Adaptive discrete clustering based whale optimization for energy efficient cluster head selection in internet of things-enabled wireless sensor networks

Abstract Repeated reception, aggregation, and forwarding place an energy burden on cluster heads in Internet of Things-enabled wireless sensor networks. We address fixed-cardinality cluster-head selection with an Adaptive Discrete Clustering Based Whale Optimization Algorithm (AD-CWOA). The method retains standard Whale Optimization Algorithm search dynamics and adds random-key decoding, progress-adaptive mutation, cluster-head swap refinement, and stagnation-triggered population restoration for discrete subset search. A four-component minimization objective evaluates residual-energy risk, intra-cluster distance, cluster-head-to-base-station distance, and cluster-load imbalance. The major-revision evaluation comprises 300 simulation executions: 140 fitness-weight sensitivity runs, 60 cluster-head-count runs, and 100 five-method comparison runs, with 20 paired 200-node topologies per condition. At K = 15, AD-CWOA gives FND, HND, and LND values of 1823.80 ± 63.64, 2191.00 ± 13.48, and 2511.80 ± 51.35 rounds, respectively, and cumulative source-data-equivalent throughput of 436820.25 ± 1889.11 packets. Relative to Original CWOA, FND increases by 3.92%, but LND decreases by 1.77%. The first node failure is delayed, although the network reaches complete depletion sooner. None of the twelve Holm-adjusted comparisons with Particle Swarm Optimization is statistically significant. The sensitivity experiments show that the same configuration does not maximize lifetime, stability, and communication efficiency. AD-CWOA offers an energy-aware CH-selection option for the tested network, with gains that depend on the chosen metric and operating configuration.