A cross-layer aging-aware task scheduling approach for multiprocessor embedded systems

This paper presents a cross-layer dynamic task scheduling approach that minimizes aging using the effective parameters of lifetime reliability from different levels of abstraction. Along with technology advances and reduction in the size of the transistors, lifetime reliability of multiprocessor-embedded systems has become a major design challenge. Moreover, the antagonistic impact of this parameter on temperature and power consumption as the other design challenges of multiprocessor systems, make its improvement more challenging. Our proposed scheduling approach is employed during system operation to manage real-time changes and improves lifetime reliability by considering temperature and power consumption as the other main constraints of multiprocessor systems design. Simulation results are performed on ESESC tool and the SPEC2006 benchmarks which show 47.1% improvement in lifetime reliability of the system as well as 11.22% and 13.57% enhancement in temperature and power consumption compared to the base schedule. Our proposed approach is also compared with the related research and its effectiveness in improving lifetime reliability is verified through several experiments.