Optimizing Resource Utilization and Improving Performance in Cloud Computing Through PSO-Based Scheduling and ACO-Based Load Balancing

Cloud computing is a technique based on the distribution of processing and storage resources among different servers. In this infrastructure, users can easily access cloud services and applications from anywhere and at any time. Although these capabilities have led to greater flexibility and availability of cloud services, issues such as scheduling and load balancing, have always been one of the main challenges in this infrastructure to achieve optimal resource utilization and provide server reliability. The process of load balancing and scheduling increases the optimal use of resources and provides a basis for improving performance. Even if the desired node is under low or high load, by using load balancing techniques, its utilization can be optimized. Therefore, in this research, a hybrid method based on Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) algorithm is used for load balancing and performance improvement. The PSO algorithm is used as a scheduling agent so that in the first step, tasks are scheduled among processing servers using this algorithm. Then, in the next step, use the ACO algorithm as a load-balancing agent. Furthermore, if required, it facilitates the execution of load-balancing operations. This solution not only enhances cloud infrastructure performance but also minimizes task execution time by preventing server congestion and the subsequent increase in task duration. Finally, the proposed solution was compared with Round Robin, Cat Swarm Optimization (CSO), Genetic Algorithm (G.A), and Ant Colony System Virtual Machine Consolidation (ACSVMC) methods, in a Cloudsim environment. It outperformed both alternatives, resulting in lower energy consumption and execution time. So, the energy consumption was lessened by 14% compared to ACS-VMC and G.A algorithms and more than 18% compared to Round Robin and CSO algorithms and also execution time was reduced by 15% compared to ACS-VMC and CSO and more than 28% compared to an R.R and G.A algorithm.