An efficient data aggregation and outlier detection scheme based on radial basis function neural network for WSN

Wireless sensor network (WSN) is used for data collection and transmission in IoT environment. Since it consists of a large number of sensor nodes, a significant amount of redundant data and outliers are generated which substantially deteriorate the network performance. Data aggregation is needed to reduce energy consumption and prolong the lifetime of WSN. In this paper a novel data aggregation scheme is proposed which is based on modified radial basis function neural network to classify the collected data at cluster head and eliminate the redundant data and outliers. Additionally, cosine similarity is used to cluster the nodes having the most similar data. The radial basis function (RBF) is adapted by Mahalanobis distance to support the outlier's detection and analysis in the multivariate data. The data collected from the sensor node at the cluster head are processed by mahalanbis distance-based radial basis function neural network (MDRBF-NN) before transferred to the based station. Extensive computer simulation with real datasets shows that the proposed scheme consistently outperforms the existing representative data aggregation schemes in terms of data classification, outlier detection, and energy efficiency.