Application of polyphase filter banks to wideband monitoring tasks

Abstract

The present paper discusses wideband monitoring and application of polyphase digital filter banks to dealing with practical tasks in this field. Various ways of classifying digital filter banks are provided including uniform DFT-modulated filter banks (direct and polyphase forms) and we also compare these banks by their performance. A particular polyphase filter bank with given characteristics is investigated in the time and frequency domains. Experimental research illustrates influence of white Gaussian noise on the quality of signal detection in filter bank channels. Experiments have been carried out on the basis of computer simulation in MATLAB and signal generation with the help of functions from standard MATLAB toolboxes. Experimental results contain plots of error of the first and second kinds versus signal-to-noise ratio at the input of a filter bank. Issues of polyphase filter bank implementation using CPU and CUDA are also considered in the paper.