Network operators must continuously explore new network architectures to satisfy increasing traffic demand due to bandwidth-hungry services, such as video-on-demand (VoD). A promising solution which enables offloading traffic consists of terminating VoD requests locally through deploying caches at the network edge. However, deciding the number of caches to deploy, their locations in the network and their dimensions in terms of storage capacity is not trivial and must be jointly optimized, to reduce costs and utilize network resources efficiently. In this paper, we aim to find the optimal deployment of caches in a hierarchical metro network, which minimizes the overall network resource occupation for VoD services, in terms of number of caches deployed across the various network levels, their locations and their dimensions (i.e., storage capacity), under limited storage capacity. We first propose an analytical model which serves as a tool to find the optimal deployment as a function of various parameters, such as popularity distribution and location of metro cache. Then, we present a discrete-event simulator for dynamic VoD provisioning to verify the correctness of the analytical model and to measure the performance of different cache deployment strategies in terms of overall network resource occupation. We prove that, to minimize resource occupation given a fixed budget in terms of storage capacity, storage capacity must be distributed among caches at different layers of the metro network. Moreover, we provide guidelines for the optimal cache deployment strategy when the available storage capacity is limited. We further show how the optimal deployment of caches across the various metro network levels varies depending on the popularity distribution, the metro network topology and the amount of storage capacity available (i.e., the budget invested in terms of storage capacity).
Analytical Model to Estimate Ride Pooling Traffic Impacts by Using the Macroscopic Fundamental Diagram
