Intelligent Workflow Scheduling for Energy-efficient and Electricity Price-aware Geo-distributed Cloud Data Center Using a Deep Learning Framework with a Novel Loss Function

Various services are provided to the users using cloud computing. The rapid growth of cloud computing leads to increased power consumption. Workflow scheduling is complex due to the dynamic and heterogenous resources in cloud environment. Several methods are used to improve the workflow scheduling but it consumes computational resources and energy for handling the different operational demands of data center. To achieve energy-efficient and electricity price-aware workflow scheduling in geo-distributed cloud data centers, this paper introduced a Taylor-based Exponential Hare Belief Network (TEXHB-Net). Initially, multiple objective parameters, such as electricity price, transmission delay, and energy consumption are considered by a Deep Belief Network with a Taylor-based Exponential Log-cosh loss function (DBN_TaELg) to handle the data center selection process. Then, TEXHB-Net handles task allocation to virtual machines by contemplating energy consumption, task priority, actual task completion time, task slack time, and reliability. Energy consumption of 14.998 J, resource utilization of 97.899%, task completion time of 5.489 s, and a delay of 0.179 s are achieved by the TEXHB-Net model.