Loss Differentiation and Congestion Control in Heterogeneous Networks

摘要

Aiming at the lack of effective packet loss differentiation mechanism in existing heterogeneous network congestion control strategies, which leads to efficient use of bandwidth, a packet loss differentiation mechanism based on Kalman filter is proposed and applied to TFRC congestion control strategy. In this algorithm, the average packet arrival delay in the stable phase and the packet arrival time interval in the packet loss phase are used to determine the type of the packet loss and the arrival interval is estimated by using Kalman filter to reduce the measurement error, and then the packet loss rate according to the packet loss type is recalculated. With the packet loss rate, the congestion control strategy starts to control the sender's rate. Experiments show that this algorithm can make congestion control policy more accurately identify the network status and improve the network utilization so that it can better meet the video streaming application requirements.