Congestion control schemes for single and parallel TCP flows in high bandwidth-delay product networks

摘要

In this work, we focus on congestion control mechanisms in Transmission ControlProtocol (TCP) for emerging very-high bandwidth-delay product networks and suggestseveral congestion control schemes for parallel and single-flow TCP. Recently, severalhigh-speed TCP proposals have been suggested to overcome the limited throughputachievable by single-flow TCP by modifying its congestion control mechanisms.In the meantime, users overcome the throughput limitations in high bandwidth-delayproduct networks by using multiple parallel TCP flows, without modifying TCP itself.However, the evident lack of fairness between the high-speed TCP proposals (orparallel TCP) and existing standard TCP has increasingly become an issue.In many scenarios where flows require high throughput, such as grid computingor content distribution networks, often multiple connections go to the same or nearbydestinations and tend to share long portions of paths (and bottlenecks). In such casesbenefits can be gained by sharing congestion information. To take advantage of thisadditional information, we first propose a collaborative congestion control scheme forparallel TCP flows. Although the use of parallel TCP flows is an easy and effectiveway for reliable high-speed data transfer, parallel TCP flows are inherently unfairwith respect to single TCP flows. In this thesis we propose, implement, and evaluatea natural extension for aggregated aggressiveness control in parallel TCP flows.To improve the effectiveness of single TCP flows over high bandwidth-delay product networks without causing fairness problems, we suggest a new TCP congestioncontrol scheme that effectively and fairly utilizes high bandwidth-delay product networksby adaptively controlling the flow??s aggressiveness according to network situationsusing a competition detection mechanism. We argue that competition detectionis more appropriate than congestion detection or bandwidth estimation. We furtherextend the adaptive aggressiveness control mechanism and the competition detectionmechanism from single flows to parallel flows. In this way we achieve adaptive aggregatedaggressiveness control. Our evaluations show that the resulting implementationis effective and fair.As a result, we show that single or parallel TCP flows in end-hosts can achievehigh performance over emerging high bandwidth-delay product networks without requiringspecial support from networks or modifications to receivers.