A high-resolution wideband digital channelizer for software radio systems using high-order perfect reconstruction filterbanks

High-resolution wideband digital channelizer is the key component for extraction and reconstruction of multichannel signals in software radio systems. It has been proved that complex-exponential modulated (CEM) perfect reconstruction (PR) filter-bank is an efficient structure for the wideband channelizer, in which the Parks-McClellan (PM) algorithm based design method for the PR prototype filter and PR filterbank is effective when the number of sub-channels is less than 1024. However, due to the limitation of the PM algorithm for the filters with extremely high-order or narrow transition band, it doesn´t work well when the number of sub-channels increases further. In this paper, we propose an efficient design method for such high-order PR filterbanks, which is based on the classical Frequency Sampling (FS) method and the criterion of least mean square error (LMSE). By this proposal, the designs for any high-order PR filters and related high-resolution wideband channelizers can be guaranteed with excellent reconstruction performance. Some numerical simulations are provided to verify the effectiveness of the proposed method, even when the number of sub-channels reaches to 8192.