An improved human memory algorithm with multi-directional and chaotic approaches for global optimization and energy-efficient cluster head selection in WSNs.

Authors

Mahmoud Abdel-Salam, Wael A Gab-Allah, Eman Mohamed Eldaydamony, Ahmed Atwan

Published in

Scientific reports. Volume 16. Issue 1. Jul 16, 2026. Epub Jul 16, 2026.

Abstract

Wireless Sensor Networks (WSNs) play a crucial role in infrastructure monitoring across domains such as smart grids, industrial automation, and environmental sensing. However, energy efficiency remains a key challenge due to the limited battery life of sensor nodes. This work addresses the energy-efficient cluster head (CH) selection problem, formulated as a dynamic optimization task. Human Memory Optimization (HMO) is chosen as the foundation for its low memory footprint and adaptive learning, which align well with WSN constraints. Its memory-based recall mechanism naturally balances exploration and exploitation with minimal parameter tuning, making it suitable for decentralized CH decisions. On the other hand, traditional metaheuristic algorithms (MAs) suffer from premature convergence and fall into local optima, which makes them less effective in addressing the dynamic and energy-sensitive nature of CH selection in WSNs. To overcome these challenges, this work develops an improved version of the HMO, named Adaptive Enhanced Human Memory Optimization (AEHMO). The proposed AEHMO integrates four targeted strategies tailored for the WSN CH selection context: (1) Adaptive parameters dynamically adjust exploration and exploitation phases, enabling the algorithm to respond to changing node energy levels and topological shifts; (2) the Multi-Directional Mutation Strategy (MDMS) increases the diversity among CH candidates, ensuring broader spatial coverage and preventing clustering imbalances; (3) Dynamic Drift Search (DDS) enhances global exploration, allowing AEHMO to discover energy-efficient CH configurations across the entire network field early in the process; and (4) Chaotic Reverse-based Learning (CRL) introduces structured randomness to help the algorithm escape from suboptimal CH arrangements that could otherwise lead to rapid energy depletion or coverage holes. AEHMO was validated on the CEC2017 benchmark suite, showing superior performance in high-dimensional optimization. Applied to WSN CH selection, it achieved an average energy consumption of 0.395 J, FND at 1150 rounds, HND at 1853, and LND at 3136 rounds in the 150-node scenario. Furthermore, AEHMO was validated under a large-scale real-world heterogeneous WSN scenario comprising 1200 sensor nodes with a three-tier node architecture and a mobile sink strategy, achieving an FND of 362 rounds, HND of 1903 rounds, and LND of 56,946 rounds-outperforming all competing algorithms in total network lifetime. These results confirm AEHMO's superiority in energy efficiency, scalability, and adaptability under both homogeneous and heterogeneous dynamic WSN conditions.

PMID:
42463783
Bibliographic data and abstract were imported from PubMed on 17 Jul 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.