A Multi-Path Approach to Deterministic Backhaul for Geographically Distributed 5G Campus Networks

The number of localized 5G campus network deployments is growing. For remote deployments, e.g. rural factories, it is not feasible to lay fiber cables. To try to add to their backhaul options, operators may start using commercial satellite internet connections or Fixed Wireless Access (FWA). Thus leaving them with multiple backhaul options. However, all of these connections are still best effort transport networks, unsuited for critical applications, and unable to provide determinism. This paper presents an investigation into the use of multiple paths to provide deterministic backhaul, in geographically distributed 5G Campus deployments, over unreliable shared transport networks.Our contributions are twofold. First, we present a design and an implementation of an approach which takes a 5G flow’s requirements and calculates which backhaul paths to use, via a binary optimization equation. Second, we propose a deployment architecture designed specifically for use in distributed 5G Campus Networks. This solution is investigated, on a custom testbed which emulates multiple satellite backhaul links. Through it’s novel use of multi-path packet replication in 5G backhaul, our approach was able to demonstrate a greater performance than an oracle approach for single link selection. It increased the proportion of time during which a test flow’s latency, jitter, and packet loss requirements could be met by 35%, 30%, and 200% respectively.