Quantitative Evaluation of Application Layer Protocols for Edge-Centric Real-Time IoVT Applications

Many applications are moving to edge computing in order to benefit from its enhanced performance and saved bandwidth. The overall performance and quality of data transmissions in these edge-based applications highly depend on the choice of application layer protocols. This is especially crucial for those dealing with continuous and high volumes of data transmissions such as Internet of Video Things (IoVT) applications. While previous works have investigated various trade-offs involved in choosing application layer protocols for IoT applications, their trade-offs for IoVT are overlooked. So the question still remains as to whether the same protocols observed to be well-suited for IoT would be appropriate choices for IoVT especially those with timing constraints in edge computing environments. In this paper, we present a quantitative evaluation of the performance, throughput, resource, and energy consumption of various protocols including HTTP (both HTTP/1 and HTTP/3 over QUIC), CoAP, WebSockets, AMQP, MQTT, DDS, and XMPP for real-time video data transmissions in an edge computing environment. We perform the comparisons while analyzing the sensitivity of the results with regard to video resolutions and transmission rates. Our comparisons reveal that promising protocols in IoT such as MQTT and CoAP, are not necessarily the best choices for real-time edge-based IoVT applications. The trade-offs discussed in this paper based on various metrics provide valuable insights for system architects dealing with latency sensitive edge-centric IoVT applications to choose the best protocol according to their system needs and constraints.