Using Node Clustering and Genetic Programming to Estimate Missing Data for Marine Environmental Monitoring

Chen, X.; Jin, Z.; Lin, G., and Yang, X., 2015. Using node clustering and genetic programming to estimate missing data for marine environmental monitoring.

Missing sensor data is a challenge in wireless sensor network (WSN) for marine environmental monitoring, which may lower the performance of WSN. Therefore, estimation of the missing sensor data has received growing attention in WSN recently. In this paper we propose a novel data estimation approach based on Node Clustering and Genetic Programming (NCGP) to estimate the missing sensor data for WSN. NCGP consists of node clustering module and data estimation module. We first set up clusters according to data similarity and spatial similarity in the node clustering module, and then in the data estimation module, Genetic Programming is applied to mine the relationship among the nodes in the same cluster. Our proposed approach can efficiently retrieve the missing data by the help of the historical round reading and/or current round reading depend on the number of the nodes in the data missing cluster. The experiments have demonstrated the efficacy of NCGP in terms of five error statistics, wave plot and scatter plot compared to the existing techniques.