No abstract

Solar power can extend the lifetime of wireless sensor networks (WSNs), but it is a very variable energy source. In many applications for WSNs, however, it is often preferred to operate at a constant quality level rather than to change application behavior frequently. Therefore, a solar-powered node is required adaptation to a highly varying energy supply. Reconciling a varying supply with a fixed demand requires a good prediction of that supply, so that demand can be regulated accordingly. We describe two energy allocation schemes, based on time-slots, which aim at optimum use of the periodically harvested solar energy, while minimizing the variability in energy allocation. The simpler scheme is designed for resource-constrained sensors; and a more accurate approach is designed for sensors with a larger energy budget. Each of these schemes uses a probabilistic model based on previous observation of harvested solar energy. This model takes account of long-term trends as well as temporary fluctuations of right levels. Finally, this node-level energy optimization naturally leads to the improvement of the network-wide performance such as latency and throughput. The experimental results on our testbeds and simulations show it clearly.

No abstract

No abstract

No abstract