SRRA: A Novel Skewness-Based Algorithm for Cloudlet Scheduling

Cloud computing enables developers to deploy and host applications without focusing on installing and maintaining the infrastructure. The developers can utilize the services provided by the cloud service providers (CSPs) to offer scalable solutions to customer applications. As a result, CSPs are deluged with different batches of cloudlets (tasks) from diverse customer applications. Therefore, developing an algorithm that selects and processes applications intelligently to minimize the execution time and maximize the throughput becomes challenging. Many researchers have shown the round-robin (RR) scheduling algorithm variants to tackle this problem. One such variant is the dynamic RR heuristic algorithm (DRRHA) that utilizes the mean of the burst times (BTs) of cloudlets in the ready queue (RQ) to calculate the time quantum (TQ). However, DRRHA has not considered skewness. This paper introduces a novel skewness-based RR algorithm (SRRA) for cloudlet scheduling. The algorithm dynamically determines the TQ for each cloudlet based on the skewness of the BTs of cloudlets in the RQ. The algorithm has two variants: SRRA with minimum TQ (SRRA-Min) and SRRA with median TQ (SRRA-Med). The two variants of the proposed algorithm exhibit improved performance in terms of total execution time (TET) and throughput compared to DRRHA, individually and collectively. These comparisons are conducted using CloudSim Plus under two scenarios: constant skewness with varying cloudlets and constant cloudlets with varying skewness.