Energy and Security Awareness Task Scheduling based on Fuzzy System in Cloud Computing

The increasing popularity of cloud computing environments makes task scheduling as a critical problem and a hot research topic. It is necessary to decrease the energy related costs and enhance the lifespan of high performance computing resources used in cloud data centers. Moreover, the high quality of security service is increasingly critical for security-sensitive applications that work with large-scale data files such as bioinformatics. We propose a new task scheduling algorithm that includes: 1) analyzing task execution time based on the load of data centers; 2) modeling the resource utilization; 3) calculating security cost based on the failure probabilities; 4) evaluating power consumption based on the linear model; and 5) analyzing the closeness centrality of data centers to improve data retrieval time. Finally, it designs a fuzzy inference system with five inputs (i.e., total execution cost, resource utilization cost, security cost, energy consumption, and centrality) in order to assign a merit value to each data center for task execution. Cloud is a dynamic environment and there is not accurate information at every moment. Therefore, fuzzy inference is a good choice for predicting the behavior of the system and scheduling decisions. The simulation results indicate that the proposed algorithm obtains superior performances respectively in waiting time, success rate, energy consumption, and degree of imbalance around 14%, 12%, 15%, 11% on average than other similar methods under high load condition. Consequently, the proposed strategy has potentials to enhance the performance of QoS delivery since it can effectively utilize cloud resources.