An efficient hybrid genetic algorithm to design finite impulse response filters

• Design of 1D FIR filters by conventional methods is not optimal.
• This work sets out to propose a hybrid genetic approach for FIR filters design.
• The local search allows the GA to escape from local optima and premature convergence.
• Hybrid GA significantly reduces the computation cost and improves results.

Although genetic algorithms (GAs) have proved their ability to provide answers to the limitations of more conventional methods, they are comparatively inefficient in terms of the time needed to reach a repeatable solution of desired quality. An inappropriate selection of driving parameters is frequently blamed by practitioners. The use of hybrid schemes is interesting but often limited as they are computationally expensive and versatile. This paper presents a novel hybrid genetic algorithm (HGA) for the design of digital filters. HGA combines a pure genetic process and a dedicated local approach in an innovative and efficient way. The pure genetic process embeds several mechanisms that interact to make the GA self-adaptive in the management of the balance between diversity and elitism during the genetic life. The local approach concerns convergence of the algorithm and is highly optimized so as to be tractable. Only some promising reference chromosomes are submitted to the local procedure through a specific selection process. They are more likely to converge towards different local optima. This selective procedure is fully automatic and avoids excessive computational time costs as only a few chromosomes are concerned. The hybridization and the mechanisms involved afford the GA great flexibility. It therefore avoids laborious manual tuning and improves the usability of GAs for the specific area of FIR filter design. Experiments performed with various types of filters highlight the recurrent contribution of hybridization in improving performance. The experiments also reveal the advantages of our proposal compared to more conventional filter design approaches and some reference GAs in this field of application.