Synthesis of low-power embedded software using developmental genetic programming

Abstract

A method of synthesis of software for low-power real-time embedded systems is presented in this paper. A function of the system is specified in the form of the task graph, then it is implemented using embedded processors with low-power and high-performance cores. The power consumption is minimized using the developmental genetic programming. The optimization is based on finding the makespan, satisfying all real-time constraints, for which the power consumption is as low as possible. We present experimental results, obtained for real-life examples and for some standard benchmarks. The results show that our method gives better solutions than makespans obtained using existing methods.