Heterogeneous real-time embedded software optimization considering hardware platform

Abstract

This paper classifies embedded software into four models according to hardware platform and execution time. We propose an algorithm, CMPCH (Cost Minimization with Probability for Configurable Hardware) that efficiently solves the configurable hardware, no-fixed execution time model, which is the most complicated model among the four models we proposed according to hardware platform and execution time. CMPCH can solve other three models also. Our approach fully takes advantage of configurable hardware and the soft real-time feature to improve the system performance. Experimental results show our approach achieves significant cost-reduction comparing with previous work.