Методи та моделі підвищення завадостійкості інформаційно-телекомунікаційних систем медичного призначення

Abstract

In the thesis, the analysis of modern medical information-telecommunication system (ITS) structures, communication channel types, health information interchange techniques, biomedical signal types, optimal digital filtration methods have been carried out, and necessity of development of new methods of optimal non-stationary noisy signal (including the negative SNR) filtration without having a priori information about signal and noise characteristics has been justified. The method and corresponding structural-analytical model of optimal signal filtration in stationary noise environments based on the minimization of the difference between filtered and approximated signals by the criterion of minimum mean absolute error have been developed. To improve an information signal recovery accuracy in the case of non-stationary noise, the method and corresponding structural-analytical model of the adaptive sampling by the criterion of the minimization the absolute difference of two consecutive values of signal-to-noise ratio estimated for noisy signal optimal filtration and approximation results (estimated correspondingly for two consecutive sampling rates) have been developed. The structural analytical model of advanced data communication in medical information-telecommunication systems has been developed. Software, which includes a module for generating information and noise components of noisy signal, a noisy signal processing module, a filtration and approximation result evaluation module, has been developed. A series of experiments on evaluating results of the stationary (the sum of four harmonics) and non-stationary (electrocardiogram, electroencephalogram, modulated binary sequence) noisy digital signal optimal filtration in the presence of high additive fluctuation (Gauss distribution) and impulse (Bernoulli distribution) noise levels has been carried out. The developed methods of optimal signal sampling and filtration allow us to restore information signal without having a reference signal and noise characteristics (blind signal processing, BSP), to reduce power consumption in the autonomous biomedical equipment, to reduce the bit error ratio (BER), to reduce the medical system hardware (biomedical sensors, transmitters/receivers, storages, etc.) quality demands.