Title: Optimised load balancing in cloud environments: a hybrid zebra and horse optimisation algorithm integrated with deep reinforcement learning

Authors: Mohammad Sharfoddin Khatib; Nazish Khan; Sadia Patka

Addresses: Department of Computer Science and Engineering, Anjuman College of Engineering and Technology, Sadar, Nagpur – 440001, Maharashtra, India ' Department of Computer Science and Engineering, Anjuman College of Engineering and Technology, Sadar, Nagpur – 440001, Maharashtra, India ' Department of Computer Science and Engineering, Anjuman College of Engineering and Technology, Sadar, Nagpur – 440001, Maharashtra, India

Abstract: Efficient load balancing is crucial in cloud computing (CC) to optimise resource utilisation and minimise delays. The proposed multi-objective hybrid zebra-horse optimisation algorithm-based load balancing mechanism (HZHOA-LB) integrates evolutionary optimisation with deep reinforcement learning, unlike existing methods such as HDWOA-LB, FIMPSO-LB, QODA-LB, and MMHHO-LB, which rely solely on metaheuristics. By combining zebra foraging for exploration and horse defensive mechanisms for exploitation, HZHOA-LB ensures an optimal balance between global and local search. Additionally, deep Q-network (DQN)-based decision-making enables adaptive load balancing by treating it as a Markov decision process (MDP), allowing real-time workload adjustments. This hybrid approach achieves a lower makespan (395 ms), reduced energy consumption (48.57 J), improved response time (12.41 ms), and optimised resource utilisation, significantly outperforming existing techniques in dynamic cloud environments.

Keywords: cloud computing; load balancing; multi-objective optimisation; hybrid zebra-horse optimisation algorithm; HZHOA; deep Q-network; DQN; degree of imbalance; DoI.

DOI: 10.1504/IJCC.2026.154437

International Journal of Cloud Computing, 2026 Vol.15 No.2, pp.181 - 205

Received: 15 May 2024
Accepted: 05 Mar 2025
Published online: 29 Jun 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article