On reliable data delivery in stochastic energy harvesting wireless sensor networks

In wireless sensor networks (WSNs), sensor nodes with energy harvesting components have motivation to expend additional energy in conditions of excess, since the surplus energy would be wasted due to energy buffer overflow. In this paper, we focus on using such surplus energy to adjust the redundancy level of erasure codes, so that the data delivery reliability can be improved while the network lifetime is still well conserved. For a single flow, we formulate the problem as the maximisation of end-to-end packet delivery probability under energy constraints. Considering the energy profile as a stochastic process, we propose a Lyapunov optimisation based algorithm called the erasure coding scheduling algorithm (ECSA) to solve the problem. Through a combination of both theoretical analysis and simulation, we show the effectiveness of ESCA in yielding a near-optimal data delivery reliability.