NFV Placement in Resource-Scarce Edge Nodes

Multi-access Edge Computing (MEC) is a new networking paradigm considered to be one of the enablers of 5G networks. In particular, it allows for network operators to provide low latency services by moving the service logic from centralized datacenters to small distributed locations at the edge of a network. However, computing and storage resources at these edge nodes are scarce and thus efficient resource allocation becomes an essential building block in any MEC orchestration solution.In this paper we address one particular challenge in this domain - how to place network functions at the edge nodes in a way that maximizes the customers benefit. Thus, we formulate the Capacitated MEC Allocation Problem (CMAP) and provide multiple algorithms with analytical performance guarantees for this problem. Furthermore, we use extensive simulations to evaluate the performance of our algorithms in realistic scenarios and show that they outperform both the analytical worst case guarantees, as well as currently used network function placement methods.