Intrinsic Evolution of Predictable Behavior Evolvable Hardware in Dynamic Environment

Abstract

Sensor electronics performance is susceptible to static and dynamic deviations. Even laser trimming still can?t deal with all the deviations. Recently, analog reconfigurable electronics offers a solution to compensate these effects. The state of the art uses genetic algorithm (GA) to find an arbitrary topology to fulfill the given specifications, which can cause hardware with unpredictable behavior. In case of any environmental change, the state of the art starts the evolution from scratch. Considering the robustness of the reconfiguration approach, we used the particle swarm optimization (PSO) [13] as an alternative to GA for reconfiguration of programmable sensor electronics. In this paper, we extend our work to investigate the PSO methods for dynamic environment in which the hardware can track the environmental change without starting from scratch. We run the algorithm on a real hardware (intrinsic evolution). Our hardware was designed in such a way that its performance is predictable by employing standard circuit topologies.