Energy neutral operation of WSNs can be achieved by exploiting the idlenessof the workload to bring the average power consumption of each node below theharvesting power available. This paper proposes a combination ofstate-of-the-art low-power design techniques to minimize the local and globalimpact of the two main activities of each node: sampling and communication.Dynamic power management is adopted to exploit low-power modes during idleperiods, while asynchronous wake-up and prediction-based data collection areused to opportunistically activate hardware components and network nodes onlywhen they are strictly required. Furthermore, the concept of "model-basedsensing" is introduced to push prediction-based data collection techniques asclose as possible to the sensing elements. The results achieved onrepresentative real-world WSN case studies show that the combined benefits ofthe design techniques adopted is more than linear, providing an overall powerreduction of more than 3 orders of magnitude.
展开▼
