Modeling and performance evaluation of DTN-integrated publish–subscribe middleware for UAV-assisted IoT: MQTT vs DDS

Unmanned Aerial Vehicle (UAV)-assisted Internet of Things (IoT) data collection faces fundamental challenges arising from intermittent connectivity, constrained energy budgets, and the degradation of publish–subscribe middleware performance due to opportunistic aerial communication patterns. This study presents a simulation-based performance evaluation of integrating Delay-Tolerant Networking (DTN) routing with two publish–subscribe protocol stacks, Message Queuing Telemetry Transport (MQTT) and the Data Distribution Service/Real-Time Publish–Subscribe (DDS/RTPS), for opportunistic aerial networks. We develop a discrete-event simulator that models UAV mobility, dual-tier ZigBee/Wi-Fi communications, link capacity estimation, per-bit energy accounting, and byte-accurate protocol overhead for both middleware stacks, including reliability overheads and DDS message batching. Across more than 1200 simulation runs spanning fleet sizes, sensor densities, and coverage areas, we observe that Spray and Focus DTN routing reduces end-to-end latency by 31% relative to baseline (non-DTN) protocols. A key finding is that enabling the DDS Batching quality of service (QoS) policy achieves 32% lower energy per delivered message than MQTT S&F, due to reduced protocol overhead. Fleet scalability analysis reveals that increasing the UAV count from 4 to 12 improves delivery ratio by 106% but increases per-message energy by up to an order of magnitude, exposing a critical reliability–energy trade-off. Our analysis further reveals that, within the evaluated mobility and density configurations, the Focus phase of the DTN mechanism is rarely invoked (<3% of forwarding events), suggesting potential for protocol simplification, which is confirmed through a direct comparison with a Spray and Wait variant that achieves equivalent PDR and latency while reducing communication energy by 13%–40% across the evaluated configurations.