A Multi-objective Dynamic Scheduling Approach for IoT Task Offloading on Amazon EC2 Spot Instances

Nowadays, Internet of Things (IoT) applications have expanded to include smart cities, agriculture, e-health, industry, smart transport, etc. This large number of sensors and widespread applications will generate huge amounts of data that requires processing for analysis and decision-making. Therefore, considering the cloud computing model for the processing of IoT tasks with loose deadlines that do not require hard-real time processing can be an option. Generally, offloading tasks to the cloud will entail costs that must be optimized and reduced by intelligent mechanisms. Consequently, considering cloud computing instances with dynamic pricing referred to as spot instances can significantly reduce the processing costs. Although, these instances offer a considerable price saving compared to on-demand instances, they can be evicted by the cloud providers which poses a scheduling challenge. In this paper, we propose a dynamic scheduling method for IoT task offloading on Amazon EC2 spot instances. The proposed method considers the task's predicted execution time and deadline to specify tasks that can be mapped on spot instances. The experimental results denote that the proposed method leads to a considerable reduction in the execution costs while increasing the number of successful tasks executed before the deadline and decreasing task turnaround time.