Non-preemptive chaotic cat swarm optimization scheme for task scheduling on cloud computing environment

With exponential growth in the number of customers accessing the cloud services, scheduling tasks at cloud datacenter poses the greatest challenge in meeting end-user's quality of service (QoS) expectations in terms of time and cost. Recent research makes use of metaheuristic task scheduling techniques in addressing this concern. However, metaheuristic techniques are attributed with certain limitation such as premature convergence, global and local imbalance which causes insufficient task allocation across cloud virtual machines. Thus, resulting in inefficient QoS expectation. To address these concerns while meeting end-users QoS expectation, this paper puts forward a non-preemptive chaotic cat swarm optimization (NCCSO) scheme as an ideal solution. In the developed scheme, chaotic process is introduced to reduce entrapment at local optima and overcome premature convergence and Pareto dominant strategy is used to address optimality problem. The developed scheme is implemented in the CloudSim simulator tool and simulation results show the developed NCCSO scheme compared to the benchmarked schemes adopted in this paper can achieve 42.87%, 35.47% and 25.49% reduction in term of execution time, and also 38.62%, 35.32%, 25.56% in term of execution cost. Finally, we also unveiled that a statistical significance on 95% confidential interval has shown that our developed NCCSO scheme can provide a remarkable performance that can meet end-user QoS expectations.