Multi-objective Load Balancing in Cloud Computing Environment Using Efficient Deep Reinforcement Learning with Hybrid Optimization Algorithm

In the dynamic era of cloud computing, load balancing is crucial for distributing workloads across multiple resources to optimize performance, utilization and reliability. Multi-objective mechanism considered various criteria’s like, quality of services, reliability, cost efficiency and performances. Existing models has some issues like, insufficient utilization of computational resources leads to some serves being overloads, poor performances, high energy consumption, increased operational cost and environmental impacts. In order to solve these existing issues, a novel deep reinforcement learning (DRL) with hybrid optimization algorithm (Hy-Coop) for load balancing to solve issues like complexity in high dimension. This proposed approach enhance accuracy and speed of the model. The combination of coati and osprey algorithm effective in balancing exploration and exploitation stage. The proposed model outperforms existing models in terms of throughput, achieved higher values across all task ranges. The proposed model reached 3.3 task/s, significantly higher than the CO algorithm at 1500 tasks. When compared to existing models, the proposed approach obtained high reward values of 35%. The proposed model obtained degree of imbalance (DOI) of 28.012, energy consumption of 7.564 kJ.