Efficient Design Approaches for Sharp Pseudo-Quadrature Mirror Filter banks using Hybrid Evolutionary Algorithms

Extremely efficient filter banks with satisfactory performances and low implementation intricacy are inevitable tools in multirate signal processing. This paper presents, computationally efficient design approaches for pseudo quadrature mirror filter banks with thin transition bandwidth. The prototype filter is developed by employing frequency response masking (FRM). The filters of the FRM filter are designed using Parks McClellan algorithm and weighted constrained least square approximation, which leads to exceptionally frequency selective filters. To reduce the implementation complexity, the coefficients are represented using canonic signed digit representation. This can degrade the characteristics of the filter bank. Hence evolutionary algorithms are utilized to upgrade the performance. In this paper, the different modified evolutionary algorithms are appropriately chosen to develop hybrid evolutionary algorithms. Improved performances in the discrete space and low implementation intricacy are obtained using hybrid evolutionary algorithms.