Iterative Method of Radiosignals Detection based on Decision Statistics

Abstract

Intensive computerization of radio systems stimulates the use of various technologies in radar and communication systems, which makes it difficult to solve the problem of signal detection and requires the use of modern processing algorithms in radio monitoring systems with the possibility of modifying their parts according to a specific type of signal and interference. To solve this problem, an iterative method for detecting radio signals based on decisive statistics was developed. The key idea of the proposed method is using of statistical differences between the signal and noise not according to the values of the samples of the received signal, but according to some decisive statistics from them. The essence of the method is to transform the sample of the received radio signal using some operator, followed by the calculation of the decisive statistics and its comparison with some threshold. When the threshold is exceeded, the maximum value of the sample from the converted sample is discarded and the procedure is repeated until the value of the decision statistics becomes less than the threshold. The discarded samples refer to the signal, and the rest to noise. The type of operator is selected based on a priori information about the signal and increases its contrast against random noise. Decisive statistics should have small scattering characteristics, and the distance between its values for noise and the signal mixture should be as large as possible for a given signal-to-noise ratio. A study of the developed method for detecting narrowband signals in the frequency domain showed that the Fourier transform is the optimal form of the operator, and the coefficient of variation is optimal decisive statistic. The developed iterative method for the frequency domain allows detecting narrowband signals at unknown values of noise power in the dynamic range, which is limited only by the level of the side lobes of the window function, when the analysis frequency band is loaded no less than 60%.