Swarm Intelligence for Electrical Design Space Exploration

Abstract

Today's high speed electrical systems exhibit ever increasing complexity generation after generation. This increased complexity results in additional design parameters which the system designer must choose carefully to obtain the optimum design. Often, the number of these design variables is large enough that a brute-force search of the design space is not feasible. Statistical techniques like design of experiments (DoE) cannot accurately find the best and worst case corners. This paper introduces the concept of swarm intelligence for the first time for electrical design space exploration. Specifically, the discrete particle swarm optimization (PSO) is used to arrive at an optimum combination of design parameters for various electrical interfaces. The PSO algorithm is shown to have great potential as a robust and efficient alternative to statistical techniques currently used in high speed electrical design optimization.