Scalable Server Location for Distributed Interactive Applications Under Budget Constraints

Mobile Edge Computing (MEC) has emerged as a popular paradigm for the support of distributed interactive applications (DIAs), promising continuous service provision and availability to mobile users, that is otherwise not easily attainable under the centralized cloud-only model. Yet, in order to alleviate the app provider from unnecessary capital expenditure, a judicious location of the edge servers must be adopted, that can effectively cope with both the network demands and the deployment cost, and still abide by the budget constraints of the app provider. In this paper, we formally formulate the problem as a budget-constrained facility location problem. We then propose a distributed and scalable algorithm, that employs elastic server relocation, to address it. Trace-driven simulations under various MEC deployment scenarios verify the expected behavior and testify to the algorithm’s accountability in obeying the budget limits and converging to a placement solution close to the optimal in finite time slots, rendering it suitable for DIAs ecosystems.