WEIGHTED BOOST SPEARMAN CORRELATIVE DUAL CLUSTER HEAD FOR ROBUST TRANSMISSION IN WIRELESS SENSOR NETWORK

Abstract

Wireless Sensor Network (WSN) consists of tiny, low-powered sensors communicating with each other via multi-hop wireless links for efficient data transmission. Clustering of nodes is utilized in WSN to reduce number of nodes for transmitting data packets to the sink node. Several researchers have been carried out on dual cluster head selection for data transmission in WSN. In these cases, cluster member transmits data packets to Primary Cluster Head (PCH), followed by which it transmits to the Secondary Cluster Head (SCH). Finally, the SCH transmits the data packets to the base station. Despite improvement observed during data transmission, data packet delivery ratio and delay was less focused. In order to address these issues, a machine learning technique called, Weighted Boost Spearman Correlative Dual Cluster Head Selection (WB-SCDCHS) is proposed for performing resource optimized data transmission in WSN. With the initial number of sensor nodes being setup in WSN, residual energy and memory availability of every sensor node is computed. Here, Weighted Boost Agglomerative Brown Clustering (WBABC) model is employed to form strong cluster by combining weak learner, with Hierarchical Agglomerative Brown Cluster (HABC) considered as the weak learner. The HABC groups the nodes based on the residual energy and memory availability to form clusters. The second step forms the cluster head selection performed using Spearman Rank Correlative Coefficient, where sensor node with high residual energy and bandwidth availability is selected as the cluster head.  In every cluster, the sensor nodes are ranked by identifying the relationship between two variables (i.e., nodes) based on residual energy and bandwidth availability. The sensor node ranking first is considered as the primary cluster head and sensor node ranking second is considered as the secondary cluster head of that particular cluster. Finally, the secondary cluster head finds the neighboring secondary cluster head with higher bandwidth availability for transmitting data to base station (BS). This in turn helps to improve the packet delivery ratio. Experimental evaluation is conducted on factors such clustering accuracy, packet delivery ratio, delay, energy consumption and packet drop rate with respect to number of sensor nodes and packets.