DOC PREVIEW
UCCS CS 622 - Comparative Study of Various TCP Versions Over a Wireless Link With Correlated Losses

This preview shows page 1-2-3-4-5 out of 14 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 14 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 14 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 14 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 14 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 14 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 14 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

370 IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 11, NO. 3, JUNE 2003Comparative Study of Various TCP Versions Over aWireless Link With Correlated LossesFarooq Anjum, Member, IEEE, and Leandros TassiulasAbstract—In this paper, we investigate the behavior of thevarious Transmission Control Protocol (TCP) algorithms overwireless links with correlated packet losses. For such a scenario,we show that the performance of NewReno is worse than theperformance of Tahoe in many situations and even OldTahoe ina few situations because of the inefficient fast recovery methodof NewReno. We also show that random loss leads to significantthroughput deterioration when either the product of the squareof the bandwidth-delay ratio and the loss probability when in thegood state exceeds one, or the product of the bandwidth-delayratio and the packet success probability when in the bad state isless than two. The performance of Sack is always seen to be thebest and the most robust, thereby arguing for the implementationof TCP Sack over the wireless channel. We also show that undercertain conditions the performance depends not only on thebandwidth-delay product but also on the nature of timeout,coarse or fine. We have also investigated the effects of reducingthe fast retransmit threshold.Index Terms—Correlated losses, packet train model, perfor-mance analysis, TCP algorithm, TCP over wireless.I. INTRODUCTIONTRANSMISSION Control Protocol (TCP) is the transportprotocol used by many internet-based applications,including http, ftp, telnet, etc. TCP is a reliable end-to-endwindow-based transport protocol designed for the wirelinenetworks characterized by negligible random packet losses.The way that TCP works is that it keeps increasing the sendingrate of packets as long as no packets are lost. When packetlosses occur, e.g., due to the network becoming congested, TCPdecreases the sending rate. Thus, TCP infers that every packetloss is due to congestion and, hence, backs off in the form ofreducing the send window. Extending TCP as used over thewireline links to the wireless links may not be an efficientsolution due to the different characteristics of the wirelineand the wireless links. This is because wireless networks arecharacterized by bursty and high channel error rates, unlikethe wireline networks. Due to this, the throughput of a TCPconnection over a wireless link suffers. In spite of this, theTCP protocol is still used to transfer data over the wirelessManuscript received June 3, 1999; revised April 3, 2002; approved byIEEE/ACM TRANSACTIONS ON NETWORKING Editor T. V. Lakshman. Thiswork was supported in part by the National Science Foundation underCAREER Award NCR-9502614 and by the Air Force Office of ScientificResearch under Grant 95-1-0061. This paper was presented in part at the ACMSIGMETRICS, Atlanta, GA, 1999.F. Anjum is with Telcordia Technologies, Morristown, NJ 07960 USA(e-mail: [email protected]).L. Tassiulas is with the Department of Electrical Engineering, Uni-versity of Maryland, College Park, MD 20742-3285 USA (e-mail: [email protected]).Digital Object Identifier 10.1109/TNET.2003.813033link, though a lot of attention is currently being given to thedesign of a better protocol over wireless links [2]–[4], [7], [15],[19]. Because of the difficulty of modeling the TCP protocolanalytically, many of these studies have been simulation based.On the other hand, it is not possible to obtain insight intothe effects of particular parameters on the behavior of TCPusing simulations of specific settings. Further, investigations toimprove TCP or design a better transport protocol can becomeless cumbersome given a simple and accurate analytical modelfor TCP.The first step toward the design of a better transport pro-tocol for wireless networks has to be a better understanding ofthe way TCP works over wireless links. This would reveal thereasons for the inefficiency of TCP over wireless links. Therehave been several efforts recently on the analytical study of TCPover wireline [12] as well as over wireless links [9]–[11], [20].Two classes of random losses have been considered, indepen-dent identically distributed (i.i.d.) and correlated. The effect ofi.i.d. packet losses on TCP performance is studied in [9]. Theyaddress the TCP versions Tahoe, Reno, and NewReno, but onlyin the context of a local network scenario. They also evaluatethe various protocol features such as fast retransmit and fast re-covery. Unlike in this paper, the packet transmission times areassumed to be exponentially distributed, as is the transmissiontime of the packet on the lossy link. Further, they also modelthe congestion avoidance phase probabilistically in which eachacknowledgment (ack) causes the window to be incremented byone with a certain probability. Note that both these assumptionshave to be resorted to in this approach so as to carry out themean cycle time analysis. In this study, the authors also considerthe less frequent case where the size of the receiver windowis the constraint on the sender’s window increase and not thebandwidth delay product of the link. Thus, this study precludesthe study of the basic TCP mechanism whereby the windowsize is increased until there is a loss due to congestion. Thisloss is caused on account of the bandwidth delay constraint. In[13], Mishraet al. consider only OldTahoe over a link with i.i.d.losses. Lakshman and Madhow [11] consider Tahoe and Reno ina regime where the bandwidth-delay product of the network ishigh compared to the buffering in the network. They show usingapproximate analysis that random independent packet loss leadsto significant throughput deterioration when the product of theloss probability and the square of the bandwidth-delay productis larger than one.In [10], Kumar and Holtzman consider the behavior of TCPTahoe and OldTahoe in the presence of correlated packet losses.The results obtained are applicable only in case of very lowbandwidth wireless links. More emphasis is placed on analyzinga link layer solution, of hiding the losses from the transport layer1063-6692/03$17.00 © 2003 IEEEANJUM AND TASSIULAS: TCP VERSIONS OVER WIRELESS LINK WITH CORRELATED LOSSES 371by link layer retransmissions, to the problem faced by TCP overthe wireless link. The approach used is also similar to the ap-proach used in [9] and, hence, suffers from the drawbacks notedearlier. Zorzi and Rao [20] study TCP OldTahoe assuming acorrelated loss model. The approach followed by them


View Full Document

UCCS CS 622 - Comparative Study of Various TCP Versions Over a Wireless Link With Correlated Losses

Documents in this Course
Fast TCP

Fast TCP

34 pages

Load more
Download Comparative Study of Various TCP Versions Over a Wireless Link With Correlated Losses
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Comparative Study of Various TCP Versions Over a Wireless Link With Correlated Losses and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Comparative Study of Various TCP Versions Over a Wireless Link With Correlated Losses 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?