Efficiency Analysis of TCP Alternatives

Performance Analysis of TCP Variants

Sanjay Rajashekhar ([email protected] neu. edu)

Tiantong Deng ([email protected] neu. edu)

Ganesh Aithal Parampalli Narayana ([email protected] neu. edu)

Abstract— TCP is the main and most widely used transfer

protocol to get reliable communication. Because of its common need, experts have been studying and suggesting new TCP

variants trying to improve it is behavior to queuing and

congestion to create it utilize the most available bandwidth while preserving may well level of justness towards additional protocols. This paper aims at evaluating and comparing the performance in the TCP versions (Tahoe, Sparks, NewReno and Vegas) based upon

congestion, queuing and fairness. We properly choose cases to investigate the throughput, intra- and inter-protocol fairness of these TCP versions. We dispute through evaluation that TCP Vegas, having its better band width estimation structure, uses the network methods more efficiently and fairly than any other TCP versions. Simulation the desired info is given that support the benefits of the analysis.

Key Words — TCP Tahoe, Reno, NewReno, Vegas, Regular Bit

charge (CBR), Blockage, NS-2, TTL, SACK, RTT, SACK, REDDISH,

UDP and DropTail.

We. INTRODUCTION

In networking, data and info are most significant in true to life scenario. The clear comparative ideas in different

perspectives of congestion windowing mechanism plays a big

position. To have a accurate comparison among different

components of over-crowding windowing, simulation of data

transfer and drop with different protocols & alternatives and the appropriate mathematical evaluation of data copy are need to.

Transmission Control Protocol can be described as widely used connection

oriented travel layer protocol that provides a trusted packet delivery over an unreliable network. Several different variations have been developed in order to improve congestion control.

In this daily news we assess the efficiency of various TCP

variants (TCP Tahoe, TCP Reno, TCP NewReno, and TCP

Vegas). When we discuss performance we are dealing with

the congestion control, flow control, and fairness between the alternatives. Every TCP variant contains a special characteristic. TCP Tahoe has a characteristic of gradual start, blockage avoidance. TCP Reno

own fast retransmission and quickly recovery. TCP NewReno

tools fast restoration and fast retransmission pertaining to partial acknowledgement. And TCP Vegas gives efficient

efficiency and retransmission is based on type of

acknowledgement and Round Trip Time (RTT).

The pair of experiments are conducted in NS-2 simulator with

animated network. The three experiments gives the comparison between your variants depending on:

1 . Congestion: comparing the throughput, bundle drop rate,

latency inside the network if the two client transfer bouts

at a Constant Bit Charge (CBR).

installment payments on your Fairness: establishing the band width in the sites

when two different TCP variant and used. Preferably use of

diverse TCP variant must be fair with each other.

a few. Queuing/Flow Control: Compare the queuing protocol

RED and Drop Butt with TCP Reno advertising SACK. Analyze

the end-end latency and fairness in bandwidth.

2. METHODOLOGY

A NS-2 (Network Simulator) is definitely the tool which in turn we have utilized

in this project. NS discrete-event network sim, primarily used in research and teaching. Natursekt provides substantial support pertaining to simulation of TCP, redirecting, and multicast protocols more than wired and wireless (local and satellite) networks. The tool is employed to measure the throughput, latency, packet damage rate and compare fairness among the TCP Variants – Tahoe, Sparks,

NewReno and Vegas. NS-2 is a powerful tool which simulates the packet circulation in the network and assess it as works on subject oriented TCL (Tool Command word Language). In this

experiment NS-2 runs upon Linux Platform. The reason why NS2 simulator is employed for this test is because they are designed to picture:

п‚· Supply flow, queue build up and packet drops.

п‚· Process behavior: TCP slow commence, self-clocking,

congestion...

References: [1] Sally Floyd and Truck Jacobson, " Random early on detection gateways

for blockage avoidance, " IEEE/ACM Ventures on

Networking, vol. 1, pp. 397-413, August 1993.

[2] W. Richard Dahon, TCP/IP Illustrated, Volume 1: The Protocols.

Reading, Ma: Addison-Wesley, year 1994.

[3] Kevin Fall and Sally Floyd, " Simulation-based

Comparisons of Tahoe, Reno, and BAG TCP”

[4] http://www.iaeng.org/publication/IMECS2009/IMECS200

9_pp351-353. pdf