Improved PSO algorithm integrated with opposition-based learning and tentative perception in networked data centres

© 2020 IEEE. Particle swarm optimization (PSO) algorithms have low-quality initial particle swarm, which is generated by a random method when handling the problem of task scheduling in networked data centres. Such algorithms also fall easily into local optimum when searching for the optimal solution. To address these problems, this study proposes combining opposition-based learning (OBL) and tentative perception (TP) with PSO; the proposed method is called OBL-TP-PSO. This algorithm uses reverse learning to generate the initial population, such that the quality of the initial particle swarm can be improved. Before the particle speed and location are updated, the TP method is used to search for the individual optimum around each particle, thereby reducing the possibility of missing the potential optimal solution during the process of searching. In this manner, the problem in which the PSO algorithm easily falls into the local optimal is effectively solved. To evaluate the performance of the proposed algorithm, simulation experiments are performed on CloudSim toolkit. Experimental results show that in comparison with other algorithms (namely, Min-Min, Max-Min and PSO algorithm), the proposed OBL-TP-PSO algorithm has better performance in terms of the total execution time, load balancing and quality of service.