Improving TCP performance over heterogeneous networks : The investigation and design of End to End techniques for improving TCP performance for transmission errors over heterogeneous data networks.

摘要

Transmission Control Protocol (TCP) is considered one of the most important protocolsudin the Internet. An important mechanism in TCP is the congestion controludmechanism which controls TCP sending rate and makes TCP react to congestionudsignals. Nowadays in heterogeneous networks, TCP may work in networks with someudlinks that have lossy nature (wireless networks for example). TCP treats all packetudloss as if they were due to congestion. Consequently, when used in networks thatudhave lossy links, TCP reduces sending rate aggressively when there are transmissionud(non-congestion) errors in an uncongested network.udOne solution to the problem is to discriminate between errors; to deal with congestionuderrors by reducing TCP sending rate and use other actions for transmissionuderrors. In this work we investigate the problem and propose a solution using anudend-to-end error discriminator. The error discriminator will improve the currentudcongestion window mechanism in TCP and decide when to cut and how much toudcut the congestion window.udWe have identified three areas where TCP interacts with drops: congestion windowudupdate mechanism, retransmission mechanism and timeout mechanism. All ofudthese mechanisms are part of the TCP congestion control mechanism. We proposeudchanges to each of these mechanisms in order to allow TCP to cope with transmissionuderrors. We propose a new TCP congestion window action (CWA) for transmissionuderrors by delaying the window cut decision until TCP receives all duplicate acknowledgmentsudfor a given window of data (packets in flight). This will give TCP a clearudimage about the number of drops from this window. The congestion window size isudthen reduced only by number of dropped packets. Also, we propose a safety mechanismudto prevent this algorithm from causing congestion to the network by usingudan extra congestion window threshold (tthresh) in order to save the safe area whereudthere are no drops of any kind. The second algorithm is a new retransmission actionudto deal with multiple drops from the same window. This multiple drops actionud(MDA) will prevent TCP from falling into consecutive timeout events by resendingudall dropped packets from the same window. A third algorithm is used to calculateuda new back-off policy for TCP retransmission timeout based on the network¿s availableudbandwidth. This new retransmission timeout action (RTA) helps relating theudlength of the timeout event with current network conditions, especially with heavyudtransmission error rates.udThe three algorithms have been combined and incorporated into a delay baseduderror discriminator. The improvement of the new algorithm is measured along withudthe impact on the network in terms of congestion drop rate, end-to-end delay, averageudqueue size and fairness of sharing the bottleneck bandwidth. The results show thatudthe proposed error discriminator along with the new actions toward transmissionuderrors has increased the performance of TCP. At the same time it has reduced theudload on the network compared to existing error discriminators. Also, the proposeduderror discriminator has managed to deliver excellent fairness values for sharing theudbottleneck bandwidth.udFinally improvements to the basic error discriminator have been proposed byudusing the multiple drops action (MDA) for both transmission and congestion errors.udThe results showed improvements in the performance as well as decreases in theudcongestion loss rates when compared to a similar error discriminator.