Abstract
The rapid proliferation of the Internet of Things (IoT) has intensified the demand for lightweight and scalable communication protocols, with Message Queuing Telemetry Transport (MQTT) emerging as a dominant standard for resource-constrained environments. Despite its widespread adoption, MQTT exhibits critical limitations in security, scalability, and performance under dynamic and high-load conditions. This dissertation addresses the fundamental gap between theoretical security mechanisms and their practical effectiveness in real-world IoT deployments.This research presents a comprehensive experimental and analytical investigation of MQTT systems, focusing on three core challenges: (i) authentication and authorization enforcement, (ii) absence of native flow-control mechanisms leading to backpressure and system instability, and (iii) performance degradation caused by wildcard-based subscription patterns. A controlled and reproducible testbed was implemented using an Eclipse Mosquitto broker and custom C++ clients to evaluate system behavior under varying workload intensities and security configurations.
For authentication and authorization (RQ1), the study quantifies enforcement accuracy, latency overhead, and access control reliability under both correct and misconfigured Access Control List (ACL) policies. The results demonstrate that while strong authentication mechanisms improve system integrity, misconfigurations can significantly degrade both security and performance.
To address system instability (RQ2), this dissertation proposes a novel backpressure-driven adaptive flow-control algorithm that dynamically regulates the publishing rate based on real-time congestion indicators, including latency, queue growth, and delivery success rate. Experimental results show that the proposed approach effectively prevents broker overload, stabilizes queue behavior, and significantly improves message delivery reliability compared to default MQTT operation.
For wildcard routing behavior (RQ3), the research provides a quantitative analysis of the impact of wildcard subscriptions on broker performance. The findings reveal that multi-level wildcard patterns substantially increase CPU utilization, routing complexity, and message latency, particularly under high-load conditions. A structured mitigation strategy is introduced to control wildcard-induced performance degradation.
Collectively, the contributions of this dissertation include: (1) a systematic experimental framework for evaluating MQTT security and performance, (2) a lightweight adaptive flow-control mechanism for congestion mitigation, and (3) an empirical characterization of wildcard routing overhead in IoT messaging systems. The results demonstrate that integrating adaptive control and structured security enforcement is essential for achieving scalable, reliable, and secure MQTT-based IoT communication.