Using Data Recovery Models for Multi-shift Scheduling in Wireless Sensor Networks

Energy efficiency is a key problem in wireless sensor networks. Keeping only a portion of nodes active and putting the others into sleep mode can conserve energy. In order to maintain satisfactory data quality, we recover the would-be sensed data for sleeping nodes. In this paper, we exploit spatial correlation among densely deployed nodes and use such correlation for data recovery. Loss in data quality caused by data recovery is the criterion for our proposed polynomial-time active nodes selection algorithm. Considering the possible energy consumption unbalance, we further develop a multi-shift scheduling scheme. The multi-shift scheduling is formulated as a constrained mini-max optimization problem. We validate these algorithms using a real-world data set and observe very satisfactory results.