Performance Analysis and Machine Learning-based Classification of DDS vs MQTT IoT Communication Protocols

Selecting the right communication protocol is crucial for the performance and scalability of an IoT system. In this paper, we compare two popular publish-subscribe protocolsData Distribution Service (DDS) and Message Queuing Telemetry Transport (MQTT)-under identical settings, because we evaluate the protocols based on bandwidth usage, latency, reliability, and resource utilisation. Our experiments model realistic traffic rather than idealised ones, and we also create temporal and QoSinspired features to better characterise each protocol over time. We use multiple machine learning models, including ensembles, to show that protocol-specific traffic can be classified with extremely high accuracy, and bandwidth- and delivery-related features consistently have the largest impact on the predictions. Taken together, the results provide practical, data-driven guidelines for choosing and operating protocols in next-generation IoT deployments.