Outlier analysis-pattern mining (OAPM) algorithm in wireless sensor networks

In this work, a novel clustering and aggregating algorithm is proposed to reduce energy consumption in wireless sensor networks (WSN) by focusing on regulating the inter-cluster traffic and data aggregation at the cluster heads. Previous works have aimed at reducing the intra-cluster traffic by choosing the node at the mean position inside as cluster head. But intra-cluster distance is small compared with inter-cluster distance. Hence reducing inter-cluster traffic will have a greater impact in power optimization. In our approach, inter-cluster traffic is reduced by clustering where cluster heads are chosen by outlier analysis. To reduce energy consumption further, we have implemented data aggregation using pattern mining at the cluster heads. Aggregation techniques proposed so far are based on probability and they arrive at implementing a static timer for data aggregation. Because of varying network traffic, a static timer results in constant delay. Implementing a dynamic timer is a possible solution for improvising the performance of data aggregation. Outlier analysis together with aggregation using pattern mining helps to reduce energy consumption.