Dynamic RL-ACO: Reinforcement Learning-based Ant Colony Optimization for Load Balancing in Cloud Networks
Satyanarayana Nimmala, M Ramchander, Maragoni Mahendar, Pinnapureddy Manasa, D. Durga Bhavani, K Raghavendar
Abstract
Effective load balancing is essential in cloud computing to maximize resource usage and guarantee uninterrupted service delivery. The dynamic nature of cloud environments is a problem that traditional methods frequently overlook. To improve load balancing in cloud networks, this paper presents Dynamic RL-ACO, which combines Ant Colony Optimization (ACO) and Reinforcement Learning (RL). Real-time load distribution strategies are predicted by RL, and task allocation is dynamically adjusted by ACO, which emulates the efficient resource use of ants during foraging. In his paper, we have used the CloudSim simulations with the Google Cluster Data to verify the effectiveness of Dynamic RL-ACO. Dynamic RL-ACO achieves a 15 percent reduction in response time, 20 percent improvement in resource utilization, 25 percent faster adaptation to workload changes, and 15 percent reduction in energy consumption when compared to traditional algorithms like Round Robin, Least Connections, and static ACO. These outcomes demonstrate the scalability and resilience of Dynamic RL-ACO, highlighting its potential as an effective, adaptable, and resilient load-balancing solution for contemporary cloud computing environments