Simultaneous thermal and timeliness guarantees in distributed real-time embedded systems

Distributed real-time embedded systems have stringent requirements for key performance properties, such as end-to-end timeliness and reliability, in order to operate properly. In recent years, with the continuously decreasing feature size and increasing demand for computation capabilities, today’s real-time embedded systems face an increasing probability of overheating and even thermal failures. As a result, their temperature must be explicitly controlled for improved reliability. While a variety of control algorithms have been proposed for either real-time guarantees or thermal management in an isolated manner, this paper proposes a coordinated control solution that can provide simultaneous thermal and timeliness guarantees for distributed real-time embedded systems running in unpredictable environments. To achieve desired control functions for different generations of processors, our control solution includes two thermal controllers that are designed for processors with and without DVFS support, respectively. In addition, a novel coordination design is proposed to allow the thermal and timeliness control loops to run on their respective desired small timescales for prompt control actions and yet achieve theoretically guaranteed control accuracy and system stability. We implement our control solution in a real system and present empirical results to show that our solution provides dynamic thermal and timeliness guarantees simultaneously. Our simulation results also demonstrate the efficacy of our control solution in large-scale real-time systems.