Designing a Genetic Algorithm for Function Approximation for Embedded and ASIC Applications
James W. Hauser, Carla Purdy · Conference proceedings · 2006
In embedded systems and application specific integrated circuits (ASICs) that typically do not have a floating-point processor, measured data or function-sampled data is commonly described by an analytic function derived using standard numerical methods. The resultant errors are not caused by rounding but by translating a real solution to a restricted fixed-point environment. We have previously described a genetic algorithm that discovers a superior piece-wise polynomial approximation with coefficients restricted to the integer target space. In this paper we discuss details of the genetic algorithm implementation.