Practical Criteria for Scheduling CPU-Bound Jobs in Mobile Devices at the Edge

The ubiquity of mobile devices, their strong computational capabilities -frequently underused by the owners- and their close relationship to IoT environments, make them valuable providers of computing power at the edge, i.e. as an aid to deal with remote computing servers high-delay communication. Then, researchers have proposed to take advantage of the connection regularity of groups of mobile devices to fixed wireless networking infrastructures, e.g., access points, to scavenge mobile computing resources. However, among the key concerns to achieve such exploitation in an efficient way is providing new scheduling criteria for computations able to deal with the challenges posed by mobile devices finite energy sources. The value of considering battery-related parameters for ranking devices capabilities against not using them has been hardly investigated in the literature. Then, in this paper, we analyze several job scheduling criteria and provide statistical evidence that support our findings. The experiments are performed via a simulation software that employs energy consumption traces derived from real mobile devices, hence practical utility is ensured.