Towards Optimising Transport Protocols on the 5G Edge for Mobile Augmented Reality

Mobile augmented reality (MAR) achieves immersive real-time experiences by offloading computation-intensive computer vision tasks. 5G NR (5G) networks and edge computing enable optimised latency and enhanced throughput for MAR offloading. However, efficiently leveraging the 5G edge also requires optimising the data link between MAR devices and server machines. We report preliminary experiments of optimising transport protocols (i.e., UDP and TCP) to understand how we could further modify the data link between MAR clients and servers. Preliminary analysis shows that when using a real-world 5G testbed, our evaluation indicates that for our settings, TCP with default configuration parameters has the lowest round-trip time on 5G, with a median of 15.8\pm10.3 ms. Then by increasing the protocol buffer sizes to 100 KB and 1000 KB, the packet latency and jitter decrease while throughput increases for some media resolutions. We further discuss optimising transport protocol parameters could improve the service quality and experience for MAR applications.