Performance Analysis of One Tier and Two Tier Fuzzy Systems for Cluster Head Election in Wireless Sensor Networks

Sensor nodes are devices deployed in target environments to sense and forward data to base stations for different applications. These form wireless sensor networks (WSNs) applied in agriculture, health systems, underground and underwater monitoring, and body area networks. The nodes are battery-powered and, in most applications, not rechargeable. Consequently, the nodes no longer participate in the network once the energy is depleted. The small size of these nodes poses a challenge of limited energy storage, thus making research on energy efficiency a critical area in wireless sensor networks. Clustering techniques are a popular strategy in energy management, and recent research has utilized fuzzy logic in cluster head selection. One-tier and two tier fuzzy systems used in cluster head selection are compared. Four input variables, residual energy, centrality, move speed, and pause time, determine the chance of a node being elected as cluster head. The one-tier feeds all four inputs to one fuzzy inference system. In contrast, the two-tier feeds the speed and time into the first fuzzy inference system, giving an output of the mobility factor, then feeds this as input in addition to the residual energy and centrality to the second fuzzy inference system. Performance is measured based on the network lifetime, energy consumption, and fuzzy system complexity. The one-tier system is found to have higher accuracy for four input variables, leading to lower energy consumption but at the expense of higher design complexity.