Lifetime Improvement of Real-Time Embedded Systems by Battery-Aware Scheduling

Many embedded systems and mobile devices use batteries as their energy suppliers. The lifetime of these devices is thus dependent on the battery behavior. Accordingly, battery management besides reducing the energy consumption of the respective system helps to increase the efficiency of such systems. Maximizing the battery lifetime is a quiet challenging problem due to the nonlinear behavior of batteries and its dependence on the characteristics of the discharge profile. This paper employs dynamic voltage scaling (DVS) to extend the lifetime of battery-operated real-time embedded systems. We propose a battery-aware scheduling algorithm to maximize the lifetime and efficiency of the battery. The proposed algorithm is based on greedy heuristics suggested by battery characteristic and power consumption of tasks to employs DVS. Two methods are used to evaluate the mentioned algorithms; the first one is based on the cost function derived from a high-level analytical model of battery, and the second one is based on Dualfoil, a low-level li-ion battery simulator. Experimental results show that the system lifetime can be increased about 4.3% to 19.6%in various situations (in terms of system workload and tasks power consumption).