Optimizing QoS in Wireless IoT Networks: A Cross-Layer based Experimental Study

Rising demand for Quality of Service (QoS) guarantees in wireless networks, driven by new use cases and technologies, necessitates innovative solutions to efficiently manage resources and meet diverse requirements. Traditional approaches addressing single-layer parameters provide limited gains, as cross-layer parameters significantly influence QoS metrics like throughput, latency, and packet loss. Employing a structure learning approach via mutual information and entropy concepts, we uncover interrelationships between layer parameters and their impact on QoS metrics. Our study demonstrates how parameters like Contention Window, Transmit Power, Queue Size, ACK Timeout, and MCS values impact network performance in WiFi-based IoT networks.
Based on our findings, we have proposed an architecture that can exploit these cross-layer interactions and relationships in a SDN controlled network to develop an AI based improved QoS management solution.