A Hierarchical Multi-Agent Reinforcement Learning With a Heterogeneous Metaheuristic Aware Resource Allocation in Big Data-Cloud

Abstract

The rapid advancement of technology like Internet of Things (IoT) and Cloud computing (CC)
based heterogeneous environment required dynamic resource management system. The
complexity of IoT-Cloud is increasing due to abundance of dynamic data dissemination that
create poor performance like high energy consumption (EC), workload imbalancing, poor
adaptability, and fail to handle SLA violations. The two primary contributions of the proposed
work are the Optimized Priority-Aware Hierarchical Multi-Agent Deep Q Network (OPHMDQN)
and the Adaptive Multi-Objective Dung Beetle Optimization Algorithm (ADBOA). The
multi-agent strategies increase the scalability and adaptability of resource management
through local and global hierarchies. The approach integrates information-based decisionmaking
and priority-aware allocation while accounting for SLA requirements, system
constraints, and job complexity to optimise resource generation, utilisation, allocation, and
task scheduling. In comparison to existing optimization and Reinforcement Learning (RL)
techniques, experimental results show that the proposed OPHM-DQN-ADBOA framework
consistently reduces EC (up to 30 % lower), execution delay, and SLA violations while
improving resource utilisation and LB. The ADBOA enhances the proposed model through
optimal multi-objective training, reducing EC, SLA violation, and cost while improving
resource utilization and scheduling efficiency. As a results, the model achieves high scalability
and adaptability in heterogeneous IoT-Cloud resource management.