A novel approach based on genetic algorithm for pipelining of recursive filters

Look-ahead pipelining is an approach for pipelining IIR filters by adding cancelling poles and zeros to the transfer function. Several modifications of the original look-ahead pipelining have been presented by different authors, Soderstrand et al. (1995) and Lim and Lin (1992), in order to obtain more stable filters with smaller hardware. In this paper a method based on Genetic Algorithm (GA) for optimization of the look-ahead pipelining technique is presented. The presented technique offers improvement over the previously published techniques both in reducing the hardware complexity, and the magnitude of superfluous poles. Finally by allowing extra hardware in the form of shift and add rather than full multipliers further improvements in the magnitude of superfluous poles is obtained with small additional hardware.