Exploring delay/area trade-offs of an LDI filter using a natural based algorithm

This paper uses a novel technique to investigate speed, area, and accuracy trade-offs in a digital lossless discrete integrator (LDI) filter. The work concentrates on the implementation of LDI filters with different bit widths rather than their numerical properties. The Jiggle algorithm is based on Darwinian and accelerated evolution ideas, and instead of starting with a single seed, a re-generative multiple-seed approach is taken. The Jiggle algorithm is used for micro-architecture exploration after scheduling and allocation of the functional units of a signal flow graph. The approach selects from a library containing a wide range of implementations to provide delay/area trade-off information for the architecture in question