RU2584493C2 - Method of detecting and correcting falsifications of phonograms based on khurgin-yakovlev presentation - Google Patents

Abstract

FIELD: information security.

SUBSTANCE: invention relates to protecting information from unauthorised change of content of audio record, specifically, to protection of audio records of falsifications and recovery of their initial content. Said technical result is achieved using Khurgin-Yakovlev presentation, which implies transmission groups of thinned signal readings and a derivative, shifted relative to each other by means of special delay line for a certain number of samples. Initial voice information in area of falsification of with certain loss of quality is reduced due to possibility of obtaining readings of signal readings derivative.

EFFECT: technical result consists in reduction of mean-square error signal recovery as compared to synthesising filter built based on V.A.Kotelnikov theorem at 40-60 % at equal orders filters, with additional noise-immunity of reconstructed speech signal in comparison with algorithm based on Kotelnikov theorem at 1, 2 -1, 5 dB.

1 cl, 1 dwg

Description

The invention relates to the field of information protection from unauthorized changes in content, and in particular to a technology for protecting phonograms from falsifications and restoring their original content.

Various systems and devices for protecting and transmitting information are known. For example, RF patent No. 2370898 [1] describes a method of protecting information from unauthorized access in communication systems. The technical result is to prevent the interception of key information near the receiving antenna. The basis for the formation of symmetric encryption keys is a set of random numbers that are generated at the receiving and transmitting points of communication by counter-measurements of the random propagation time of sounding radio signals from the transmitter to the receiver. The keys are generated by irreversible transformations of a large set of measurements of the propagation time of sounding radio signals from the transmitter to the receiver, accumulated over a given time interval of the operation of information security equipment. In this case, the accumulated set of measurements is continuously replenished with the next measurement results.

The disadvantages of this method include the lack of protection against artificial or natural interference and distortion in the communication channels, the lack of an estimate for the probability of imposing false information, in this connection there is no guarantee of the exact receipt of the decryption key.

RF patent No. 23800 [2] describes a method of information protection based on two-way transmission and subsequent detection of sounding radio signals that carry time stamps of their radiation, tied to previously reduced time scales, and emanating from communication devices located at both ends of the radio line, performing irreversible mathematical transformations over numerical sequences, encryption and decryption of messages using a key, characterized in that they perform the operation of mutual recognition of the exchange information clocks and synchronize the time scales of the communication devices of the participants, set the time interval for the operation of information protection tools during which two sets of binary equivalents of the digitized measurement results of a random natural process of phase changes of the received sounding signals are accumulated at both ends of the radio link, which are a superposition of propagating in a multipath environment of partial rays with unpredictable phase incursions, over accumulated sets of binary equi Each of the communication devices performs irreversible mathematical transformations that coincide at both ends of the radio link, form a pair of identical copies of the symmetric encryption key at the receiving and transmitting ends, accumulated sets of binary equivalents replenish the binary equivalents of the digitized results of the next measurements of the phase of the sounding radio signals, moreover, as a key distribution mechanism symmetric encryption uses the reciprocity properties of the conditions of multipath propagation of radio waves.

This patent proposes protection against interference on the communication line by synchronizing the operation of mutual recognition of participants in information exchange on special time scales, because mathematical operations are performed synchronously with the accumulation of binary analogs of measurements, respectively, they are performed according to a certain algorithm, which can be mathematically detected and decrypted. In addition, users must have sophisticated highly sensitive equipment, in the event of failure of which, protection and transmission of protected information is impossible.

As a prototype, a method for generating a phonogram of speech information is considered, in which voice and control marking signals are recorded on a carrier with the possibility of determining the integrity of the phonogram according to the results of its time-frequency visualization, characterized in that a continuous sequence of sonophilic frames is formed simultaneously with the phonogram of the speech information, in each of its frames, spectral analysis using a number of parallel blocks for the formation of harmonics of the control marks frequency signals in the range of sound frequencies, the spectral power of each frame is determined, the harmonics are modulated in amplitude by the spectral power level of each sonofilm frame with a random distribution of their phases so that the power of each harmonic of the control marking signals is weaker than the most powerful frequency-time components of the speech signal (PC ), at least 10 dB, and summarize the indicated harmonics of the control marking signals and the PC before recording them on the medium (Eurasian patent patent Organizations 002 728 B1 20020829 EA, G11B 27/00 / G11B 27/36) [3].

The disadvantage of this method is the ability to identify the structure and filtering of marking signals in the process of analysis of the phonogram. In addition, the addition of additional marking signals degrades the quality of the original phonogram and requires additional processing to clear it.

:In contrast to the known method, the inventive method proposes to form a PC in the form of k groups of thinned out M times samples of the signal f _c (i, M, k) and delayed thinned M times samples of the first M-1 derivatives $$f_P^l\left(i+\left(M-\mathrm{one}\right)N_R+N_0,M,k\right)$$

:

where N _g is the number of samples of the signal or derivative in the group; k = 1 ... [N ₁ / MN _g ] is the number of the group of samples of the thinned signal or its derivative within one analysis segment with a length of N ₁ samples; i = 0 ... N _G -1 is the reference number in the group, α _l is the normalization coefficient, N ₀₁ is the delay value of the sample group of the lth derivative with respect to the signal sample groups.

In FIG. 1 shows a block diagram of a PC phonogram formation algorithm according to the proposed law.

Thus, the invention consists in the fact that samples of derivatives formed in the form of PC samples and samples obtained in blocks 1 are subjected to decimation with a coefficient M (blocks 2) and separate groups are formed consisting of signal samples and derivatives of fixed duration N _G (block 3). The groups of samples of the derivatives are shifted relative to the group of samples of the signal by a certain number of samples (blocks 4) and multiplied by a certain coefficient α _l (blocks 5), and then using the adder (block 6), the original signal is restored according to expression (1).

It was shown in [4] that the samples of the derivative PC have less energy than the samples of the signal (on average 3 times), therefore, the samples of the derivative are masked by the samples of the signal.

Moreover, in the case of recovery by an attacker at the output of speech information using a synthesizing filter based on the theorem of V.A. Kotelnikov, provided low quality speech. When phonograms are expertized on the basis of the Khurgin-Yakovlev representation, the groups of samples of the derivative are shifted back relative to the groups of the decimated signal. Using this conversion, it is possible to detect fraud as well as restore the original signal.

As was shown in [4], the Hurgin – Yakovlev representation makes it possible to simplify the implementation of synthesizing filters with finite impulse responses, providing a decrease in the mean square error of signal reconstruction compared to a synthesizing filter constructed on the basis of V.A. Kotelnikov, 40-60% with equal orders of filters. In addition, the application of this algorithm provides additional noise immunity of the reconstructed speech signal in comparison with the algorithm based on the Kotelnikov theorem by 1.2-1.5 dB.

In [5], it was shown that the Khurgin-Yakovlev algorithm provides high-quality signal recovery when using decomposition to the third order. Thus, it is possible to carry out phonogram formation using three methods: using the first-order Hurgin-Yakovlev representation (M = 2), using twice-decimated signal samples and the first derivative, using the second-order representation (M = 3), when the use of three times decimated samples of the signal, the first and second derivatives; as well as when using a third-order representation (M = 4), when using four times decimated samples of the signal, the first, second and third derivatives.

, $${\widehat{S}}_{п1}\left(\omega \right)$$

, $${\widehat{S}}_{п2}\left(\omega \right)$$

, $${\widehat{S}}_{п3}\left(\omega \right)$$

полученными на основе известных свойств преобразования Фурье:The detection of falsified groups of phonogram samples can be made by comparing the spectra of the current samples of the signal S _c (ω) and its first three derivatives S _p1 (ω), S _p2 (ω), S _p3 (ω) with their estimates $${\widehat{S}}_c\left(\omega \right)$$

, $${\widehat{S}}_{P\mathrm{one}}\left(\omega \right)$$

, $${\widehat{S}}_{P2}\left(\omega \right)$$

, $${\widehat{S}}_{P3}\left(\omega \right)$$

obtained on the basis of the known properties of the Fourier transform:

Using formulas (2), it is possible in the future to restore the original sound file with some loss of quality, but with sufficient speech intelligibility.

List of sources

1. Karpov A.V., Sidorov V.V., Sulmir A.I. Patent RU No. 2370898. A way to protect information.

2. Sidorov V.V., Sulimov A.I., Sherstyukov O.N. Patent RU No. 2423800. Information protection method based on two-way transmission and subsequent detection of sounding radio signals.

3. Zhenilo M.V., Zhenilo V.R., Zhenilo S.V., Kirin V.I. Patent EA 002728 B1 20020829. A method of forming a phonogram of speech information.

4. Kirillov S.N., Dmitriev V.T. Realization capabilities and noise immunity of the signal recovery procedure based on the Khurgin-Yakovlev algorithm // Radio engineering. 2003. - No. 1. - S. 73-75.

5. V.T. Dmitriev, D.I. Lukyanov. Masking algorithm based on the Khurgin-Yakovlev representation using second and third order derivatives. Bulletin of RGRTU, 2012, No. 4. - S. 13-17.

Claims (1)

A method for detecting and correcting phonogram falsifications based on the Khurgin-Yakovlev representation, characterized in that when generating the phonogram, groups of twice decimated samples of the signal and its first derivative are used relative to each other, and fraud is detected when comparing the sum of the ratios of the spectral power density of the difference of samples decimated signal and the first derivative and their estimates for the spectral power of the samples themselves with a threshold previously determined experimentally for For groups of samples after the reverse shift, and in the event that a decision is made about falsification, the samples of the decimated signal and the first derivative are evaluated from the decimated samples of the first derivative and signal, and then the falsified values of the decimated signal and the first derivative are replaced with the obtained estimates of the signal and the first derivative.

Images