A backpressure-driven flow control model for stabilizing MQTT communication in IoT systems

This paper investigates the stability challenges of the Message Queuing Telemetry Transport (MQTT) protocol in Internet of Things (IoT) environments under varying traffic loads and network conditions. While MQTT is widely adopted for its lightweight design and efficiency, it lacks effective application-level flow-control mechanisms, which can lead to broker congestion, increased latency, and message loss when publishing rates exceed the effective processing capacity of the experimental communication pipeline. This study demonstrates that a lightweight backpressure-driven adaptive flow-control mechanism can substantially improve MQTT communication stability under controlled experimental conditions. The approach dynamically adjusts the message transmission rate based on real-time system indicators, including end-to-end latency, delivery success rate, and congestion signals. Unlike existing approaches that rely on static configurations or simulation-based evaluations, this study employs a controlled experimental testbed using an Eclipse Mosquitto broker and C + + MQTT clients to compare default MQTT behavior with adaptive publishing under normal, high-load, and adverse network conditions. Experimental results from the baseline MQTT configuration without adaptive control demonstrate that uncontrolled publishing leads to queue buildup, significant latency growth, and reduced delivery reliability as system load increases. In contrast, the proposed adaptive mechanism effectively regulates message rates, reduces congestion effects, and maintains more stable throughput and higher delivery success rates. Statistical validation confirms that improvements are significant ( p < 0.01). These findings indicate that application-layer flow control can significantly enhance MQTT communication stability without requiring modifications to the protocol or broker implementation, making it suitable for practical IoT deployments.