Multi-Objective Particle Swarm Optimization with Energy-Density Centroid Based Deputy Cluster Head Selection Scheme for IoT based Sensor Networks

Abstract

In today's world, most applications are equipped with IoT. Wireless sensor networks are the major component in IoT. Generally, the IoT-based sensor nodes can be sensing and convey the information to a base station via a wireless channel. Forest fire monitoring system is one of the safety-critical applications which can implement the fire monitoring process using IoT based sensor networks. Energy preservation is the major open issue in forest fire monitoring system. Usually, in sensor environment, the information exchange between the sensor nodes to the base station via multi-hop network with multiple cluster heads. Thus, the cluster head can have an extra burden to manage this type of heavy data load sometimes the cluster head node becomes a dead state or data loss/delay. In a forest fire monitoring, every sensor node’s lifetime is crucial because the timely report from a cluster head node to the base station can make a quality of a service otherwise it may happen ecological or environmental damage. Energy preserved cluster head is an important role for better routing in a forest fire monitoring. The first phase of this paper provides an energy-efficient clustering and cluster head selection using Multi-Objective Particle Swarm Optimization (MOPSO). From the previous studies, the MOPSO is efficient for cluster head selection but the heavy density area creates a more number of cluster heads in a specific area. The numbers of cluster heads are the major consideration because it may lead to consuming high energy. Thus to prevent the exceeding of the number of cluster heads, the new Energy-Density Centroid based Deputy Cluster Head Selection Algorithm (EDCDCHsa) is proposed in a second phase. This proposed can able to select the cluster head with 5% of the total number of nodes in an efficient way. The proposed method can reduce the number of cluster head selection and the data load of the cluster head. This proposed algorithm is evaluated in a simulation part with state-of-the-art protocols and gives better efficient, reliable, and cost-effective in a forest fire monitoring system than the present approaches for cluster head selection.