RU2430417C1 - Radio signal recognition method - Google Patents

Abstract

FIELD: information technology. ^ SUBSTANCE: radio signal recognition method involves giving standard radio signals first, discretisation, quantisation and frame wavelet transformation of the sequence of quantised readings of standard radio signals. A standard radio signal attribute vector is then generated from the wavelet coefficients of the standard radio signals, after which the recognised radio signal is received and its attribute vector is generated in the same way as that of standard radio signals. The received signal is then identified by comparing its attributes with attributes of standard radio signals, where for each time reading of the radio signal, the maximum wavelet coefficient is picked up from corresponding wavelet coefficients at filter outputs, onto which the rest of the wavelet coefficients corresponding to the given time reading of the radio signal are standardised, and average power values of the wavelet coefficients obtained at the output of each filter are chosen as attribute vectors. The received radio signal is identified through successive modulo subtraction of elements of its attribute vector from attribute vectors of each of the standard radio signals, and the recognised radio signal is the incident standard radio signal, the difference between attribute vectors of which is minimal. ^ EFFECT: high efficiency while maintaining the required probability of correct recognition. ^ 12 dwg

Description

The invention relates to pattern recognition, namely to methods for recognizing radio signals, in particular to methods for recognizing the form and modulation parameters of radio signals. The method can be used in the technical means of recognition of radio signals of phase manipulation.

A known method for the recognition of radio signals based on the singular decomposition of the pseudo-frequency-time distribution (pseudo-CWR) Wigner-Ville [N.M. Marinovic, G. Eichmann. An expansion of Wigner distribution and its applications. - Proc. IEEE ICASSP-85, 1985, pp.1021-1024], in which the energy distribution matrices (REs) of the reference radio signals are preliminarily generated on the basis of the Wigner-Ville pseudo-FEC, their spectral decomposition is performed, the parameters of the reference radio signals are formed, then a recognized radio signal is received, discretize and quantize it, form the matrix of the energy distribution of the received radio signal, select the signs of the received radio signal, compare them with the parameters of the reference radio signals and identify the received radio signal from the results of the comparison.

The disadvantage of this method is the relatively low probability of correct recognition (the probability of correct recognition is the relative frequency of making the right decision when assigning the received radio signal to one of the reference classes. The correct recognition event is the opposite (additional) to the error recognition event (P _dec = 1-P _osh ) - see J. Tu, R. González, Principles of Pattern Recognition, Translated from English - Moscow: Mir, 1978, pp. 142-152) of radio signals of complex time-frequency structure, as well as for signals when exposed to noise and interference, which is due to the features used for the recognition of pseudo-FWM Wigner-Ville [Cohen L. Time-frequency distribution. Review // TIIER, 1989, v.77, No. 10. S.72-121]. The probability of correct recognition is reduced due to the appearance of interference background and false power peaks in the Wigner-Ville pseudo-FWM, which distort the real picture of the RE signal in frequency-time coordinates.

A known method of recognizing radio signals according to the patent of the Russian Federation No. 2261476, IPC ⁷ G06K 9/00 from 09/27/2005, in this method, the reference radio signals are pre-set. Then, the reference radio signals are discretized, quantized, and a continuous wavelet transform (VP) operation is performed on them in order to obtain an RE matrix. After that, for each matrix RE form the vector RE. Then, for all obtained RE vectors, the total covariance matrix is calculated. After that, spectral decomposition of the RE matrices of the reference radio signals is performed by calculating the eigenvectors and eigenvalues of the total covariance matrix. Then a truncated matrix of eigenvectors is formed by choosing the eigenvectors of the general covariance matrix corresponding to its maximum eigenvalues. When forming the parameters of the reference radio signals, a truncated matrix of eigenvectors is multiplied by the RE vectors of the reference radio signals, and the average values of the obtained products are used as the parameters of the standards. After that, a recognizable radio signal is received, it is sampled, quantized, and then the continuous VP operation of its quantized samples is performed. Then, an RE vector is formed from the RE matrix, and a truncated matrix of eigenvectors is multiplied by its RE vector to extract features of the received radio signal. The calculation results are taken as signs of recognition of the received radio signal, which are sequentially compared with the parameters of the previously obtained standards. The comparison results serve as the basis for deciding on the correlation of the recognized radio signal to one or another class.

The disadvantage of the considered method is the relatively low responsiveness (speed) (speed is the time it takes for the system to transition from a certain initial state to the desired final state; one of the system quality ratings is see Dictionary on Cybernetics. Kiev: Ukr. Sov. Encyclopedia, 1979, 623 p. // S.89.) of the recognition process itself, due to the need to perform continuous VP operations, perform spectral decomposition of RE matrices and form a truncated matrix of eigenvectors that are associated with a significant amount computing operations.

The closest analogue in technical essence to the claimed one is the method of recognition of radio signals according to RF patent No. 2356064, IPC ⁷ G06K 9/00 of 05/20/2009. In the closest analogue, reference radio signals are pre-set. Then, for each reference radio signal, its RE matrix is formed. To this end, reference radio signals are sampled, quantized, and then perform a frame operation (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 pp.) VP, by filtering their quantized readings by means of filters, the bandwidth 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, an RE matrix is formed, with the rows of the RE matrix of each reference radio signal being the VC obtained at the output of the filters. And from the RE matrices of the reference radio signals, their feature vectors are formed by line-wise concatenation of all the VC generated RE matrices. After that, a recognizable radio signal is received, from the quantized samples of which an RE matrix and a feature vector are formed, similarly to the reference radio signals. The received radio signal is identified by subtracting modulo its feature vector from the feature vectors of each of the reference radio signals. The recognized radio signal is considered incident with the reference radio signal, the difference of feature vectors with which is minimal.

The disadvantage of the prototype method is the relatively low efficiency (speed) caused by the need to select unimportant VCs when forming a feature vector and the relatively low probability of correct recognition when exposed to noise and interference, due to the uneven distribution of their power across the VC of the generated matrix of the RE [V. Dyakonov Wavelets . From theory to practice. - M .: SOLON-R, - 2002.448 s.].

The purpose of the claimed technical solution is to develop a method for recognizing radio signals, providing increased recognition efficiency while maintaining the required probability of correct recognition, by pre-normalizing the VC of the radio signals coming from the filter outputs every time they are subjected to the next time reference when forming its RE matrix. Moreover, as a normalizing value, the maximum VK is selected each time only among those VK that appear at this time at the filter outputs [Dvornikov SV, Saukov A.M. Modification of time-frequency descriptions of non-stationary processes based on exponential and power functions // Scientific Instrument Making. T.14. 2004. No2. S.57-66]. The specified normalization method is defined as temporary normalization, since for the RE matrix, VK normalization is carried out along the coordinate axis of the filter outputs for each value of the coordinate axis of time samples). And the choice as elements of the vectors of signs of the average power values of the VC obtained at the output of each of the filters.

This goal is achieved by the fact that in the method of recognition of radio signals pre-set L≥2 reference radio signals. For each l-th reference radio signal, where l = 1, ..., L, an RE matrix М _{l is formed} , for which it is sampled, quantized, and then the frame VP operation of the sequence of its quantized reports is performed using K≥2 filters. Why the passband ΔФ _{k of the} k-th filter, where k = 1, ..., K, is selected from the condition ΔФ _k = 2 ^(k-1) ΔФ, where ΔФ is the bandwidth of the first filter. Then, from the VC of the l-th reference radio signal received in each k-th frequency band Δ Ф _k , a feature vector of the l-th reference radio signal is formed. After that, a recognizable radio signal is received and its feature vector is formed in the same way as for the l-th reference radio signal. Then, the received radio signal is identified by comparing its features with those of the reference radio signals.

Moreover, for each time reference of the radio signal from the number of corresponding VCs at the filter outputs, the maximum is allocated, to which the remaining VCs are normalized, corresponding to this time reference of the radio signal. And as the elements of the feature vectors, the average values of the VC power obtained at the output of each of the filters are selected. Moreover, the received radio signal is identified by sequentially subtracting modulo elements of its feature vector from the feature vector elements of each of the L reference radio signals and the recognized radio signal is considered to be incident with the reference radio signal, the difference of feature vectors with which is minimal.

Thanks to the new set of essential features in the claimed method, the number of computational operations is reduced during the formation of feature vectors by eliminating the selection procedures of insignificant VC, as well as the choice of the average VC power at the output of each filter as the element of the feature vector. This achieves a reduction in the procedures for the process of recognition of radio signals and thereby provides higher efficiency. In addition, normalizing the VC at the filter outputs for each time reference of the radio signal (time normalization) made it possible to reduce the negative effect of additive and multiplicative noise, which ensured the stability of the generated feature vectors and as a result increased the probability of correct recognition.

The claimed method is illustrated by drawings, which show:

figure 1. The principle of calculating the RE matrix of a radio signal based on a frame VP;

figure 2. A fragment of the test signal S ₁ (t) (signal without noise);

figure 3. A fragment of the test signal Z ₁ (t) (signal Z ₁ (t) represents the signal S ₁ (t) in multiplicative and additive noise);

figure 4. The matrix of RE test signal S ₁ (t);

figure 5. Matrix RE of the test signal Z ₁ (t) without operations of time normalization;

Fig.6. The matrix of RE test signal Z ₁ (t) after the operations of temporary normalization;

Fig.7. The vector of the difference of the feature vectors of the test signal S ₁ (t) and test signal Z ₁ (t) without applying the time normalization operation;

Fig.8. The vector of the difference of the feature vectors of the test signal S ₁ (t) and test signal Z ₁ (t) after applying the time normalization operation;

Fig.9. The spectrum of the test signal S ₁₁ (t);

figure 10. The spectrum of the test signal S ₁₂ (t);

11. The spectrum of the recognized test signal Z ₁₁ (t) representing the test signal S ₁₁ (t) in noise at a signal-to-noise ratio (SNR) of 10 dB;

Fig.12. The vector of the difference of the feature vectors of the signals S ₁₁ (t) and Z ₁₁ (t) and the vector of the difference of the feature vectors of the signals S ₁₁ (t) and S ₁₂ (t).

The implementation of the claimed method is explained as follows.

Pre-set L reference radio signals, the number and types of which cover the possible number and types of real radio signals to be recognized.

Then, a set of procedures is performed to form a feature vector of each l-th reference radio signal, where l = 1, ..., L. For this, each reference radio signal is sampled and quantized. The sampling and quantization procedures for analog signals are known and described, for example [V.Grigoriev. Signal transmission in foreign information technology systems. - SPb .: VAS, 1998, pp. 83-85], and the quantized samples of the reference sequences of radio signals are formed in accordance with the requirements of calculating statistical estimates [Mathematical Encyclopedic Dictionary. M .: Sov. Encyclopedia, 1988, 847 p .; G.Corn, T.Corn. Math reference. Per. from English - M .: Nauka, 1977, pp. 638-643]. The sample length N (the value of discrete reports of radio signals n = 1, ..., N) is selected within 256 ... 16384, depending on the requirements for the probability of correct recognition and processing time (the sample length must be a multiple of 2 ⁱ , where i is an integer) . The larger N, the higher the probability of correct recognition, but the processing time increases.

Then a set of matrices RE {M ₁ ... M _L } is formed, for which a frame IP operation is performed on the quantized samples of reference radio signals. The operation of the frame VP consists in filtering the samples of the quantized radio signal using a combination of K≥1 filters (figure 1). The total number K of filters in this case is determined taking into account the conditions:

,

,

where ΔF is the width of the spectrum of the radio signal; ΔF ₁ - bandwidth of the first filter [V. Dyakonov. Wavelets. From theory to practice. - M .: SOLON-R, 2002, 448 p., Pp. 117-121]. In turn, the bandwidth ΔF _{k of the} k-th filter, where k = 1, ..., K, is selected from the condition

ΔФ _k = 2 ^(k-1) ΔФ ₁ .

Such a choice of filter bandwidths provides a complete overlap in the frequency spectrum of the radio signal with a wavelet filter system, and at the same time, the description redundancy inherent in a continuous VP is avoided [V. Dyakonov. Wavelets. From theory to practice. - M .: SOLON-R, 2002, 448 p., Pp. 104-107]. In this case, the dimension of the RE M matrix is (K × N), where K is the number of filters, N is the number of discrete reports of the signal.

After that, an operation of temporary normalization is performed on the RE matrices of the reference radio signals. The operation of temporary normalization consists in the fact that, in the matrix of REs, the VK are normalized for each temporary value of N along the coordinate K (K is the number of filters). Since the RE matrix can be considered as a set of N VK vectors (N is the number of time samples), the operation of temporary normalization of the RE matrix will be equivalent to N operations of normalization of VK vectors. The operation of normalizing a vector is known and described in [G. Korn, T. Korn. Math reference. - M .: Nauka, 1977, 831 p.].

Then, feature vectors are formed for each of the reference radio signals by averaging the power of the VC at the outputs of each of the K filters. The averaging operation is known and described in [G. Korn, T. Korn. Math reference. - M .: Nauka, 1977, 831 p.].

, выполняют над ним все описанные действия, которые выполнялись над эталонными радиосигналами, и формируют вектор признаков принятого радиосигнала

аналогично, как и для эталонных радиосигналов.After that, a recognized radio signal is received.

, perform on it all the described actions that were performed on the reference radio signals, and form a vector of signs of the received radio signal

similarly as for reference radio signals.

с вектором признаков каждого из эталонных радиосигналов {m₁(k)…m_L(k)}. Идентификация может быть реализована, например, путем вычитания по модулю из вектора признаков принятого радиосигнала векторов признаков каждого из L эталонных радиосигналов

. Процедуры принятия решения являются известными и описаны, например, в [Я.Фомин, Г.Тарловский. Статистическая теория распознавания образов. - М.: Радио и связь, 1986, стр.30-46; Ю.Сато. Обработка сигналов. Первое знакомство. / Пер. с японского, под ред Ёсифуми Амэмия. - М.: Издательский дом «Додека-XXI», 2002, 176 с. с.41-54]. Распознаваемый радиосигнал считают инцидентным одному их L эталонных радиосигналов, с использованием одного из решающих правил, например, когда разница между векторами признаков минимальна.Recognized radio signal is identified by comparing its feature vector.

with the feature vector of each of the reference radio signals {m ₁ (k) ... m _L (k)}. Identification can be realized, for example, by subtracting modulo from the feature vector of the received radio signal the feature vectors of each of the L reference radio signals

. Decision-making procedures are well known and are described, for example, in [Y. Fomin, G. Tarlovsky. Statistical theory of pattern recognition. - M .: Radio and communications, 1986, pp. 30-46; Y. Sato. Signal processing. First meeting. / Per. from Japanese, edited by Yoshifumi Amemiya. - M.: Dodeka-XXI Publishing House, 2002, 176 p. p. 41-54]. A recognized radio signal is considered to be incident to one of their L reference radio signals, using one of the decisive rules, for example, when the difference between the feature vectors is minimal.

The choice of averaged VC values from the outputs of the K filters after the time normalization operation as the feature vector elements for the radio signals is due to the fact that the use of the time normalization operation reduces the negative impact of additive and multiplicative noise on the generated RE matrix [Dvornikov SV, Saukov A.M. Modification of time-frequency descriptions of non-stationary processes based on exponential and power functions // Scientific Instrument Making. T.14. 2004. No2. S.57-66].

тестового сигнала S₁(t). На фиг 5 представлена матрица РЭ

тестового сигнала Z₁(t) без применения операции временного нормирования, а на фиг 6 - матрица РЭ

тестового сигнала Z₁(t) после применения операции временного нормирования. Применение операции временного нормирования позволило уменьшить величину негативного влияния от шумов аддитивного и мультипликативного характера, что подтверждают значения векторов разности

(фиг.7) и

(фиг.8). Здесь m_S(k) - вектор признаков тестового сигнала S₁(t); m_Z(k) - вектор признаков тестового сигнала Z₁(t) без применения операции временного нормирования;

- вектор признаков тестового сигнала Z₁(t) после применения операции временного нормирования.So, in Fig.2 and 3 presents a fragment of the test signal without noise - S ₁ (t) and in noise - Z ₁ (t). Figure 4 shows the matrix RE

test signal S ₁ (t). On Fig 5 presents the matrix RE

test signal Z ₁ (t) without the use of the operation of temporary normalization, and in Fig. 6 - matrix RE

test signal Z ₁ (t) after applying the operation of temporary normalization. The use of the operation of temporary normalization allowed us to reduce the magnitude of the negative effect of noise of additive and multiplicative nature, which is confirmed by the values of the difference vectors

(Fig.7) and

(Fig. 8). Here m _S (k) is the vector of features of the test signal S ₁ (t); m _Z (k) is the vector of features of the test signal Z ₁ (t) without applying the operation of time normalization;

- the vector of signs of the test signal Z ₁ (t) after applying the operation of temporary normalization.

и

. Здесь

- вектор признаков сигнала S₁₁(t) после применения операции временного нормирования;

- вектор признаков сигнала S₁₂(t) после применения операции временного нормирования;

- вектор признаков сигнала Z₁₁(t) после применения операции временного нормирования.The studies showed that for signals with a close frequency-time structure S ₁₁ (t) and S ₁₂ (t), the developed method allows to obtain a rather high contrast of the vectors of their signs. Figure 9 shows the F ₁₁ (f) spectrum of the signal S ₁₁ (t). Figure 10 shows F ₁₂ (f) is the spectrum of the signal S ₁₂ (t). 11 shows F ₃₃ (f) is a spectrum of a recognized signal Z ₁₁ (t) representing a test signal S ₁₁ (t) in noise at an SNR of 10 dB. 12 shows the modules of the difference vectors

and

. Here

- the signal vector of the signal S ₁₁ (t) after applying the operation of temporary normalization;

- the signal vector S ₁₂ (t) after applying the operation of temporary normalization;

- the signal vector of the signal Z ₁₁ (t) after applying the operation of temporary normalization.

меньше разности

.For the case under consideration, R ₁₁ (k) = 0.087, and R ₁₂ (k) = 0.437. Therefore, it is possible to uniquely identify the received signal Z ₁₁ (t), since the difference

less difference

.

Claims (1)

A method for recognizing radio signals, namely, that L≥2 reference radio signals are predefined, is formed for each l-th reference radio signal, where l = 1, ..., L, the energy distribution matrix M ₁ , for which it is sampled, quantized, and then performed an operation of a frame wavelet transform of the sequence of its quantized reports using K≥2 filters, for which the passband ΔФ _{k of the} k-th filter, where k = 1, ..., K, is selected from the condition ΔФ _k = 2 ^(k-1) ΔФ, where ΔФ is the bandwidth of the first filter, then from the wavelet coefficient The l-th reference radio signal received in each k-th frequency band ΔФ _k forms the feature vector of the l-th radio reference signal, after which a recognized radio signal is received and its feature vector is generated in the same way as for the l-th radio reference signal, then the received a radio signal by comparing its features with those of a reference radio signal, characterized in that for each time reference of the radio signal from the number of corresponding wavelet coefficients at the outputs of the filters, the maximum which is normalized by the remaining wavelet coefficients corresponding to a given time count of the radio signal, and the average values of the power of the wavelet coefficients obtained at the output of each filter are selected as elements of the feature vectors, and the received radio signal is identified by sequentially subtracting modulo elements of its feature vector from the elements of vectors features of each of the L reference radio signals, and the recognized radio signal is considered incident with the reference radio signal, the difference of vectors when signs with which is minimal.

Images