Energy and timing aware synchronous programming

Abstract

The synchronous paradigm is widely used for the design of safety critical systems. Such systems, especially in the medical devices domain, must meet strict timing requirements while also ensuring long battery life. As a consequence, they are subject to very strict constraint both regarding their WCRT (Worst-Case Reaction Time) and their WCEC (Worst-Case Energy Consumption, the equivalent constraint for the energy consumption). Many techniques exist to compute an upper bound on the WCRT, but few techniques exist that address both the WCRT and the WCEC. We propose here a static analysis framework where conventional WCRT analysis interacts with a DVFS (Dynamic Voltage Frequency Scaling) algorithm to minimize also the WCEC of the given synchronous program. Our algorithm is able to compute the Pareto front of non-dominated solutions in the (WCRT, WCEC) space. Experimental results reveal that the proposed approach is scalable in terms of analysis time while providing more non-dominated solutions compared to two existing approaches. To the best of our knowledge, the proposed approach is the first to produce energy and timing aware synchronous programs.