Autonomous Queue Management System in Software-defined Routers for Sensor Networks

Recent advancements in sensor networks, the Internet of Things (IoT), and machine communication technologies have led to a significant increase in network traffic across various domains. However, this growth has brought forth the critical challenge of bufferbloat, particularly in next-generation sensor networks. Bufferbloat, characterized by excessive buffering in network equipment resulting in high latency and jitter, severely impacts service quality, leading to outages and interruptions. Despite existing active queue management schemes, such as RED, PIE, and CoDel, bufferbloat remains a persistent issue since existing solutions consider limited parameters. To address this issue, we propose a novel autonomous queue management system called dynamic port switching and queue management (DPSQM), designed specifically for software-defined routers in sensor networks. Unlike traditional solutions that react to congestion after it occurs, DPSQM proactively intervenes based on the likelihood of bufferbloat, integrating multiple parameters, including average queue length, current delay, and port utilization. Through extensive simulations using the NS-3 network simulator, we demonstrate that DPSQM outperforms existing methods in terms of packet loss rate and average queue delay, thereby significantly enhancing network efficiency and performance. Moreover, we showed that our proposed port-switching algorithm successfully switches ports when network congestion and bufferbloat occur under different traffic densities.