COMMITTEE CHAIR: Dr. Cajetan Akujuobi

TITLE: EDGE-NATIVE MULTI-AGENT SYSTEM FOR LATENCY-CRITICAL ANOMALY DETECTION AND QOS OPTIMIZATION IN 6G SMART MANUFACTURING

ABSTRACT: The increasing reliance on wireless connectivity in smart manufacturing places stringent demands on network latency, reliability, and adaptability that exceed the capabilities of static or threshold-based quality-of-service (QoS) control mechanisms. While 5G standalone networks enable industrial deployments, emerging 6G environments are expected to introduce greater variability, tighter latency constraints, and more frequent performance degradation events, necessitating autonomous and learning-driven control strategies. This dissertation proposes an intelligent multi-agent framework for real-time anomaly detection and QoS optimization in industrial wireless networks. The framework integrates edge-based monitoring, unsupervised anomaly detection, and reinforcement learning–driven control to enable proactive and adaptive network management. Network telemetry collected from a Firecell 5G standalone testbed is used as an empirical baseline, while a scaled 6G-emulated dataset is employed to evaluate robustness under more stringent performance conditions. Anomaly detection is performed using Isolation Forest and deep autoencoder models to identify deviations in latency, throughput, and packet loss. QoS control is formulated as a Markov Decision Process and addressed using reinforcement learning agents that dynamically adjust network parameters to mitigate detected anomalies. The proposed multi-agent system is evaluated against static and threshold-based baselines across multiple operating regimes. Experimental results demonstrate that the learning-based approach achieves improved latency stability, faster throughput recovery, and reduced packet loss under both moderate and stress-intensive conditions. The framework maintains consistent control behavior across 5G and 6G-oriented scenarios, indicating scalability and robustness under increased network volatility. These results validate the effectiveness of multi-agent, learning-driven QoS control for next-generation industrial networks and provide a practical pathway toward autonomous network management in 6G-enabled smart manufacturing environments.

Keywords: Intelligent multi-agent framework, edge-based monitoring, 5G standalone testbed, QoS, 6G, autonomous network management

Room Location: Electrical and Computer Engineering Department Conference, Room 315D