International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 7
Simulation Analysis of a New Startup Algorithm for TCP New Reno
Ahmed Yusuf Tambuwal
1
, Aminu Bui Muhammed
2
1,2
Computer Science Unit, Usmanu Danfodiyo University, Sokoto (UDUS)
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Standard TCP (New Reno) is vulnerable to
startup effects that cause loss of connection setup packets or
result in long round trip time (RTT) greater than 1-second.
When either of these events occurs, TCP New Reno resets its
congestion state by reducing initial congestion window (IW)
and slow-start threshold (ssthresh) values to 1 and 2
maximum segment size (MSS) respectively. In this condition,
TCP requires multiple round trips to complete delay-
sensitive transactions, thus resulting in poor user-
experience. This paper presents a new congestion control
algorithm that makes TCP more responsive by increasing its
robustness against startup losses. Our main contribution in
this paper is performing extensive simulation studies to
investigate dynamics of the proposed algorithm. The main
result obtained shows that an average latency gain of 15
RTTs can be achieved at up to 90% link utilisation, with a
packet loss rate (PLR) of 1%.
Key Words: Transmission Control Protocol, Congestion
Control, Startup, Responsiveness, Short-Lived Applications
1. Introduction
The transmission control protocol (TCP) 1 is the main
protocol used on the Internet for reliable delivery of data
packets between communicating hosts. A TCP client
initiates a connection to a remote Internet server using a
three-way handshake (3WHS) procedure. In general, the
client then submits a data request, which is processed by
the server resulting in a data response. Once data
transmission starts, TCP attempts to maximize throughput
without causing congestion on the network. Several works
have focused on designing new TCP algorithms with better
throughput performance such as in 2345.
Conversely, throughput performance is not the main
requirement of short-lived interactive applications (e.g.
web browsing and E-commerce), which account for a
majority of TCP flows 67. Quite different from bulk
transfers, interactive applications demand speedy delivery
of few data chunks across the Internet within short delay
bounds. Despite many algorithms proposed to solve this
important problem 89101112, it still remains an open
challenge for TCP.
This paper proposes a new algorithm that aims to make
standard TCP (New Reno) more responsive by increasing
its robustness against startup losses. TCP New Reno
interprets the loss of connection setup packets (i.e. SYN or
SYN-ACK) as a signal for serious network congestion,
prompting a sender to reduce its initial congestion
window (IW) to 1 maximum segment size (MSS) and its
slow-start threshold (ssthresh) to 2 MSS 13. This response
increases latency of short-lived interactive applications by
several round trips, thus significantly reducing end-user
Internet experience.
While ignoring the SYN congestion signal and starting with
very large IW and ssthresh values negates TCP
conservative principles, there are strong motives to use a
less conservative approach. Firstly, random packet loss is
quite a common occurrence when data traverses
wireless/mobile network links e.g. due to high contention
between multiple users sharing the radio channel, poor
weather conditions, or when a mobile host is obstructed
and suffers temporary link outages 1415. Also, network
middle boxes such as firewalls, proxies, and network
address translators, can erroneously drop SYN packets
due to suspicion of unwanted or malicious traffic 1617. In
a more general context, TCP inherently causes loss of
packets (including the SYN and SYN-ACK) when probing
for available capacity and trying to maximize throughput
181920.
This paper investigates performance of a newly proposed
algorithm called ǮTCP SYN Loss ȋTSLȌ Startup Algorithmǯ
that uses a halving congestion response function during
startup, which is less conservative than the current
standard. After connection setup is completed, standard
TCP congestion control is applied for the data transfer
phase. Our proposed algorithm uses the following set of
instructions.
ሺ ሻ ሼ
ሺ
ሻ
(
)
ሽ
The rest of this paper is organized as follows: Section 2
presents the simulation setup that we use for the study.
Section 3 then presents and discusses extensive
simulation results before the paper concludes with a
summary in section 4.
2. Simulation Setup
In order to investigate dynamics of the proposed algorithm,
sets of simulations are performed with realistic models
that represent Internet connections at different congestion
levels. The network topology, application traffic, and
transport model are discussed below.
2.1 Network Topology
An institution network normally has enough
bandwidth to carry its own traffic. Similarly, the Internet