two antennas, r 12 ,r 21 YesR x The non-main diagonal elements in the matrix represent the cross-correlation values ​​of the spectrum data of the two antennas, and |·| represents the modulus value of the complex number.

[0176] Step 3: Compare the detection statistic with the decision threshold Th. If Det>Th, the frequency point is considered to be "occupied" by the user, and the perception result is "1". Otherwise, the frequency point is considered to be "idle", and the perception result is "0".

[0177] Step 4: Repeat steps 1 to 3 until all frequency points in the frequency band are detected, and store the sensing results in the buffer SSoutcomeSet.

[0178] Sending the perception results: adding the perception results and the perception time node information in the perception result buffer SSoutcomeSet together for storage, and then sending the perception results and the perception time node information to the base station.

[0179] Part III: Decision Threshold Correction Process

[0180] In the zero-mean Gaussian white noise scenario, when there is no user signal, the noise power of the two receiving antennas of the sensing terminal is expressed as P n , so the autocorrelation values ​​of the antenna noise data are all P n There is a weak correlation between the noise data on the two lines, and their cross-correlation value can be expressed as:

[0181]

[0182] Where ρ represents the correlation coefficient, which is a random number that is only related to the noise properties but not to the noise power, and it is assumed that |ρ| < 1. The autocorrelation value and cross-correlation value in the above detection statistic Det are replaced by the noise power P n and the correlation coefficient ρP n Substituting, we get:

[0183]

[0184] It is determined that when there is no user signal, the detection statistic is only related to the noise correlation coefficient ρ between antennas, and has nothing to do with the noise power. At this time, the Det value can be used as a theoretical decision threshold. The correlation coefficient ρ is not a fixed value and is difficult to accurately estimate. Therefore, an approximate value Th of the theoretical decision threshold can be set, which is a fixed value and has nothing to do with the noise power.

[0185] If the distribution law of the noise remains unchanged, for the "idle" frequency point, in this embodiment, it can be determined that the statistical probability of Det>Th is fixed and unchanged, which is called the false alarm probability P f The false alarm probability is an important indicator for measuring spectrum sensing performance, indicating the probability that the sensing system will mistakenly judge the spectrum not occupied by users as "occupied". f The value is negatively correlated, that is, the larger the Th value, the higher the P f The smaller the value, the smaller the Th value P f The larger the value.

[0186] In this embodiment, the value of the decision threshold Th is first set to 1, and then spectrum sensing is performed according to known idle frequency point information to calculate the false alarm probability, and the decision threshold is corrected according to the target false alarm probability and the current false alarm probability.

[0187] Specific process of threshold correction: Fig.10 As shown:

[0188] Step 1: Turn on the decision threshold correction function. The sensing terminal first obtains the idle frequency point information, and then caches the spectrum data of the current signals collected by the two receiving antennas;

[0189] Step 2: The sensing terminal calculates the detection statistic Det according to the above steps, and compares Det with the new decision threshold newThreshod;

[0190] Step 3: Count the idle frequency points, and if Det is greater than the new decision threshold, the false alarm counter is incremented by 1. After scanning all idle frequency points in the spectrum, the current false alarm probability is equal to the false alarm counter value divided by the total number of scanned frequency points.

[0191] Step 4: The current value of the correction factor is related to the correction factor value sensed last time and the current false alarm probability and the target false alarm probability. If the current false alarm probability is greater than 1.1 times the target false alarm probability, the correction factor is increased by 0.1 / s. If the current false alarm probability is less than 0.9 times the target false alarm probability, the correction factor is reduced by 0.1 / s. If both the last false alarm probability and the current false alarm probability are greater than the target value or both the last false alarm probability and the current false alarm probability are less than the target value, s remains unchanged. If the last false alarm probability is greater than the target value and the current false alarm probability is less than the target value or the last false alarm probability is less than the target value and the current false alarm probability is greater than the target value, s is equal to twice the last s value.

[0192] Step 5: The new decision threshold newThreshod is equal to the correction factor multiplied by Th. In the threshold correction stage, 100 spectrum sensings are performed, and the average value of the threshold correction factors from the 81st to the 100th is calculated as the final correction factor value.

[0193] Part 4: Edge Computing Device Processing

[0194] After receiving the perception results, the edge computing device first groups them according to the perception time information, and then groups the perception results D in the same group. i (1,2,…,480), contains 480 frequency point results, the value is 0 or 1, i represents the number of the sensing terminal, and the fusion is performed to obtain the fusion result D ot (1,2,…,480), t is the serial number related to the perception time, and the fused result D is stored according to the serial number t ot (1,2,…,480). The schematic diagram of group fusion is as follows Fig.11 As shown:

[0195] The same set of sensing result data is fused, that is, in the same time interval, for a certain frequency point, as long as the number of sensing terminals with a sensing result of "1" accounts for a proportion greater than or equal to V of the total number of sensing terminals, the fusion result is "1", indicating that the frequency point is "occupied", otherwise, the fusion result is "0", indicating that the frequency point is "idle". In this embodiment, the total number of sensing terminals is 5, and V is equal to 0.6.

[0196] The specific implementation process of fusion judgment: Fig.12 As shown:

[0197] Step 1: Obtain a set of grouped perception result information, including the perception results of multiple terminals.

[0198] Step 2: Count the total number of terminals terminals1 with a perception result of "1" for frequency point number i and the total number of all perceived terminals terminalsAll.

[0199] Step 3: Calculate terminals1 / terminalsAll and compare it with the ratio V. If terminals1 / terminalsAll is greater than or equal to the ratio V, the fusion result is "1", that is, the frequency point is considered to be "occupied", otherwise the fusion result is "0".

[0200] Step 4: According to step 2-step 3, the perception results of other frequency point numbers are fused, and the fusion results of 480 frequency points are stored.

[0201] Step 5: According to Step 1 to Step 4, integrate the perception results of other perception time intervals, that is, other groups and each frequency point.

[0202] Step 6: Spectrum law analysis: A binary prediction neural network model is trained based on the historical data of spectrum sensing results and the current sensing results.

[0203] In this embodiment, the LSTM neural network model is used to complete the training. LSTM has a memory unit that can store information for a long time, so that the model can capture the long-distance dependencies in the sequence. LSTM also has an input gate, a forget gate, and an output gate, so that the model can better process long sequence data. Therefore, the LSTM model can achieve the prediction of spectrum usage for the next time step based on a large amount of historical data.

[0204] The first step is to load the fusion result data. The fusion result data is an array data_table with a dimension of M×480, and the array elements are 0 or 1. In data_table, each column from left to right corresponds to a frequency point number from 0 to 479, and each row from top to bottom corresponds to a perception time sequence number from 0 to M-1. Convert data_table to floating-point type data input_data. Extract the data from the zth row to the z+tw-1th row from input_data as a training sequence and store it in seq(z,:,:), where ":" means traversing the entire data space on this dimension of the array, z=0,1,…,Mt w -1, the sizes of the three dimensions of seq are Mt w ,t w and 480. Extract the z+tth w The row result is stored as a label in label(z,:), and the two dimensions of label are (Mt w ) and 480. The training sequence seq(z,:,:) and the corresponding label(z,:) are stored in the data set SSDataSet. In this embodiment, M is equal to 1090, t w Equals 50.

[0205] Step 2: Remove the last 40 datasets from SSDataSet, each of which contains seq and its corresponding label. The removed datasets are used as the test set TestDataSet for final prediction. The remaining datasets in SSDataSet are randomly selected in a ratio of 9:1 and stored in the training set TrainDataSet and the test set ValidDataSet. Then TrainDataSet and ValidDataSet are added to the iterators train_loader and valid_loader respectively according to the size of batch_size.

[0206] Step 3: Set the model parameters such as the number of input and output features, hidden layer size, number of LSTM layers, and create an LSTM model. The model parameters are shown in Table 1.

[0207] Table 1

[0208] Parameter name Parameter Value Number of input features (input_size) 480 Output feature number (output_size) 480 Activation Function Sigmoid function Hidden layer size (hidden_size) 50 Number of LSTM layers (num_layers) 4 Initial learning rate (learning_rate) 0.005 Optimizer Adam Loss Function Binary Cross Entropy Function

[0209] Step 4: Randomly extract seq and label data from train_loader and store them in X_batch and Y_batch respectively as the sequence set and label set during training. Send X_batch to the created model to get the output result outputs. Outputs is the value mapped by the S-shaped growth curve function (sigmoid function for short) as the activation function, and its value is between 0 and 1. Then pass outputs and Y_batch into the binary cross loss function (BCELoss function for short) to calculate the loss value.

[0210] Step 5: Calculate the input gate, forget gate, output gate, candidate memory cells, and linear layer gradients based on the optimizer and loss function. The optimizer subtracts the corresponding gradients multiplied by the learning rate to update the parameters.

[0211] Step 6. After completing 5 trainings for all samples, extract the seq and label data from valid_loader and store them in X_batch and Y_batch respectively as the sequence set and label set for model evaluation. Use the currently trained model, loss function and validation set data to calculate the model evaluation loss value and evaluation accuracy to evaluate the model training status.

[0212] The model evaluation accuracy is calculated as follows:

[0213] The outputs of the model evaluation are mapped to binary values ​​of 0 and 1 and stored in map_outputs. The dimensions of map_outputs and Y_batch are both batch_size×480, with the same data type and elements of 0 or 1. The absolute value of map_outputs-Y_batch is calculated, and then all elements of the result are summed and divided by (batch_size×480) to get the error rate errrate. Finally, the evaluation accuracy is equal to 1-errrate.

[0214] Step 7. The entire training process includes training, evaluation, retraining, and re-evaluation. When the loss value of the model evaluation is lower than 0.058, the training is completed, the model is saved, and the training ends. When the number of model evaluations reaches the maximum number of training times, it means that the training has failed and the training ends. The curve of the model evaluation loss value and accuracy change during the training process is as follows: Fig.14 shown.

[0215] Step 7: Use the trained model to predict the spectrum occupancy of the following multiple time steps based on the latest spectrum sensing results of multiple different time nodes, and store them. In this embodiment, the first data set TestFirstSet in the test data set TestDataSet is used to complete the single-step prediction, and then TestFirstSet and the current prediction result are used to predict the data result of the next time step, achieving 40 steps of continuous prediction to obtain the prediction result.

[0216] Spectrum prediction steps: Fig.15 Shown

[0217] The first step is to load the test data set TestDataSet that was divided during data preprocessing. Extract the first training data set in TestDataSet, that is, seq(0,:,:), and save it to TestFirstSet. The dimension of TestFirstSet is t w ×480.

[0218] The second step is to create an LSTM model according to the model parameters during training and load the trained model.

[0219] The third step is to send TestFirstSet to the loaded model for prediction and get the predicted value. The predicted value is a value between 0 and 1. It is mapped into binary values ​​0 and 1 according to the numerical size. The mapped result is saved in oneoutcome. The dimension of oneoutcome is 1×480.

[0220] Step 4: Extract the data from the second to the last row of TestFirstSet, concatenate oneoutcome to its end, assign them to TestFirstSet, and then repeat step 3 to predict the value of the next time step.

[0221] Step 5: Repeat steps 3 and 4. When the prediction step length reaches 40, the prediction ends and the frequency point occupancy time prediction graph is obtained, as shown in Fig.16 As shown in the middle left figure. The perception terminal has false alarm frequency points during the perception process, such as Fig.16 As shown in the middle right figure, the data comes from the label value in TestDataSet. The edge computing device can effectively eliminate false alarm frequencies during spectrum prediction.

[0222] Comparing the prediction results with the actual spectrum occupancy, the 40-step continuous spectrum prediction accuracy curve is shown in Fig.17 As shown:

[0223] Step 8: Send idle frequency prediction results. According to the base station command, the predicted idle frequency and its corresponding time interval are sent to the base station. If the sensing terminal requests to obtain idle frequency information, the base station is responsible for sending the idle frequency information to the corresponding terminal.

[0224] The present invention is based on a communication network constructed by two-antenna communication devices, and establishes a spectrum sensing system for discrete shared frequency bands. The spectrum sensing method can effectively detect the user occupancy of 480 frequency points in the 223-235MHz frequency band. The sensing terminal performs spectrum sensing and sends the sensing results to the base station. The base station forwards it to the edge computing device for spectrum analysis and prediction to obtain the spectrum occupancy within a period of time. Compared with the traditional spectrum sensing method, the present invention uses a narrowband spectrum sensing method with lower complexity to ensure higher detection accuracy of spectrum sensing. Data fusion and spectrum prediction at the edge computing device can meet the needs of real-time spectrum analysis, and can eliminate some false alarm frequencies, which provides a guarantee for improving the spectrum utilization rate within the frequency band.

[0225] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A spectrum sensing system for discrete shared frequency bands, characterized in that: The system includes: at least one base station, at least one edge computing device and several sensing terminals, each sensing terminal is provided with at least two receiving antennas, a dual-channel amplification and filtering circuit, a dual-channel digital-to-analog converter and an internal data processing module, the internal data processing module includes an in-phase orthogonal signal demodulation module, a secondary mixing module, a filtering sampling module, a fast Fourier transform module, a frequency selection grouping module, a spectrum sensing module and a sensing result sending module connected in sequence, wherein: Multiple sensing terminals collect spectrum data and perform spectrum sensing, and send the sensing results to the base station. The base station then forwards the sensing results to the edge computing device for fusion processing of multiple sensing results. The edge computing device fuses multiple sensing results and predicts the spectrum occupancy of subsequent multiple time steps through a pre-trained binary prediction model, and sends the spectrum prediction results to the base station. The base station selects idle spectrum frequency points based on the spectrum prediction results to communicate with the sensing terminal. Each sensing terminal collects two channels of RF data. The RF data comes from two antennas receiving RF signals. After amplification and filtering by a dual-channel amplification and filtering circuit, a dual-channel digital-to-analog converter is used to sample the time-domain RF signal. The in-phase orthogonal signal demodulation module performs IQ demodulation on the time-domain RF signal to obtain IQ data. The IQ data is transmitted to the secondary mixing module for secondary mixing to divide the signal spectrum into two sub-bands. The fast Fourier transform module performs fast Fourier transform on the signals of the two sub-bands to obtain spectrum data of the two antennas. The spectrum data is grouped according to frequency points through the spectrum sensing module, and then spectrum sensing is performed by the spectrum sensing module. Finally, the spectrum sensing result and sensing time information are sent to the base station through the sensing result sending module.

2. The spectrum sensing system for discrete shared frequency bands according to claim 1, characterized in that: The in-phase orthogonal signal demodulation module in each sensing terminal performs IQ demodulation on the discrete shared frequency band RF signal to obtain IQ data, wherein the in-phase orthogonal signal demodulation module respectively combines the two-way RF signal data of antenna 1 and antenna 2 collected by the dual-channel mode converter with the primary mixing carrier frequency f c1 Multiply the two signals to get the collected IQ data, and mix the carrier frequency f c1 Denoted as local carrier 1, it includes frequency f c The I carrier and Q carrier of the c1 n s / f s1 ) and -sin(2πf c1 n s / f s1 ), where n s =1,2,…,N s , N s Indicates the length of data collected in one sampling cycle of the ADC, f s1 is the ADC sampling frequency; the IQ demodulation module outputs IQ data and generates baseband signal IQ data after filtering and sampling, which is recorded as baseBandSignal[m], where m represents the antenna number and its value is 1,2.

3. The spectrum sensing system for discrete shared frequency bands according to claim 2, characterized in that: The secondary mixing module of each sensing terminal performs secondary mixing on the baseband signal in the obtained IQ data baseBandSignal. The secondary mixing carrier frequency f c21 , f c22 , denoted as local carrier 2, where The secondary mixing module mixes the baseBandSignal of antenna 1 with the two carriers f of local carrier 2. c21 , f c22 Multiply them respectively to obtain the mixed IQ data of sub-band 1 and sub-band 2 corresponding to antenna 1, recorded as mixFreqSignal[m1][1] and mixFreqSignal[m1][2], where m1 represents the antenna number and its value is 1; The secondary mixing module mixes the baseBandSignal of antenna 2 with the two carriers f of local carrier 2. c21 , f c22 Multiply them respectively to obtain the mixed IQ data of sub-band 1 and sub-band 2 corresponding to antenna 2, recorded as mixFreqSignal[m2][1] and mixFreqSignal[m2][2], where m2 represents the antenna number and its value is 2; Among them, local carriers 2 are: Where n1 = 1, 2, ..., N1, N1 is the baseband signal IQ data length, f s2 is the baseband digital sampling frequency, which is equal to N1 divided by the sampling period.

4. The spectrum sensing system for discrete shared frequency bands according to claim 3, characterized in that: The filter sampling module of each sensing terminal includes a filter sampling module 1 and a filter sampling module 2, which are respectively arranged at the output end of the IQ demodulation module and the secondary mixing module to filter out the frequency signal of irrelevant frequency bands and sample to reduce the IQ data rate, wherein: The filtering and sampling module 1 uses a Butterworth low-pass digital filter to filter the IQ data output by the IQ demodulation module, and then uses the sampling module to sample the filtered IQ data. The sampling multiple is equal to the ADC sampling frequency divided by the FFT point number divided by the subcarrier interval and then divided by 2. The IQ data after filtering and sampling is the baseband signal IQ data baseBandSignal[m]; The filter sampling module 2 uses a Butterworth low-pass digital filter to filter the IQ data output by the secondary mixing module. The IQ data output by the secondary mixing module is expressed as mixFreqSignal[m][n f ], where m represents the antenna number, and its value is 1,2; n f Indicates the sub-band number, the value is 1,2; the sampling module is used to sample the filtered IQ data, and its sampling multiple is 2; the filtered and sampled IQ data is recorded as subBandSignal[m][n f ].

5. The spectrum sensing system for discrete shared frequency bands according to claim 4, characterized in that: The fast Fourier transform module of each sensing terminal converts subBandSignal[m][n f ] The IQ data of the two antennas in sub-band 1 and sub-band 2 are respectively transformed by FFT to obtain the spectrum data of antenna 1 and antenna 2, which are recorded as subCarrierData[m][n f ].

6. The spectrum sensing system for discrete shared frequency bands according to claim 5, characterized in that: The frequency selection and grouping module of each sensing terminal performs frequency selection and grouping on the spectrum data corresponding to the two antennas in sub-band 1 and sub-band 2 in subCarrierData based on a single frequency bandwidth B. After grouping, spectrum data in different groups carry spectrum information of different frequency points. The same group contains spectrum data corresponding to the two antennas of one or more sampling periods. The processing process is as follows: 1) Calculate the starting position of the spectrum data of the lth frequency point startpos l , l represents the frequency point number; 2) In the Tth sampling period, the position is selected from the spectrum data subCarrierData of the two antennas as startpos l ,startpos l +1,…,startpos l +B-1 spectrum data set x (l)1T (1,2,…,B) and x (l)2T (1,2,…,B) is stored as a set of data in a cell of the spectrumDataSet data buffer; 3) Store the frequency point data in the vertical expansion direction from small to large according to the frequency point number l, and then store the frequency point data of other signal acquisition periods in the horizontal expansion direction from small to large according to the symbol sampling period number T.

7. The spectrum sensing system for discrete shared frequency bands according to claim 6, characterized in that: The spectrum sensing module of each sensing terminal uses the spectrum sensing algorithm to calculate the detection statistics of the spectrum data of each frequency point, and then compares it with the decision threshold to determine whether each frequency point is occupied. The specific processing process is as follows: 1) Take out a set of data of one or T sampling periods of the lth frequency point in the horizontal direction of the spectrumDataSet data buffer and directly splice it into a two-row N-column array. The first row of data in the array is the spectrum data of antenna 1, and the second row of data is the spectrum data of antenna 2. Calculate the correlation matrix R of the array x , and its calculation formula is as follows: Where x1 and x2 are both 1×N arrays, corresponding to the spectrum data of antenna 1 and antenna 2, [·] H represents the conjugate transpose of the matrix, R x is a 2×2 complex matrix; 2) According to R x The autocorrelation and cross-correlation values ​​in the matrix are used to calculate the detection statistic Det, as shown in the following formula: Among them, r 11 ,r 22 YesR x The main diagonal elements of the matrix represent the autocorrelation values ​​of the spectrum data of the two antennas, r 12 ,r 21 YesR x The other elements in the matrix represent the cross-correlation values ​​of the spectrum data of the two antennas, and |·| represents the modulus value of the complex number; 3) The false alarm probability is estimated by the ratio of the number of frequency points misjudged as "occupied" in all frequency points of spectrum sensing to the total number of actual idle frequency points. Then, the decision threshold is obtained based on the target false alarm probability and the detection statistic calculation formula. The process is as follows: a) According to the detection statistic calculation formula, set the theoretical value of the decision threshold, denoted as threshold_theory; b) Perform spectrum sensing based on known idle frequency information, obtain judgment results, and calculate the false alarm probability; c) Compare the target false alarm probability with the current false alarm probability, fine-tune the decision threshold, and record the temporary value of the decision threshold, recorded as threshold_tmp; d) after repeating steps b) and c) for a preset number of times, calculate the average of a portion of temporary decision threshold values ​​threshold_tmp as the decision threshold used for subsequent spectrum sensing; 4) Compare the detection statistic with the current decision threshold. If the detection statistic is greater than the decision threshold, the frequency point is considered to be "occupied" by the user. Otherwise, the frequency point is considered to be "idle". 5) Repeat steps 1) to 4) until all frequency points in the frequency band are detected.

8. The spectrum sensing system for discrete shared frequency bands according to claim 7, characterized in that: In the spectrum sensing module, the decision threshold can be corrected according to the target false alarm probability and the current false alarm probability. The detection statistic Det is only related to the noise correlation coefficient ρ between antennas and has nothing to do with the noise power. The Det value is used as the theoretical decision threshold value, and an approximate value Th of the theoretical decision threshold is set. The statistical probability of Det>Th is fixed and unchanged, which is called the false alarm probability P. f , Th value and P f The value is negatively correlated, that is, the larger the Th value, the higher the P f The smaller the value, the smaller the Th value P f The larger the value is, the initial value of the decision threshold Th is set to 1, and then spectrum sensing is performed based on the known idle frequency information, the false alarm probability is calculated, and the decision threshold is corrected based on the target false alarm probability and the current false alarm probability. The process is as follows: Step 1: The sensing terminal first obtains idle frequency information, and then caches the spectrum data of the signals currently collected by the two receiving antennas; Step 2: The sensing terminal calculates the detection statistic Det according to the above steps, and compares Det with the new decision threshold newThreshod; Step 3: Count the idle frequency points. If Det is greater than the new decision threshold, the false alarm counter is incremented by 1. After scanning all idle frequency points of the spectrum, the current false alarm probability is equal to the false alarm counter value divided by the total number of scanned frequency points. Step 4, determining a correction factor, wherein the current value of the correction factor is determined according to the correction factor value sensed last time and the relationship between the current false alarm probability and the target false alarm probability; Step 5: The new decision threshold newThreshod is equal to the correction factor multiplied by Th.

9. The spectrum sensing system for discrete shared frequency bands according to claim 1, characterized in that: The base station manages the system spectrum resources and coordinates the communication services of each terminal. In the spectrum sensing service, the base station receives the spectrum sensing results of each sensing terminal for multiple consecutive sensing times and sends them to the edge computing device; In the data communication service, the base station obtains the idle frequency prediction results from the edge computing device and determines the frequency for data communication with the terminal. When the perception terminal requests to obtain the idle frequency information, the base station sends the idle frequency information to the corresponding perception terminal; The edge computing device interacts with the base station to receive spectrum sensing results and send idle frequency prediction results. The edge computing device groups and fuses the sensing results of each terminal according to its time interval. The fusion rule is that in the same time interval, for a certain frequency point, if the proportion of the number of sensing terminals with the sensing result of "occupied" to the total number of all sensing terminals is greater than or equal to the set proportion V, the fusion result is "occupied", otherwise the fusion result is "idle"; The edge computing device uses the long short-term memory neural network model as the basic model to train a binary prediction model based on historical perception data and current spectrum perception data; The edge computing device then predicts the frequency occupancy in one or more subsequent time intervals based on the binary prediction model and the latest perception results; Finally, the prediction results of idle frequency points within the required time interval are sent to the base station.

10. A sensing method of a spectrum sensing system for discrete shared frequency bands based on any one of claims 1 to 9, characterized in that: The method comprises the following steps: Collect spectrum data and perform spectrum sensing through multiple sensing terminals, and send the sensing results to the base station; The sensing results are then forwarded to the edge computing device through the base station for fusion processing of multiple sensing results; The edge computing device is used to fuse multiple sensing results and then use a pre-trained binary prediction model to predict the spectrum occupancy of multiple subsequent time steps, and the spectrum prediction results are sent to the base station; Finally, the base station selects idle spectrum frequency points according to the spectrum prediction results to communicate with the sensing terminal; Among them, each sensing terminal collects two channels of RF data internally. The RF data comes from two antennas receiving RF signals. After amplification and filtering by a dual-channel amplification and filtering circuit, a dual-channel digital-to-analog converter is used to sample the time-domain RF signal. The in-phase orthogonal signal demodulation module performs IQ demodulation on the time-domain RF signal to obtain IQ data. The IQ data is transmitted to the secondary mixing module for secondary mixing to divide the signal spectrum into two sub-bands. The fast Fourier transform module performs fast Fourier transform on the signals of the two sub-bands respectively to obtain spectrum data of the two antennas. The spectrum data is grouped according to frequency point selection by the spectrum sensing module, and then the spectrum sensing module performs spectrum sensing. Finally, the spectrum sensing result and sensing time information are sent to the base station through the sensing result sending module.

Images