An efficient nonuniform channelizer based on Adjacent Channel Merging

An efficient nonuniform channelization method based on ACM (Adjacent Channel Merging) is proposed to extract large amount of radar signals with different band widths. The neural networks based on cosine basis functions are used to train the LP prototype filter coefficients in order to satisfy the ACM condition. Use this LP prototype filter to form windowed FFT uniform channelizer, then simply add the outputs of the adjacent channels and the nonuniform channelizer can be achieved. The proposed channelizer is simply composed of an FFT component and a few adders, so it has a smaller number of computations than DDC and TPFT channelizers. The computer simulation results given at last shows that the proposed method also has a good channelizing performance.