A Dynamic Task Scheduling Algorithm Based on Learning Automata for Cloud Computing

In the realm of cloud computing, suboptimal usage of system resources and an imbalance in task scheduling can result in inefficient resource utilization and system overload. In response to these challenges, we introduce the Dynamic Task Scheduling Algorithm based on Learning Automata (DTSALA). This algorithm harnesses the adaptive learning mechanism inherent in Learning Automata (DTSALA). Throughout the iterative looping process of the algorithm, the action probability distribution model undergoes continuous updates triggered by reward and punishment signals. The adaptive learning mechanism serves to mitigate resource wastage and prevent system overload by dynamically adjusting task scheduling actions based on prevailing system conditions. To assess the efficacy of the DTSALA algorithm, we conducted simulations comparing its performance with that of other widely used task scheduling algorithms, such as Max-Min, Min-Min, Genetic Algorithm (GA), Whale Optimization Algorithm (WOA), and Hybrid Whale Genetic Algorithm (HWGA). The simulation results unequivocally demonstrate the superior performance of the DTSALA algorithm, showcasing improvements in load balancing rate, task execution time, and system resource utilization when compared to these algorithms. By achieving a heightened load balancing rate, the DTSALA algorithm ensures equitable distribution of tasks across available system resources, thereby optimizing resource utilization. Furthermore, it reduces task execution time through efficient allocation of tasks to appropriate resources, ultimately enhancing the overall system performance and user experience within the cloud computing environment.