Sparse FIR Filter Design using Iterative MOCSA

In this paper, sparse FIR digital filter design using iterative multiobjective cuckoo search algorithm (MOCSA) is presented. The least-squares frequency response error and the l1 norm of selected non-zero filter coefficients of a sparse digital filter are jointly minimized by iterative multiobjective cuckoo search algorithm until convergence. After convergence, a final round of single-objective cuckoo search is performed to minimize the least-square frequency response error of the sparse digital filter to arrive at an optimal solution. With the same set of sparsity levels, the sparse digital filter design results obtained by the iterative MOCSA indicates that smaller weighted least-squares frequency response errors can be obtained as compared to those of another optimization method.