An improved global-best-driven flower pollination algorithm for optimal design of two-dimensional FIR filter

Abstract

The design of two-dimensional (2D) digital filter is a higher-order, nonlinear, and multi-modal optimization problem. This paper presents an improved Global-best-driven Flower Pollination Algorithm, named as GFPA, for the design of 2D Finite Impulse Response (FIR) filters. Two methods have been proposed—the first method minimizes the weighted square error via GFPA and the second method finds the coefficients of one-dimensional FIR filter by GFPA before McClellan transformation. The performance of proposed algorithm has been compared with state-of-the-art algorithms and the simulation results show significant improvements. For the design of a $$15\times 15$$ circular symmetric filter, an average reduction of 55% in fitness function evaluation and 72% in execution time is observed. Further, the experiment on CEC 2014 benchmark functions demonstrates better optimal solution than existing algorithms.