An Iterative Design with Variable Step Prototype Filter for Cosine Modulated Filter Bank

Abstract

A systematic and self controlled prototype filter design approach for multichannel Cosine Modulated Near Perfect Reconstruction (NPR) filter bank is proposed in this paper. The primary goal is to design a prototype filter with enhanced performance i.e., minimum amplitude distortion and aliasing error. This algorithm approximates 3dB cutoff frequency very close to π/2M. This is achieved by selecting suitable step size which is a function of transition width. If the selection of step size is too fine, the objective function oscillates. Whereas, if step size is coarse, 3dB cutoff frequency will not be close to π/2M. This will degrade the overall performance of the prototype filter. Thus by choosing the step size as a function of transition width and varying the step size from coarser to finer level, the minimum amplitude distortion and aliasing error can be definitely achieved. The proposed filter is designed using two input parameters: number of subbands M and attenuation A and all other system parameters are derived from it to avoid heuristic inputs. Simulation results indicate better performance with reference to algorithms existing in literature. In addition, the design approach is systematic and self controlled.