The cost-aware dynamic voltage scaling approach for networked control systems

Due to the energy constraints, low-power control processor must operate in energy-efficient fashion. Dynamic Voltage Scaling (DVS) techniques has been introduced to reduce energy consumption of a processor by lowering supply voltage. However, for embedded controllers in a networked environment, minimizing energy consumption is more complex than non-networked control systems since all control loops share a communication network. In this paper, we address the problem of energy saving in a networked embedded controller. To reduce energy consumption of a embedded controller, we propose a Cost-Aware DVS (CA-DVS) approach that integrates static and dynamic techniques. The approach statically obtains the sampling period bounds that provide successful scheduling of all controller tasks running on the control processor and guarantee required control performance for each control loop. At the runtime, relative deadlines of controller tasks are dynamically assigned according to the performance of the corresponding control loops.