TCP Congestion Control Algorithms: A Detailed Analysis and Comparison

This report provides a comparative analysis of several TCP congestion control algorithms, focusing on their mechanisms for avoiding and managing network congestion. The report begins with an introduction to TCP and its role in reliable data transmission, highlighting the challenges posed by packet loss and network congestion. It then delves into the specifics of each algorithm, including TCP Tahoe, which uses a conservation of packets principle, and TCP Reno, which adds intelligence to detect lost packets early. The report also covers TCP New Reno, an improvement over Reno for multiple packet losses, TCP SACK, which addresses the limitations of New Reno, and TCP Vegas, which takes a proactive approach to congestion avoidance. Each algorithm's strengths, weaknesses, and operational details are discussed, and the report concludes with a summary of the performance and suitability of each algorithm, emphasizing the advantages of TCP Vegas in terms of reduced retransmissions and effective congestion avoidance. The report references key literature to support the analysis.

Table of Contents
Introduction...........................................................................................................................................1
TCP TAHOE.............................................................................................................................................1
TCP RENO...............................................................................................................................................1
NEW RENO.............................................................................................................................................2
TCP SACK................................................................................................................................................2
TCP VEGAS..............................................................................................................................................2
CONCLUSION..........................................................................................................................................2
References..............................................................................................................................................3

Introduction
TCP is one of the most reliable protocol for connections that are oriented or based on end to end
protocols. It has in itself all the means and mechanisms for ensuring how reliable it is for acquiring
the receiver by acknowledging the segments that are received. In most cases not all networks are
perfect as packets will always get lost on the way during routing. This may be either due to error in
the network or to the facts that there is congestion or traffic over the network hence making the
routers to drop the packets. The main reason for this report is to analyse and compare the different
congestion controls. The major area covered being the avoidance means and mechanism as they are
proposed by the TCP/IP protocols namely the Renoe, Vegas, Tahoe, SACK and new Reno etc.
(Jacobson, 2010).
TCP TAHOE
TCP Tahoe is referred to as the TCP traffic control type of an algorithm that is based on the principle
of the conservation of packets. This principle is implemented in essence that it is acknowledged by
clocking the outgoing packets before they are taken off by the receiver. This congestion control is
used in maintaining the traffic window CWD in reflecting the capacity of the network. The issues
revolving around TCP Tahoe that need to be addressed are as follows
Determining the bandwidth that is available
Ensuring that the equilibrium is always maintained
Reaction in congestion
To avoid such congestion Tahoe uses the Additive increase multiplicative decrease. The problems
with Tahoe is that it always takes a complete timeout interval in detecting a lost packet. In some
cases it might take longer due to the timeout of the coarse grain (lavobison, 2011).
TCP RENO
Reno is referred to as the retaining basic principles of the Tahoe such as the coarse grain but in real
sense it adds a lot of intelligence so that the packets that are lost may be detected at an early stage.
This make the pipe not to be emptied each and every time there is loss of a packet. If there is a high
possibility that a packet is lost Reno suggests an algorithm referred to as ‘Fast Re-Transmit’
(Henderson, 2009).
The problem with Reno is that it usually performs well when there are small losses of packets. This is
opposite with Reno especially in cases when we have multiple losses of packets, the reason being
that it may not perform well.

NEW RENO
This is simply a slight modification that makes it to be over the TCP Reno. It is improved in essence
that it is able to detect the losses of multiple packets, hence making it more efficient as compared to
RENO in situations where there is multiple loss of packets.
The problem associated with Reno is that it suffers from facts that it only takes one RTT in detecting
each packet that is lost.
TCP SACK
This is an extension of TCP Reno and works in essence that it tries to cancel and solve the problems
faced by TCP NEW RENO. This congestion control is also used in detecting multiple packets that are
lost. TCP SACK is also applied when we need to re-transmit more than one packet that is lost as per
the RTT.
The problem associated with SACK is that the current selective acknowledgements may not be
provided by receivers in implementing SACK , hence there arise a need to implement the selected
acknowledgement which might not be that easy.
TCP VEGAS
TCP Vegas is a modification of RENO. This is because it is built on facts that the proactive measures of
encountering the congestion may be much more efficient than the reactive ones. Vegas has the new
extension of re-transmitting means and mechanism (Brakmo, 2011)..
CONCLUSION.
TCP Vegas is clear and better than the others in essence that.
1. It has few re-transmissions.
2. Very good at congestion avoidance.
References
Brakmo. (2011). IEEE Journal on Selected Areas in communication. TCP Vegas: End to End Congestion
Avoidance on a global internet, 13, 1465-1490.
Henderson. (2009, april). Reno Modification to TCP’s Fast.
Jacobson. (2010). Congestion Control and Avoidance.
lavobison. (2011). Congestion Avoidance and Control”.SIGCOMM Symposium no Cummunication
architecture and protocols.