Cloud-aided edge caching with wireless multicast fronthauling in fog radio access networks

Abstract

In this paper, we investigate the total delivery latency across the fronthaul and wireless segments of a Fog Radio Access Network (F-RAN) under the assumption that cloud processor and edge nodes (ENs) are connected by a multicast fronthaul link. The total delivery latency is assessed via the Normalized Delivery Time (NDT) metric which provides a high signal-to- noise ratio (SNR) measure of the relative delivery worst-case latency with respect to an interference-free system. We derive upper and lower bounds on the achievable NDT for a F-RAN with two ENs and two users as a function of cache and fronthaul resources. The upper bound is obtained by studying the NDT achieved by delivery strategies that encompass both coded and uncoded multicast strategies on the fronthaul. The lower bound is instead derived by leveraging information theoretic converse arguments. Upper and lower bounds are shown to coincide, hence characterizing the minimum NDT, for a large range of problem parameters. Among the conclusions of this study, we demonstrate that coded multicasting is not useful for reducing the NDT for the mentioned range of parameters.