Principal Component Analysis based Clustering Approach for WSN with Locally Uniformly Correlated Data
Zakia Jellali, Leı̈la Najjar Atallah, Sofiane Cherif · 2019
Wireless Sensor Networks (WSN) is an effective data collection technology, which is increasingly recognized to be a key competency in Internet of Things (IoT) paradigm since its inception. In WSN deployment, it is interesting to reduce energy consumption in order to prolong network lifetime. Cluster based data aggregation is one of the most effective approaches for conserving energy in WSN. The data aggregation is processed at the cluster head node which relays its cluster nodes data towards the sink while exploiting some compression or aggregation rule based on these data correlation. However, cluster formation protocols either assume a uniform nodes data correlation model throughout the network or consider criteria such as energy consumption or data delivery latency. In this paper, we propose an effective nodes clustering scheme for clustering in heterogeneous WSN. Heterogeneity refers here to data correlation which is realistically supposed not uniform over the network. The scheme aims to split the large network into irregular sub-networks, regarded as blocks. In each of such blocks, the nodes measurements are as uniformly correlated as possible. Our contribution is aligned with Principal Component Analysis (PCA) technique by exploiting the spatial variation of data correlation. The obtained evaluation results confirm that our proposed clustering approach outperforms the existing clustering schemes in terms of partition accuracy and fidelity.