Improving TCP smoothness by synchronized and measurement-based congestion avoidance

Abstract

In this paper, we observe that although multiplicative decrease is necessary to accomplish fairness in congestion control, it does not inevitably sacrifice system throughput, as long as the system operates between the knee and the cliff, according to an equation. However, even when the system throughput is relatively stable, end users of real-time applications do not necessarily experience a smooth traffic, mainly caused by the unsynchronized window adjustments due to random congestion indications. We analyzed and evaluated the negative impact of random window adjustments on smoothness, short-term fairness, and even long-term fairness measured by a novel fairness metric defined in this paper. Therefore, we propose an experimental congestion avoidance mechanism to improve TCP smoothness for media-streaming applications. The mechanism relies on a fine-grained RTT estimation to measure the network condition, and coordinates the upward and backward window adjustments to abolish the damage of unsynchronized window control. Congestive packet drops are reduced by a new control parameter g, and the bottleneck queue length can also be controlled in an end-to-end way. Simulation results confirm that the new mechanism enhance significantly the smoothness and fairness, without a cost of responsiveness. In fact, by enabling a new parameter d, the responsiveness can be even enhanced when the bandwidth is under-utilized.