Service Allocation in a Mobile Fog Infrastructure under Availability and QoS Constraints

Abstract

The next generation of mobile networks, namely 5G, and the Internet of Things (IoT) have brought a large number of delay sensitive services. In this context Cloud services are migrating to the edge of the networks to reduce latency. The notion of Fog computing, where the edge plays an active role in the execution of services, comes to meet the need for the stringent requirements. Thus, it becomes of a high importance to elegantly formulate and optimize this problem of mapping demand to supply. This work does exactly that, taking into account two key aspects of a service allocation problem in the Fog, namely modeling cost of executing a given set of services, and the randomness of resources availability, which may come from pre-existing load or server mobility. We introduce an integer optimization formulation to minimize the total cost under a guarantee of service execution despite the uncertainty of resources availability.