Multiplierless Evolutionary Filter Design
Benoît Châtelain, François Gagnon · 2007
Evolutive algorithms have demonstrated their potential as optimizers in a wide variety of applications. Automated evolutionary design of analog filters, antennas, logical gates and micro-electro-mechanical systems (MEMS) has resulted in unexpected but efficient topologies and configurations. In this paper, we present an automated design procedure for digital filters based on the use of a genetic algorithm (GA) and high level primitives such as delays, bit shift operators and adders. Given the performance criteria, the proposed algorithm autonomously decides on the components use and circuit configuration. Compared to traditional infinite impulse response (IIR) and canonical signed digits (CSD)-IIR filters, synthesis results show that the evolutionary designed (ED) filter can attain a twofold increase in speed and requires less hardware resource.