EDR2: A sink failure resilient approach for WSNs

Data collection, redistribution and retrieval are essential components of wireless sensor networks (WSNs). In dense WSN deployments, the sensor data are usually sent to a sink that can be reached through one or multiple hops. In the case where communications with the sink are disrupted due to various reasons, the data must be stored in the network for later retrieval. When considering in-network storage, we must redistribute the data among an energy-constrained network with sensors that have a low storage capacity. In previous works, the data redistribution problem has been studied, but the focus was only on the redistribution costs while the data retrieval costs (which have been analyzed in other works as an independent problem) were ignored. We recognize that these two problems should be studied in concert and therefore, in this paper, we combine both data redistribution and retrieval into a single problem. We propose a graph transformation, formulate the problem as a minimum cost flow optimization problem and use linear programming to find the optimal solution. Moreover, we introduce an algorithm named EDR²: energy-efficient data redistribution and retrieval. EDR² is a distributed energy-efficient algorithm for in-network storage and later retrieval in WSNs. To evaluate our solution on a large scale, we modeled different scenarios in a 400-node network, used the GNU Linear Programming Kit (GLPK) to obtain the optimal solutions, and ran simulations to find the solutions using our algorithm. Finally, we implemented EDR² using real sensors to demonstrate the feasibility of our algorithm. We compared EDR² with two heuristic algorithm and show that our approach is an energy-efficient solution for node selection when redistributing data in a WSN for eventual retrieval.