Design of optimal digital FIR filters using evolutionary and swarm optimization techniques

• The design of digital FIR HP and BS filters using metaheuristic algorithms, RCGA, PSO and CSA is presented.
• Optimal filter coefficients are computed on account of minimizing absolute magnitude error.
• Analysis is performed over stopband attenuation, flat passband response, least magnitude error, computation time, and fitness profile.
• Results of CSA based filters are advantageous over other designs.
• CSA is observed to be a better algorithm in terms of computation speed, convergence, less number of tuning parameters, and robustness.

Design of optimal filters is an essential part of signal processing applications. It involves the computation of optimal filter coefficients such that the designed filter response possesses a flat passband and up to an infinite amount of stopband attenuation. This study investigates the effectiveness of employing the swarm intelligence (SI) based and population-based evolutionary computing techniques in determining and comparing the optimal solutions to the FIR filter design problem. The nature inspired optimization techniques applied are cuckoo search, particle swarm and real-coded genetic algorithm using which the FIR highpass (HP) and bandstop (BS) optimal filters are designed. These filters are examined for the stopband attenuation, passband ripples and the deviation from desired response. Moreover, the employed optimization techniques are compared on the field of algorithm execution time, t-test, convergence rate and obtaining global optimal results for the design of digital FIR filters. The results reveal that the proposed FIR filter design approach using cuckoo search algorithm outperforms other techniques in terms of design accuracy, execution time and optimal solution.