Signal Reconstruction from Rechargeable Wireless Sensor Networks using Sparse Random Projections

摘要

Due to non-homogeneous spread of sunlight, sensing nodes possess non-uniformenergy budget in recharge- able Wireless Sensor Networks (WSNs). Anenergy-aware workload distribution strategy is therefore nec- essary to achievegood data accuracy subject to energy-neutral operation. Recently proposedsignal approx- imation strategies assume uniform sampling and fail to ensureenergy neutral operation in rechargeable wireless sensor networks. We proposeEAST (Energy Aware Sparse approximation Technique), which ap- proximates asignal, by adapting sensor node sampling workload according to solar energyavailability. To the best of our knowledge, we are the first to propose sparseapproximation to model energy-aware workload distribution in rechargeable WSNs.Experimental results, using data from an outdoor WSN deployment suggest thatEAST significantly improves the approximation accuracy offering approximately50% higher sensor on-time. EAST requires the approximation error to be knownbeforehand to determine the number of measure- ments. However, it is not alwayspossible to decide the accuracy a-priori. We improve EAST and propose EAST+,which, given only the energy budget of the nodes, computes the optimal numberof measurements subject to the energy neutral operation.