An Adaptive AQM Based on the Consecutive Change Detection in the Programmable Queue

With the rapid expansion of the Internet, the surge in data traffic, propelled by the exponential growth of network services and users, has heightened the risk of network congestion, security breaches, and system instability. Addressing these challenges presents stringent demands and novel complexities in queue management. However, prevailing solutions often rely heavily on average queue size thresholds while ignoring traffic variations. In this paper, CCD-AQM, an Adaptive Queue Management approach based on the Consecutive Change trend Detection in the queue size is proposed. Considering that today's programmable data plane offers promising ways for finer analysis of the queue in the hardware switches, we implement CCD-AQM on an RMT switch and analyze its resource usage. Large-scale simulations are conducted to evaluate CCD-AQM, showcasing its superior performance in queue management. The results demonstrate its ability to maintain low queue occupancy and high throughput while ensuring fairness among competing flows.