An Energy Efficient Expert System to Choose Cluster Head from Hybrid Clustering
Girija M.S. and Tapas Bapu B.R.
Wireless Sensor Networks monitor dynamic environments that change rapidly over time. To adapt to such conditions, Sensor networks often adopt machine learning techniques. Machine learning in wireless sensor networks provide technical solutions to optimize the performance of Wireless sensor networks by applying statistical and mathematical techniques. The main challenge in wireless sensor networks is to improve the lifetime of the network. Lifetime depends on the energy utilization of the sensor nodes. Most of the energy of the sensor nodes is used for data transmission. Clustering is a technique used to handle energy utilization efficiently. Hence the proposed method integrates hybrid clustering techniques, which includes K-means clustering and Naïve Bayes classification algorithm, to group the given sensor nodes into clusters. Subsequently the cluster head is chosen by performing Principal Component Analysis among the nodes in the same cluster by computing principal component score value for each and every node. Thus, sensor nodes are given a chance to become cluster head based on the high PCA score value, on a rotation basis by analyzing the distance and remaining residual energy of the sensor nodes. The network lifetime and energy consumption ratio is compared and evaluated against LEACH algorithm.
Keywords: Cluster Head, K-means Clustering, Machine Learning, Naïve Bayes classification, Principal Component analysis (PCA), Wireless Sensor Networks (WSN)