Volunteered mobile sourcing with multi-objective ant colony optimization

Volunteered computing has been one of the popular distributed computing concepts recently. The basic idea is to allow computer owners to donate the computing power and storage to scientific applications. In this research, we are interested in the utilization of volunteered mobile devices. The implementation of such a concept is complicated since it is hard to accurately estimate the execution time of workflow tasks on numerous mobile devices. To efficiently schedule application workflows can thus be a real challenge. In this paper, we proposed a practical way to construct a workflow with estimated overhead and execution time, as well as a scheduling algorithm for a highly distributed computing platform. The main idea is to effectively optimize task scheduling onto the currently available mobile devices with two objectives of maximizing both cost and execution time saved. Therefore, the cost will be covered by the volunteers. We adapt the Multi-objective Ant Colony Optimization (MOACO) algorithm in our framework. We perform an experiment with different sizes of scientific workflows under different numbers of volunteered devices. The results show a good potential in using mobile sources to minimize the energy consumption at the data center while keeping the execution time within a reasonable deadline.