Cloudlet Scheduling with Particle Swarm Optimization

Abstract

Cloud computing is a particularly interesting and truly complex concept of providing services over networks on demand. Many tools have previously been created to simulate the work of the clouds, such as CloudSim. The main use of these tools is evaluation and testing of architectures and services before deployment on network centers and hosts in order to avoid any potential failures or inconveniences. The benefits of using the pay-per-use clouds may be affected by underutilization of the already reserved resources, so the maximization of system utilization while simultaneously minimizing makespan is of great interest to cloud providers wishing to reduce costs. To minimize makespan and increase resource utilization, a hybrid of particle swarm optimization and simulated annealing is implemented inside of CloudSim to advance the work of the already implemented simple broker. The new method maximizes the resource utilization and minimizes the makespan.