Energy-Efficient Framework Based on Particle Swarm Optimization for Protection of Critical Infrastructures Using Wireless Sensor Networks

Abstract

Wireless sensor networks (WSNs) have seen increased application in recent years across medicine, the military, and the protection of critical infrastructures. A major challenge in their deployment is limited power supply, which significantly affects its operational lifetime. Several factors such as data transmission latency, packet loss ratio (PLR), and anomalies, also affect the energy-efficiency and by consequence, the lifetime of the WSN. This study proposes an energy-efficient framework based on Particle Swarm Optimization (PSO). The framework is developed in MATLAB, and is capable of detecting anomalies. The model uses Low-Energy Adaptive Clustering Hierarchy (LEACH) to assign cluster heads using k-means while anomaly detection was implemented using isolation forest and statistical z-score. Weibull is used to model wear-out pattern and signal strength while burst errors were modeled using Gilbert-Elliot. The PSO-based routing fitness function considers total energy consumption, packet loss and latency.  A cluster of 10 nodes, 15 nodes, 20 nodes, were considered. The performance of these clusters were compared with an unoptimized LEACH-based cluster of 10 nodes. A custom performance metric, energy-latency efficiency index (ELEI) per node is used to assess the performance of each cluster. The ELEI, expressed in Joules per second (J/s), is achieved by the 10, 15, and 20 nodes are 20.866, 11.436, and 78.827 respectively. Another metric, latency-normalized lifetime (LNL) at 215.78 indicates acceptable level of performance. The model performs comparatively well against bacterial foraging Optimization with Harmony Search Algorithm (BFO-HSA) and Trust Index optimized Cluster Head Routing (TIOCHR), for 100 nodes, in terms of latency.