This article is devoted to the problem of accurate detection of cardiosignal QRS-complexes for early diagnosis of various diseases of human cardiovascular system. For that purpose various algorithms based on either digital filtering methods or mathematical modeling of ECG signal particular sections are used. However, all listed methods have a number of disadvantages impairing the accuracy of QRS-complex determination. Yet wavelet transforms enabling accurate identification of local features for non-stationary signals are becoming more common in various fields of technology.

The article presents wavelet spectrogram calculation by means of various wavelets and levels of decomposition in the Wavelet Toolbox environment. Based on wavelet coefficient amplitudes, the presence of jumps, discontinuities, i.e. QRS complex can be identified. In addition, by comparing the form of the QRS complex and the graph of the scaling function of different wavelets, the most optimal wavelet is determined for identifying the QRS complex, as well as noise suppression in cardiosignals.

The obtained results can be used not only in electrocardiography, but also in solving problems of identification and processing of various types of signals. 

The article describes algorithm that allows determining scrambler type based on the signal from its output. The task of this kind is relevant for radio monitoring systems and when creating cognitive systems for digital signal receiving and processing. The identification algorithm determines the form of the scrambler both multiplicative and additive. There are no published papers providing algorithm for automatic determination of scrambler type. This article is intended to partly fill the gap. It provides a form al statement of the problem, an identification algorithm and simulation results. 

Demodulation task is encountered in many practical applications including digital signal processing and digital communications. Demodulation is connected with the communication system performance. Demodulation depends on a number of factors including signal-to-noise ratio (SNR) in the received message. In practice, it is necessary to minimize the number of errors for the given SNR and therefore new demodulation techniques are constantly developed with increased interference immunity. Demodulators aimed at for frequency-hopping spread spectrum signals have to meet special requirements since the message length can reach several ms and the number of messages can exceed several dozens.

Frequency-hopping spread spectrum is a technique of information transmission via radio channel and it is distinguished by variable carrier frequency that can change many times. The carrier frequency changes according to a pseudo random number sequence, which is available to both a sender and a recipient. This technique improves interference immunity of a communication channel.

Frequency-hopping spread spectrum is used in civil and special applications. This signal is stable to jamming (until the third side finds out the number sequence), which makes it possible to use it for special purposes (however, the signal still needs additional encryption).

Demodulation includes signal detection, synchronization, message type determination (modulation speed and modulation type), decoding, determination of autostarting and autostop combinations (for message identification), composition of the received message. The paper considers the tasks beginning with message type determination.

Message type determination can be carried out several ways: using the cross-correlation function, spectral analysis, etc. Since the structure of a synchrosequence is known, it is possible to obtain more precise results using the crosscorrelation function. Several synchrosequences are formed for each message and then we compute their cross-correlation with the received message. The analysis includes the comparison all the results of cross-correlation function computa-tion and finally we make a decision regarding the message.

Determination of autostarting and autostop combinations is performed by comparing autostarting and autostop combinations from the database. Each autostarting combination determines the receiver operation mode (single-channel or frequency-hopping spread spectrum). Determination of combinations is performed during signal demodulation.

Reception of a frequency-hopping spread spectrum signal is performed according to the frequency plan. According to this plan, the carrier frequency changes in fixed time points. After receiving the autostarting combination of frequencyhopping spread spectrum, a reception mode for frequency-hopping spread spectrum signal is switched on. After receiving the autostop combination this mode is terminated. The output of a demodulator is the message itself, modulation type, and carrier frequency.

The outcome of demodulator performance can be represented with a table. The first column of this table contains the carrier frequency, the second column contains frequency deviation, the third column - modulation type, the fourth one - message speed, the fifth one and further - the message itself.

In the paper, we provide new demodulation techniques of frequency-modulated messages for the given SNR. The developed techniques are based on spectral analysis and correlation analysis. We determine the computational complexity of the developed demodulation techniques. The total error is computed for each SNR and the selected demodulation technique using the developed MATLAB/Simulink model for a communication channel. Finally, we conclude about the best demodulation technique for the selected message type for the given SNR. 

The joint analysis of several signals is essential for better understanding of the principles underlying the complex systems dynamics. We consider three methods for estimating the stability of the relative dynamics of two surrogate processes. The first one is based on calculation of the phase synchronization coefficient S and the second one on estimation of the cross-conditional entropy CE. The third approach uses the average value of the coherence function of the two processes - the coherence coefficient C. We study the sensitivity of these methods in relation to the amplitude randomization between test processes. All methods are applied to analyze two types of normally distributed random stochastic processes, with either short-term correlations characterized by finite correlation time or long-term correlations with theoretically infinite correlation time characterized by Hurst exponents. In our research, we generate two copies of the surrogate process with either short-term or long-term correlations. Then we attribute the additive white noise to one of these copies at first with the uniform distribution and then with the Gaussian distribution and the same variance. Next, we calculate the coefficients that characterize the mutual behavior of the two test processes and estimate their statistical characteristics. It is found that the sensitivity of all methods to Gaussian additive noise is higher than that of uniform one. We show that processes with long-term correlation react more actively to the additive amplitude noise then processes with short-term correlation. The influence of Hurst exponent value for the processes with long-term correlation is expressed for the coefficients S and C. The influence of correlation time is demonstrated for the coefficients S and СЕ. Our results may be useful in investigations of the mutual dynamics of two processes belonging to the considered types. 

The use of neural network classification algorithms for solving the problem of receiving telegram-code structures is considered. The article provides comparison of the neural network classifiers analyzing the normalized input signal as well as the signal after the binary conversion. Various measures of the code distance in the space of informative features are considered. Recognition comparative results for the selected pair of symbols are given. On the basis of these results the code distance is determined, which ensures the minimum recognition error probability. The results obtained in the developed neural network classifier are compared with those obtained in correlation receivers operating in the signal time and frequency domains. The advantage of neural network algorithm is shown. The structure implementing the developed neural network classifier is provided. It is shown that the procedure for the classifier developing, k \ including selection of information signs and their amount, as well as code distance, is not of general nature and is to be performed for each set of recognizable symbols. It is stated that to generalize the received alphanumeric blocks it is necessary to use the second decision contour where current information on the reception and information on the duration of the observed symbol is supplied, which is the subject of further research.

The crucial task when summarizing magnetron power in microwave technological installations is tracking of load optimality for individual generators. The article presents a method for estimating the efficiency of magnetron generator based on the current waveform in the anode supply circuit. It is found that the change in the third harmonic current phase in the Fourier expansion and the ratio of the two current extrema adequately reflect the magnetron load change by the high-frequency path. The statement verification is carried out on an experimental setup that allows to purposefully change the reflection of coefficient from the load and compare it with the current change in the high voltage circuit. Studies are carried out for the voltagedoubling scheme. To record changes in the current waveform due to the nonlinearity of the high-voltage transformer, computer simulations are performed with the use of LTSpice program. The LG Magnetron LG 2M214 with an output power of 1 kW is used as the object of research. The magnetron power supply and control board integrated in the general microprocessor control system, is developed and tested. 

A solution is obtained for propagation constant of electromagnetic wave fundamental mode in straightangle waveguide with the cross section partially filled with multilayer nanocomposite ferroelectric film (MNESEP) on a dielectric substrate. The numerical model is based on the scalar form of the finite element method reduced to a homogeneous system of lin-ear algebraic equations with respect to nodal values of the wave scalar for one-dimensional approximation. The nanolayer dielectric properties are determined by barium concentration dependence in composition of BaxSr1-xTiO3 sol-id solution. A calculation program for several dozens of elements of partition along the wide wall of a rectangular waveguide is developed. The results of computer modeling for MNSEP with different number of layers and concentration dependences of barium are given, which allow to stabilize the propagation constant over a wide temperature range.