An Efficient Tasks Offloading Procedure for an Integrated Edge-Computing Architecture

The advent of sixth-generation networks has given rise to numerous challenges, requiring the synergistic exploitation of both ground and air edge computing facilities. This paper considers an integrated ground-air edge computing scenario where the computation offloading of a set of delay sensitive tasks has to be performed in a context where ground and air computational facilities are already involved in monitoring and control proceduers in a remote area under an unpredictable overload of computation requests, e.g., related to the management of an emergency situation. In this reference, a matching game is proposed to assign tasks to the most suitable computation nodes, in order to minimize the outage probability of the newly arrived tasks, i.e., the probability with which tasks experience a completion time greater than the corresponding deadline. To this regard, we have considered that new allocated task suffer for a waiting time due to the time needed to complete the service of all the tasks already in the ground or air computation node. As a consequence, to statistically characterize such waiting time, under proper assumptions, we have resorted to the G/G/1 queuing system model and the Lindley’s integral equation approach to define a suitable metric to formulate a tasks allocation procedure based on the matching theory. Furthermore, matching stability has been theoretically proved for the proposed approach. Finally, numerical results have been provided in order to highlight the better behavior of the proposed task allocation scheme in comparison with different state-of-the-art alternatives.