RU2551903C1 - Method of identifying radio signals - Google Patents

Abstract

FIELD: radio engineering, communication.

SUBSTANCE: method includes generating energy distribution matrices from discretised and quantised readings of reference radio signals based on wavelet frame transform thereof; from the matrices, starting from the second row, generating attribute vectors via row-wise concatenation of all wavelet coefficients; normalising elements of the attribute vectors and calculating parameters thereof, wherein the parameters are the average value of normalised amplitude values of the elements of attribute vectors, and a decision is made based on results of calculating the difference between values of parameters of an identified radio signal and reference radio signals. The identified radio signal is considered an incident reference signal, the magnitude of the difference between parameters of attribute vectors of which will be minimal.

EFFECT: method of identifying radio signals which does not require storage in memory of large arrays of values of attribute vectors of reference radio signals.

5 dwg

Description

The invention relates to methods for recognizing radio signals (PC) and can be used in technical means for recognizing the form and modulation parameters of a PC.

A known PC recognition method according to the patent of the Russian Federation No. 2356064 dated 04/24/2007. In the known method, reference PCs are pre-set. Then, for each reference PC, its energy distribution matrix (MRE) is formed. To this end, reference PCs discretize, quantize, and then perform frame ¹ operation ( ¹ Wavelet frames are a wavelet transform that uses a multiple of two scaling (in frequency) and continuous shifts (in time). - see V. Dyakonov Wavelets. From theory to practice. - M .: SOLON-R, - 2002. 448 p. 11 P.106.) Wavelet transforms (FWP) by filtering their quantized readings by means of filters, the pass-band of which each time doubles with increasing filter serial number. After that, the wavelet coefficients (VK) obtained from the output of each filter are normalized, ranked and excluded insignificant VK. The VC set is selected as insignificant, starting from the lowest, the total energy of which is 10-30% of the total energy of the entire VC set at the output of each filter, respectively. Then, from the remaining VCs, the MRE is formed, and the MRE lines of each reference PC are the VC obtained at the output of the filters. And from MRE reference PCs, their feature vectors (VPs) are formed by line-by-line concatenation of all VK formed MREs. After that, a recognizable PC is taken, from the quantized samples of which the MRE and the VP are formed in the same way as for reference PCs. The received PC is identified by subtracting modulo its VP from the VP of each of the reference PCs. The recognized PC is considered to be incident to the reference PC, with which the difference in VIs is minimal.

The disadvantage of this method is the relatively low probability of correct recognition of the PC, due to the modulation parameter, depending on the information carried by the PC. The modulating signal used to modulate the reference and recognized PC is determined by the information content (information component), and, therefore, may be different. As a result of using different modulating signals to modulate the reference PC and the recognized PC, their VIs will be different, even if the reference and recognized PC will belong to the same class of recognized PCs.

A known PC recognition method according to the patent of the Russian Federation No. 2430417 from 05.25.2010

In the known method, reference PCs are pre-set and MRE is formed for each reference PC, for which it is sampled, quantized, and then the FWP operation is performed on the sequence of its quantized reports using filters whose bandwidths are a multiple of two to the power of K, where K is an integer. Then, VCs are formed from the VC reference PC. Moreover, for each time reference PC, among the VCs corresponding to it at the filter outputs, the maximum one is allocated to which the remaining VCs are normalized, corresponding to this PC time reference, and the average VC power values obtained at the output of each filter are selected as VP elements. Then they accept a recognizable PC and form its VP in the same way as for a reference PC. Then, the received PC is identified by sequentially subtracting modulo elements of its IP from the elements of the IP of each of the reference PC. The recognized PC is considered to be incident to the reference PC, with which the difference in VIs is minimal.

The disadvantage of this method is the relatively low probability of correct recognition of the PC, due to the modulation parameter, depending on the information transferred by the PC.

The closest in technical essence to the claimed (prototype) is the "PC Recognition Method" according to the patent of the Russian Federation No. 2423735 dated 03/17/2010. In the prototype method, reference PCs are pre-set, formed for each reference PC MRE, for which it is sampled, quantized and perform the operation of the phase sequence of its quantized reports using filters whose bandwidths are a multiple of two to the power of K, where K is an integer. After this, the FVP VK is normalized with respect to its maximum value. Then, the VC of the reference PC obtained in each frequency band is ranked and the VP of the reference PC is generated. Moreover, the average VC power values obtained at the output of each of the filters are selected as VP elements. Then they accept a recognizable PC and form its VP in the same way as for a reference PC.

The received PC is identified by sequentially subtracting modulo elements of its VIs from the VIs of each of the reference PCs. The recognized PC is considered to be incident to the reference PC, with which the difference in VIs is minimal.

The disadvantage of the prototype method is that the recognition system needs to store large arrays of IP values of the reference signals in memory.

The purpose of the claimed technical solution is to develop a PC recognition method, in which it is not necessary to store large arrays of IP values of the reference signals in memory.

This goal is achieved by pre-setting reference PCs of equal duration, which are discretized and quantized, then perform an FEP operation, as a result of which an MRE is obtained for the sequence of quantized reports of each reference PC and the EP is formed for each of them by line-wise concatenation of VC obtained MRE, after then the VK VPs are normalized, then the recognized PC is received and its VP is formed in the same way as for the reference PC, after which the recognizable PC is identified by comparing its VP with the VP of each of the reference ny PC. Moreover, the VPs are formed by line-wise concatenation of VCs obtained by MRE starting from the second line, the comparison of the VPs of the recognized signal with the VPs of each of the reference PCs is carried out by comparing the calculated average value of the normalized amplitude values of the vector of the recognized signal with the calculated average value of the normalized values of the amplitude values of the vector of each of the reference PCs, and the decision is made by calculating the difference between the values of the parameters of the recognized PC and the reference PC. The recognized PC is considered to be incident to the reference PC, the modulus of the difference in the parameters of the VP with which will be minimal.

Thanks to the new combination of essential features in the claimed method provides the correct recognition of the PC. Moreover, in the inventive method, it is not necessary to store large arrays of VP values of the reference signals in the memory, since PC parameters are used to recognize the PC, calculated from the average values of the normalized amplitude values of the VP elements.

The claimed method is illustrated by drawings, which show:

FIG. 1. Time-discretized 256 samples of the reference PC on-off phase-shift keying (FM-2) (600 baud rate) S ₁ (t);

FIG. 2. Three-dimensional representation of the MRE M _{1 of the} reference PC FM-2 (600 baud) S ₁ (t), obtained on the basis of the FWP for a set of 8 filters, 256 by 8 in size, i.e. MRE contains 2048 VK;

FIG. 3. VP m ₁ (i) of the reference PC FM-2 (600 baud), composed of VK its MRE FVP. The size of the VP m ₁ (i) is equal to 1792 VK (i is the current value of the VK number);

размером 1792 ВК;FIG. 4. VP reference PC FM-2 (600 baud), composed of normalized VK its MRE FVP $${\overline{m}}_{{}_{\mathrm{one}}}\left(i\right)$$

size 1792 VK;

; ФМ-2 (500 бод) $${\overline{m}}_{{}_2}\left(i\right)$$

; ФМ-2 (400 бод) $${\overline{m}}_{{}_3}\left(i\right)$$

и распознаваемого PC, ФМ-2 (400 бод) $$\stackrel{\cap }{\overline{m}}\left(i\right)$$

.FIG. 5. Normalized EP reference PC FM-2 (600 baud) $${\overline{m}}_{{}_{\mathrm{one}}}\left(i\right)$$

; FM-2 (500 baud) $${\overline{m}}_{{}_2}\left(i\right)$$

; FM-2 (400 baud) $${\overline{m}}_{{}_3}\left(i\right)$$

and recognizable PC, FM-2 (400 baud) $$\stackrel{\cap }{\overline{m}}\left(i\right)$$

.

The implementation of the claimed method is explained as follows.

p. 1. Pre-set reference PC of equal duration, which sample and quantize.

The sampling and quantization procedures of analog PCs are known and described, for example, in RF patent No. 2356064 of April 24, 2007. For example, in FIG. 1 shows 256 samples of the reference PC S ₁ (t) FM-2 (600 baud).

Samples of sequences of reference PCs are formed in accordance with the requirements of calculating statistical estimates [Mathematical Encyclopedic Dictionary. M .: Sov. Encyclopedia, 1988. 847 p .; G. Korn, T. Korn. Math reference. Per. from English - M .: Nauka, 1977. p. 638-643]. The report sample length for reference PCs is selected within 64 ... 16384, depending on the requirements for the probability of correct recognition and processing time (the sample length must be a multiple of 2 ^V , where V is an integer, for the convenience of the subsequent calculation of the MRE, the longer the sample length , the higher the probability of correct recognition).

p. 2. Perform the FWP operation for a sequence of quantized reports of each of the reference PCs.

Operations for the implementation of FVP are known and described, for example, in RF patent No. 2423735 dated 03/17/2010. As a result of FVP operations, MREs are obtained. The procedures for the formation of MRE PC based on their FWP are known and described, for example, in RF patent No. 2356064 dated 04.24.2007. As an example, figure 2 shows a three-dimensional representation of MRE M ₁ reference PC S ₁ (t) FM-2 ( 600 baud) obtained on the basis of the FWP for a set of 8 filters (here t is the current time value).

p. 3. Form VP from VC received MRE. Moreover, EPs are built by line-by-line concatenation of VCs obtained by MRE, starting from the second line. After that, the VC in the EP are normalized.

Concatenation consists in sequentially aligning the rows of the matrix one after another. As an example in FIG. Figure 3 shows the VP m ₁ (i) of the reference PC FM-2 (600 baud) formed in the above manner (the first 256 VCs are zero, since the formation of the VP starts from the second line of the MRE).

Normalization consists in highlighting the largest VK and dividing all the other VK by the value of the largest VK.

эталонного PC ФМ-2 (600 бод), элементы которого нормированы.As an example in FIG. 4 shows the VI $${\overline{m}}_{{}_{\mathrm{one}}}\left(i\right)$$

reference PC FM-2 (600 baud), the elements of which are normalized.

The procedure for performing concatenation and standardization procedures is known, for example, see RF patent No. 2356064 dated April 24, 2007.

p. 4. Parameters are calculated from the VI. As parameters, determine the average value of the normalized amplitude values of the elements of the VP.

The calculation of the parameters is carried out according to the formula:

$${\Sigma }_n=\frac{\mathrm{one}}{N}{\displaystyle \sum _{i=\mathrm{one}}^N{\overline{m}}_n\left(i\right).\left(\mathrm{one}\right)}$$

- нормированный ВП n-го эталонного сигнала; N - число ВК в ВП.In the formula (1) $${\overline{m}}_{{}_n}\left(i\right)$$

- normalized VP of the n-th reference signal; N is the number of VK in the VP.

эталонного PC ФМ-2 (600 бод) составят ∑₁=0,158.According to formula (1), the parameters of the VP $${\overline{m}}_{{}_{\mathrm{one}}}\left(i\right)$$

reference PC FM-2 (600 baud) will be ∑ ₁ = 0.158.

и формируют его ВП $$\stackrel{\cap }{\overline{m}}\left(i\right)$$

аналогично, как и для эталонного PC. С этой целью принятый для распознавания PC $$\stackrel{⌢}{S}\left(t\right)$$

дискретизируют и квантуют, а затем последовательно выполняют п.п. 2, 3 и 4. В результате выполнения указанных процедур получают параметры $$\stackrel{⌢}{\Sigma }$$

распознаваемого PC $$\stackrel{⌢}{S}\left(t\right)$$

.p. 5. Receive a recognizable PC $$\stackrel{⌢}{S}\left(t\right)$$

and form it VP $$\stackrel{\cap }{\overline{m}}\left(i\right)$$

similarly to the reference PC. To this end, adopted for PC recognition $$\stackrel{⌢}{S}\left(t\right)$$

discretize and quantize, and then sequentially perform p.p. 2, 3 and 4. As a result of the above procedures, parameters are obtained $$\stackrel{⌢}{\Sigma }$$

recognizable PC $$\stackrel{⌢}{S}\left(t\right)$$

.

The application of the procedure for calculating VP parameters allows storing data on reference PCs as a set of values {значений _n }, the number of which will be determined by the number of classes of recognized PCs.

p. 6. The recognizable PC is identified by calculating the difference between the values of the parameters of the recognizable PC and the reference PC, the recognized PC is considered to be incident with the reference PC, the modulus of the difference in the parameters of the VP with which will be minimal.

The procedures for subtracting modulo from the feature vector of the received PC feature vectors of each of the reference PCs can be implemented by the formula:

$$R_n=\left|\stackrel{⌢}{\Sigma }-\left\{{\Sigma }_n\right\}\right|.\left(2\right)$$

To identify the recognizable PC among the results of calculations by the formula (2), find the minimum. The current index n will indicate the class (number) of the reference PC, the differences with which the recognized PC will be minimal.

Decision-making procedures are well known and are described, for example, in [Yu. Sato Signal Processing. First meeting. / per. with Japanese., edited by Esifumi Amemiya. - M.: Dodeka-XXI Publishing House, 2002. - 176 p. S. 41-54].

. Здесь n - один из порядковых номеров эталонных PC в интервале от 1 до L.A recognized PC is considered incident to one of their L reference PCs when the difference between the VP parameters is minimal $$R_n=\underset{\left|\stackrel{⌢}{\Sigma }-\left\{{\Sigma }_n\right\}\right|}{\min }$$

. Here n is one of the serial numbers of the reference PC in the range from 1 to L.

The implementation of procedures for the identification of a recognized PC in accordance with the claimed method is as follows.

Previously set reference PC of equal duration {S _n (t)} _L. The choice of the number of recognizable classes PC L is determined by the recognition conditions. Then, reference PCs corresponding to each of the L classes are sampled and quantized according to clause 1 (Fig. 1 shows a fragment of the sampled and quantized samples of the reference PC FM-2 (600 baud)). After that, according to claim 2, FEP is performed on each reference PC and a set of MRE {M _n } _L reference PCs is obtained (Fig. 2 shows a fragment of the MRE of the reference PC FM-2 (600 baud)). Then, for each of the reference PCs, a VP is formed in accordance with clause 3 (Fig. 3 shows a fragment of the VP of the reference PC FM-2 (600 baud)). The first line of MRE is not used for the formation of VP for the following reasons. In the formation of the PMF in the first line of the MRE, the VC of the noise components are predominantly localized. Therefore, their exclusion during the formation of the VP will increase their noise immunity.

(на фиг. 4 показан фрагмент нормированного и ранжированного ВП эталонного PC ФМ-2 (600 бод)). Затем, в соответствии с п. 4, из ВП эталонных PC вычисляют значения их параметров по формуле (1). В результате получают совокупность {∑_n}_L из L значений.After this, the EP is normalized and the set is obtained $${\left\{{\overline{m}}_n\left(i\right)\right\}}_L$$

(Fig. 4 shows a fragment of a normalized and ranked VP reference PC FM-2 (600 baud)). Then, in accordance with paragraph 4, the values of their parameters are calculated from the EP reference PCs by the formula (1). The result is a collection of {∑ _n } _L of L values.

для распознавания и, аналогичным эталонным PC образом, вычисляют значение его параметра $$\stackrel{⌢}{\Sigma }$$

согласно п. 5Then take a PC $$\stackrel{⌢}{S}\left(t\right)$$

for recognition and, in a similarly standard PC manner, the value of its parameter is calculated $$\stackrel{⌢}{\Sigma }$$

according to paragraph 5

.As an example, in FIG. 5 shows the EP reference PC FM-2 (600 baud); FM-2 (500 baud); FM-2 (400 baud) and recognizable PC FM-2 (400 baud) (for the convenience of interpreting the results of the VC presented in Fig. 5 VP are ranked). The recognized PC was formed separately and therefore is not a copy of the reference PC FM-2 (400 baud). After that, a recognizable PC is identified in accordance with formula (6) of clause 6. For those shown in FIG. 5 EP reference PC during the experiment obtained the following values of their parameters ∑ ₁ = 0.158; ∑ ₂ = 0.114; ∑ ₃ = 0.098, and the value $$\stackrel{⌢}{\Sigma }=\mathrm{0,101}$$

.

According to the formula (2), the recognized PC was assigned to the class PC FM-2 (400 baud), which confirms the validity of the proposed method. The experiment was carried out in accordance with the requirements of calculating statistical estimates [Mathematical Encyclopedic Dictionary. M .: Sov. Encyclopedia, 1988. 847 p .; G. Korn, T. Korn. Math reference. Per. from English - M .: Nauka, 1977. p. 638-643].

Thus, thanks to a new set of essential features, the claimed method provides PC recognition, in which it is not necessary to store large arrays of IP values of the reference signals in memory.

Claims (1)

A method for recognizing radio signals, which consists in preliminarily setting reference radio signals of equal duration, which are sampled and quantized, then perform a frame wavelet transform operation, as a result of which an energy distribution matrix is obtained for a sequence of quantized reports of each reference radio signal and a vector is generated for each of them features by row-wise concatenation of the wavelet coefficients of the obtained energy distribution matrices, after which the wavelet coefficients vect feature ores are normalized, then a recognizable radio signal is received and its feature vector is formed in the same way as for a reference radio signal, after which a recognizable radio signal is identified by comparing its feature vector with feature vectors of each of the reference radio signals, characterized in that feature vectors are formed by progressive wavelet concatenation -coefficients of the obtained energy distribution matrices starting from the second row, comparing the feature vectors of the recognized signal with the vectors the signs of each of the reference radio signals is carried out by comparing the calculated average value of the normalized amplitude values of the vector of the recognized signal with the calculated average value of the normalized values of the amplitude values of the vector of each of the reference radio signals, and the decision is made according to the results of the calculation of the difference between the values of the parameters of the recognized radio signal and reference radio signals, the recognized radio signal is considered incident reference radio signal, vector parameter difference module signs with which will be minimal.

Images