An Intelligent MobileViT-Driven ISBOA Approach for Dynamic Load Balancing in Heterogeneous Cloud Infrastructures

The increasing demand for a wide range of applications in cloud computing systems has made it challenging to distribute workloads and resources, which has raised computational costs and energy usage. To maintain peak performance and optimize resource use, load balancing must be done effectively. It ensures that network traffic is distributed evenly among servers, avoids overload, speeds up response times, and increases system reliability. Furthermore, a precise workload prediction method is necessary to guarantee effective resource usage and adaptability. However, the highly dynamic and varied character of cloud workloads is frequently overlooked by current approaches, which results in uneven task distribution, longer makespan, higher migration costs, and decreased system efficiency. Therefore, this work presents a novel approach for workload prediction and load balancing in cloud networks, incorporating MobileViT- Improved Secretary Bobcat Optimization Algorithm (ISBOA) based hybrid method. In order to forecast future workloads, the suggested approach incorporates a MobileViT-based network that captures both local and global interdependence in dynamic task patterns. The load balancing is then carried out using a hybrid optimization approach that combines the Secretary Optimization Algorithm (SOA) and the Bobcat Optimization Algorithm (BOA). Experiments are conducted on two real-world datasets, HPC2N and NASA Ames iPSC/860, and the results are compared with existing techniques. The results demonstrate the effectiveness of the proposed technique in cloud environments by showing improved performance in terms of makespan, energy consumption, resource utilization, migration cost, and other error metrics.