Optimal power and performance trade-offs for dynamic voltage scaling in power management based wireless sensor node

Dynamic voltage scaling contributes to a significant amount of power saving, especially in the energy constrained wireless sensor networks (WSNs). Existing dynamic voltage scaling techniques make the system slower and ignore the event miss rate. This results in degradation of the system performance when there is non-stationary workload at input. The overhead due to transition between voltage level and discrete voltage levels are also the limitations of available dynamic voltage scaling (DVS) techniques at sensor node (SN). This paper proposes a workload dependent DVS based MSP430 controller model used for SN. An online gradient estimation technique has been used to optimize power and performance trade-offs. The analytical results are validated with the simulation results obtained using simulation tool “SimEvents” and compared with the available AT9OS8535 controller. Based on the stochastic workload, the controller's input voltage, operational frequency, utilization, and average wait time of events are obtained.