empirical examination of tcp in manet - estirj.com · tcp agents specifically tcp-new reno,...
Post on 30-Jun-2018
219 Views
Preview:
TRANSCRIPT
ENGINEERING SCIENCE AND TECHNOLOGY INTERNATIONAL RESEARCH JOURNAL, VOL.1, NO.2, JUN, 2017
22
Corresponding author Email address: khalid.mohammadani@gmail.com
Khalid Hussain Mohammadani1, Safi Faizullah
2, Arshad Shaikh
3, Nazish Nawaz Hussaini
4, Rahat Ali Khan
5
1,3,4Department of Computer Science Isra University, Hyderabad, Pakistan.
2Department of Computer Science, Islamic University of Madina, Saudi Arabia.
5Institute of Information and Communication Technology, University of Sindh, Pakistan
Keywords: Variants of transmission control protocol, mobile ad-hoc network, routing protocol
1. Introduction
asically, the traditional mobile wireless network is
handled by some control devices to create a route from
source to destination for communication. There is no any
control device in Mobile Ad-Hoc network (MANET).
Every node of MANET is itself control device. Some
protocols help MANET to create a route from source to
destination for communication. MANET can be
exceptionally helpful in such circumstances as these don't
depend on pre-built up the infrastructural network. It is a
sort of multi-hop network [1, 2]. No need for cables to
setup the MANET network, it takes lower front cost than
the conventional wireless network. It spends low energy
and can spread at long distance for communication.
MANET is additionally collections of the movable device
because every device can move freely toward any path and
henceforth the nodes change their connections habitually.
This is a reason of MANET that it stays away from the
wide arrangement of fixed links and the transmission can
be completed in a practical manner. In Figure 1. the
collection of free nodes make a MANET. Where “S” is
source node; “F” identifies the forwarding nodes and “D”
shows the destination node. There are other nodes present
as neighboring nodes.
MANET can communicate with the internet where end
station nodes are fixed with help of MANET gateways.
Fixed or mobile node can be presented the roll or gateway
between MANET and internet. From the previous short
discussion, we can observe that MANET is the name of
such network which uses fewer resources and utilization of
MANET is not impossible for building the network. In this
paper, we evaluate the performance of TCP variants using
AODV routing protocol in MANET scenario.
Figure 1. Multi-hope mobile network
We further discuss types of TCP variants and one of the
MANET routing protocol in section 2. The related work
with TCP and TCP variants is discussed in section 3.
Section 4 defines methodology to conduct this research.
Section 5 demonstrates to results. The summary of results
is discussed in section 6. Section 7 is the last section that
gives ending remarks and followed by references.
2. Transport control protocols
Transmission control protocol (TCP) oversees ensuring that
a message is broken into the chunks that IP directs and for
reassembling the chunks again into the aggregate message
toward the end side. TCP was proposed for strong
correspondence in the network. In early time, it was
conceptualized of a wired network and henceforth the
managing standards of the strategies were with regards to
the qualities of a wired network [3]. Since the wireless
network has picked up in prevalence. Wireless networks
are distinctively more mix-up slanted than wired network
B
Abstract: Transmission control protocol (TCP) was initially intended for established systems to give the dependability of the
information conveyance. The change of TCP execution was additionally accomplished with various sorts of systems with the
presentation of new TCP variations. To decide the best TCP variant from the system multifaceted nature perspective, we simulate
some TCP variants in a real-life scenario. This work addresses the execution of TCP and TCP variants, for example, TCP, TCP-
Reno, TCP-New Reno, TCP-Vegas as sources and TCP-Sink as a goal in the network. The situations displayed in this paper are
bolstered by ad-hoc on-demand distance vector (AODV) routing protocol with the incorporation of various network
complexities of the versatile mobile ad-hoc network (MANET) utilizing random waypoint portability model. The situation
demonstrates portable nodes situated especially for a grounds zone. On the premise of simulation, we analyze End-to-End delay,
jitter, throughput, and packet deliver ratio as the quality of service (QoS) measurements.
.
Empirical Examination of TCP in MANET
ISSN (e) 2520-7393
ISSN (P) 2521-5027
Received on 1stJune, 2017
Revised on 28th June, 2017
www.estirj.org
K. H.MOHAMMADANI et.al. EMPIRICAL MODELLING OF TCP IN MANET
Copyright ©2017, ESTIRJ-VOL.1, NO.2 (22-27)
on account of a couple channel qualities. The effects of
obscuring, multipath etc. provoke higher errors and bundle
incidents in a wireless network. TCP was expected to
prompt package disasters as a sign of network obstruct and
take remedial measures properly. In a wireless network,
this enlistment is wrongly made despite when the death of a
packet in transmission. It is a result of channel setbacks and
not the bottleneck. This prompts over the top number of
retransmissions and timeout events provoking exponential
spoil of network execution in a short time span [4].
Research has been led to prompt adjustments to the
TCP configuration to consider the necessities of a wireless
domain. There is a plenty of ways that have been proposed
to alleviate the impacts of wireless channel on TCP [5].
When TCP was considered, certain directing calculations
were intended to control the activity stream and advance
the system execution. Nonetheless, with advancing wireless
systems, those directing calculations have neglected to give
ideal system execution since they were intended to manage
blockage and approaches to counteract clog. Thus, these
directing calculations were proactive in nature. It implies
they would make a move after the issue had happened.
There are various variants of TCP available. Some of them
are used in this paper.
2.1. TCP Reno
It's practically like Tahoe. It incorporates quick restoration
component. It identifies early packet loss. It sits tight for 3
copy of ACK (Acknowledgment) to retransmit. Likewise, it
doesn't decrease the congestion window too soon. Higher
data transfer capacity use was seen in the correlation of
different variations in Reno [6, 7].
2.2 TCP-New Reno
TCP Reno was adjusted to concoct new variant called TCP
New Reno. This is the calculation with minor alterations in
TCP Reno [8]. It has the benefit of asset usage to ideal.
Here, the sender gets halfway ACK, however, doesn't select
Fast Recovery, rather, expect that each next packet of last
ACK portion is lost thus take after Retransmission for that
packet. It doesn't wait for the clock to time out for touching
off for retransmission.
2.3. TCP-Vegas
TCP Vegas depends on Reno. It proficiently checks for
timeouts of clocks and conquers the issue of copy
affirmation and identifies early packets loss and TCP Vegas
is an enhanced adaptation of moderate begin. Along these
lines, it identifies bottleneck before packet loss happens. It
forces blockage shirking as opposed to blocking
identification. It has likewise changed retransmission
methods. TCP Vegas records the littlest measured RTT and
records the accessible data transfer capacity [9].
2.4 Ad Hoc On-Demand Distance Vector (AODV)
MANET is blind without routing protocols. The routing
protocols are that assign the path for packet forwarding in
MANET [19]. Ad-hoc on-demand distance vector (AODV)
routing protocol is used to conduct the evaluation of TCP
variants under MANET in this paper. AODV routing
protocol belongs to reactive and topology based families of
routing protocols. The reactive families of routing protocol
get up to make a routing path on the need for
communication between network nodes.
2.4.1. Discovery process of AODV
AODV limits the quantity of required communicates by
making paths in an on-request way [10]. At the point when
a source device goals to send information to another end
device, it needs to start a way of discovering procedure to
find the other device. A source device broadcasts a route
request (RREQ) packet to its all nearest neighbors.
Neighbor nodes of source nodes forward RREQ to their
neighbors, and so on until the destination is located. Target
node sends a route reply (RREP) packet but RREP is not
broadcasted, it passes the pre-defined route from which
target node gets the RREQ packet. The source device
begins to send actual information after getting RREP
packet utilizing a similar path [11] as a display in following
Figure 2.
Figure .2 Route Discovery process of AODV [11]
3. Literature Review
It is important to lead the related study for comprehending
fundamental and ability idea with respect to TCP variations
and MANET routing protocols. In [12] author watched
different MANET routing protocols under three TCP
variations. They quantified consequences of different QoS
parameters. Each of the routing protocol and TCP
variations is found to have performed well as a rule. One of
the variations known as the Selective Acknowledgment
(SACK), it could alter for the most part well to the
changing network sizes while the Reno performs most
capable in assorted suitability circumstances.
The work in [13], authors explored the impacts of
subscriber’s mean speed on the execution attributes of three
TCP agents specifically TCP-New Reno, Westwood, and
Cubic, in WiMAX arrange condition, under the states of
related lossy connections, route failures, and network
blockage. They noticed that TCP throughput diminishes
altogether when node movement cases connection
disappointments, because of TCP's helplessness to perceive
the contrast between connection disappointment.
In this paper [14], authors have examined the
execution of new protocol named RCP+ compare with
variations of TCP in a hybrid network situation. Their
outcomes demonstrated that RCP+ could altogether build
the productivity of congest-control-systems. The work of
[15], to explore the execution of various transmission
control protocols in MANET and wired systems. The least
impact of portability has been evaluated on TCP Westwood
that delivered lesser segments than TCP Vegas and SACK.
K. H.MOHAMMADANI et.al. EMPIRICAL MODELLING OF TCP IN MANET
Copyright ©2017, ESTIRJ-VOL.1, NO.2 (22-27)
In [16] authors concentrated through broad simulation
situations the execution attributes of various TCP agent
under the states of related wireless blunders, asymmetric
end-to-end abilities, and link congestion. Generally, their
results demonstrated that Westwood and BIC offer the best
performance over the rest TCP variations in all cases
contemplated. In the paper [17] authors analyzed different
TCP variants in the unidirectional and bidirectional
networks on the Wi-Fi scenario. Authors analyzed various
TCP variants [18]. Their motto was the attention around
TCP sender side components under appropriate buffer
management algorithm to deal with higher offered stack,
irregular misfortunes, and retransmission timeouts. Table 1.
is highlighting the significance of our work in relation to
previous research work.
Table.1. Summary of related work
Publicatio
n
TCP
Variants
QoS
Parameters
Network
Type
[12] SACK,
Reno, New
Reno
Delay,
Throughput
MANET
[13] New Reno,
Westwood &
Cubic
Throughput,
Delay, &
PDR
WiMAX
[14] Bay TCP,
FULL TCP,
TCP- Vegas,
TCP- Linux,
TCP Fack,
TCP Sack,
Fast TCP,
TCP Reno,
New Reno &
RCP+
Throughput,
Dropped
Packet,
Delay, PDR
& NRL.
Heterogeneou
s
[15] SACK,
Vegas,
Tahoe, Reno,
New Reno &
Westwood
Good put,
Fairness
Hybrid
Network
[16] New Reno,
Vegas, Reno,
Westwood &
BIC
Good put WiMAX
[17] TCP Tahoe,
Vegas, Reno,
New Reno &
Sack
Throughput
& congestion
window
Wi-fi network
[18] New Reno,
Westwood,
& CUBIC
Throughput,
PDR, Delay
& NRL
WiMAX
4. Methodology
4.1 Simulations
Network simulation is a technique that builds model
networks virtually. Simulation is utilized to test a PC
program to monitor network performance. A few
simulators are utilized for network simulation, for example,
OPNET, NS3, OMNET, and NS2 etc. In this paper, we use
NS2 for simulation purpose. NS2 is an object-oriented
simulator, written in C++, with an OTCL interpreter as a
frontend. NS2 supports various kind of (wired and
wireless) routing protocols. It supports the main two
transport agents TCP and UDP. It also supports various
type of application e.g. FTP, HTTP, WWW, CBR,
Exponential, and VBR. simplex and duplex are also
supported by NS2 as the mode of communication. We use
various types of simulation parameters in this paper. Table
2 shows the simulation parameters.
Table.2. Simulation Parameters
Simulation Parameters Parameters Values
Simulation Time 1000 seconds
Environment Size 1000.m × 1000.m
Application / Traffic FTP
Transport protocol TCP, TCP Reno, TCP New
Reno and TCP Vegas
Mobility model RWP
Maximum Speed 15 m/s
Protocols AODV
Nodes 10, 20, 30, 40 and 50
Queue Length 50
MAC 802.11b
5. Results and Discussion
This segment exhibits the results of our simulation. The
dimensions of variation included five different complexities
of a MANET network (10, 20, 30, 40 & 50) nodes, a
variation of four TCP variants (TCP, TCP New Reno, TCP
Reno and TCP Vegas) and to assess the execution of these
TCP variants through four QoS parameters have been
measured viz. Packet Delivery Ratio, Average Throughput,
e2e Delay and Jitter.
5.1 Network Throughput
Throughput is the measure of information conveyed in a
unit of time. From the graph in Fig.3. TCP behaves
similarly with TCP Reno and Vegas at initially due to low
congestion on the network. As network nodes are increased
the performance of these TCP types decreases. Initial the
performance of TCP Reno is less but as nodes increase the
TCP Reno maintains its performance and ability to deliver
gradually data than other TCP types in high complexities
scenarios.
Figure.3. Throughput (kbps) Vs No of Nodes
5.2 Jitter
Jitter is characterized as a variety in the delay of got
packets. The sending side transmits packets in a consistent
stream and spaces them equally separated. The jitter
between packets can change as opposed to staying steady
K. H.MOHAMMADANI et.al. EMPIRICAL MODELLING OF TCP IN MANET
Copyright ©2017, ESTIRJ-VOL.1, NO.2 (22-27)
due to bottleneck, inappropriate queuing and setup errors.
Fig. 4, shows the jitter of the network, initially, the jitter of
all TCP variants is similar and give a neck to neck
performance. TCP Reno decreases jitter even when
network complexity increases that means TCP Reno
maintains the network jitter at high complexity scenarios.
TCP does not control the jitter problem in all scenarios.
TCP New Reno and TCP Vegas do not control so much as
TCP Reno controls the jitter.
Figure.4. Jitter (s) Vs No of Nodes
5.3 End to End Delay
End-to-End (e2e) delay is the normal time a packet takes to
achieve the goal. It is also known latency of the network.
From the graph in Figure 5, it can be detected that the e2e
delay of TCP Vegas the remained less as compare to other
TCP variants. All variants initially increase the delay of the
network due to break the routes of the network but as
network nodes increase the more routes are available to
deliver the packets there other TCP variants try to
decreases the overall delay in the network. Initially TCP
Vegas maintains the delay but as network complexity
increases it does not control as in less complexity network
but still, it maintains delay as shown in the graph.
Figure.5. e2e Delay (s) Vs No of Nodes
5.4 Packet Deliver Ratio
Packet deliver ratio (PDR) is the proportion of packets get
to the quantity of packets sent. From the graph in Figure 6,
it can be observed that all TCP variants perform identically
and deliver the more packets in overall time. Little bit TCP
Reno still performs well as compare to other TCP variants.
It maintains itself for delivering the packets. Other TCP
variants also deliver packets but still little bit variation
occurred in the graph but it can be negligible.
Figure.6. Packet deliver ratio (PDR %) Vs No of Nodes
6. Summarized results
Our objective was to look at the four variations of
transmission control protocols to each other, not to trace the
ideal execution possible in our situations, we noticed that
the network complexities do impact the execution of TCP
variations. This conclusion is reliable with the perception
of past investigations. In any case, contrary to other
investigations that looked at variants of transmission
control protocol (TCP) under an assortment of performance
parameters, a particular victor may not be certain for our
case where a real scenario was employed since various
execution measurements and network evaluations appear to
give unstable execution rankings.
At last, Table 3. does precisely the empirical
examination of selected TCP variants for MANET network
mentioned in this paper. It gives correspondingly, the TCP
variants name, number of nodes to show the network
complexities, and the QoS metrics, where, H, M and L
mean High, Medium and Low execution, individually. In
case of PDR and throughput, they must be high for the
good performance. And in case of jitter and end to end
delay, as they are low as the performance is sufficient.
Table 3. Summary of measured results
Quality of Service Parameters
( H- M -L ) TCP
Variants
No of
Nodes PDR Throughput Jitter
e2e
Delay
TCP
10 H H L L
20 H M L H
30 H M L H
40 H M H H
50 H L H M
TCP
Reno
10 H M L M
20 H M L M
30 H M M M
40 H H M L
50 H H M L
TCP
New
10 H H L M
20 H M L H
K. H.MOHAMMADANI et.al. EMPIRICAL MODELLING OF TCP IN MANET
Copyright ©2017, ESTIRJ-VOL.1, NO.2 (22-27)
Reno 30 H M M H
40 H M M H
50 H M M M
TCP
Vegas
10 H H L L
20 H M L L
30 H M M L
40 H L H L
50 H L H L
7. Conclusion and future directions
We executed the various transmission control protocols
known as TCP-variants in this paper. We have especially
run around MANET to examine the execution of the TCP
variants. The less impact of network complexity has been
analyzed on TCP Reno. TCP, TCP New Reno, and TCP
Vegas degraded the throughput in high network
complexities and they have got more packet delay time in
term of an end to end delay and jitter time to deliver the
packets as compared to TCP Reno. TCP Reno could be
better from network complexity as the quality perspective
of QoS metrics in the MANET.
In the future, we will compare TCP variants in the
integrated wireless network in relation to different reactive
and proactive routing protocols with multiple types of
mobility models.
References
[1] Mohammadani, Khalid, et al. "A Survey on Integrated
Wireless Network Architectures." International Journal of
Computer Applications 79.4 (2013).
[2] Jabeen, Qamar, et al. "Performance improvement in multihop
wireless mobile ad-hoc networks." the Journal Applied,
Environmental, and Biological Sciences (JAEBS) 6 (2016):
82-92.
[3] Chowdhury, Kaushik R., Marco Di Felice, and Ian F.
Akyildiz. "TCP CRAHN: A transport control protocol for
cognitive radio ad hoc networks." IEEE Transactions on
Mobile Computing 12, no. 4 (2013): 790-803.
[4] Zhou, Dizhi, Wei Song, Ping Wang, and Weihua Zhuang.
"Multipath TCP for user cooperation in LTE networks."
IEEE Network 29, no. 1 (2015): 18-24.
[5] ABBASI, S., et al. "Performance Analysis of MANET
Routing Protocols with UDP and TCP under VBR
traffic." Sindh University Research Journal-SURJ (Science
Series) 47.4 (2015).
[6] Domański, Adam, Joanna Domańska, Michele Pagano, and
Tadeusz Czachórski. "The Fluid Flow Approximation of the
TCP Vegas and Reno Congestion Control Mechanism." In
International Symposium on Computer and Information
Sciences, pp. 193-200. Springer International Publishing,
2016.
[7] Bisoy, Sukant Kishoro, and Prasant Kumar Pattnaik.
"Fairness between TCP Reno and TCP Vegas in wired and
wireless network." International Journal of Computational
Systems Engineering 3, no. 1-2 (2017): 14-26.
[8] Cui, Yong, Lian Wang, Xin Wang, Yisen Wang, Fengyuan
Ren, and Shutao Xia. "End-to-end coding for TCP." IEEE
Network 30, no. 2 (2016): 68-73.
[9] Abolfazli, E., and Vahid Shah-Mansouri. "Dynamic
adjustment of queue levels in TCP Vegas-based networks."
Electronics Letters 52, no. 5 (2016): 361-363.
[10] ABBASI, S.; MOHAMMADANI, K.H, Hussain.Z, Awan
J.H SHAH.R. “Evolution of V2V Routing Protocols in
realistic scenario of National Highway NH-5, Pakistan”.
Science International Lahore, 2016, 4711-4714
[11] Iqbal, Muddesar, Muhammad Shafiq, Jin-Ghoo Choi, Hasina
Attaullah, Khawar Akram, and Xingheng Wang. "Design
and Analysis of a Novel Hybrid Wireless Mesh Network
Routing Protocol." International Journal of Adaptive,
Resilient and Autonomic Systems (IJARAS) 5, no. 3 (2014):
20-39.
[12] Rao, Nageswara. "Functioning Estimation of Tcp Variants
and Directing Etiquettes in Flexible Ad-Hoc Set of
Connections." IJRCCT 4, no. 8 (2015): 609-619.
[13] Dhaliwal, Gurvinder, Gagandeep Singh, and Sumit Kumar.
"Enhancing Quality of Service over 802.16 Network Using
Variants of TCP." 2015
[14] Goswami, Bhargavi H. "Experimental Based Performance
Testing of Different TCP Protocol Variants in comparison of
RCP+ over Hybrid Network Scenario." International Journal
of Innovations & Advancement in Computer Science IJIACS
ISSN (2014): 2347-8616.
[15] Elmannai, Wafa, Abdul Razaque, and Khaled Elleithy.
"Simulation based Study of TCP Variants in Hybrid
Network." arXiv preprint arXiv:1410.5127 (2014).
[16] Tsiknas, Konstantinos, and George Stamatelos.
"Comparative performance evaluation of TCP variants in
WiMAX (and WLANs) network configurations." Journal of
Computer Networks and Communications 2012 (2012).
[17] Chawhan, Manish Devendra, and Avichal R. Kapur.
"Performance Evaluation of TCP Variants in Wi-Fi Network
Using cross layer design protocol and Explicit Congestion
Notification." International Journal of Wireless & Mobile
Networks (IJWMN) Vol 3 (2011): 112-124.
[18] Chhabra, A., MVEC, J.I., Dhiman, A. and Joshi, M.,
Performance Evaluation of Variants of TCP Based on Buffer
Management over WiMAX. Ashish Chhabra et al ,
International Journal of Computer Science &
Communication Networks,4 (5), 17-24, 2014.
[19] Shaikh, Arshad, Danish Vasan, and Khalid H. Mohammadani.
"Performance Analysis of MANET Routing Protocols-A
Comparative Study." International Journal of Computer
Applications 83.7 (2013).
About authors
Khalid Hussian Mohammadani is a Computer Science
Professional with over five-year experience in teaching
networking, programming in C, Web development and
other CS and Telecommunication courses. His area of
interest includes wireless sensor network, mobile adhoc
network, image processing and blooming taxonomy
technology. He worked as a trainer in two workshop of
Network Simulator NS2 that held by department of
Telecommunication, Mehran university, Jamshoro, Sindh.
He obtained his BS in Telecommunication from University
of Sindh, Jamshoro, Sindh, Pakistan in 2008. He also
received M. Phil. Degrees in Computer Science from Isra
University, Hyderabad, Sindh Pakistan in 2015. At present
he is pursuing Ph.D. in Computer science at Isra university
Hyderabad, Sindh , Pakistan.
K. H.MOHAMMADANI et.al. EMPIRICAL MODELLING OF TCP IN MANET
Copyright ©2017, ESTIRJ-VOL.1, NO.2 (22-27)
Dr. Safi Faizullah received his Ph.D. in Computer Science
from Rutgers University, New Brunswick, New Jersey,
USA in 2002. He also received MS and M. Phil. Degrees in
Computer Science from Rutgers University, New
Brunswick, New Jersey, USA in2000 and 2001,
respectively. Dr. Faizullah also earned his BS and MS
degrees in Information and Computer Science from
KFUPM, Dhahran, KSA in 1991 and 1994,respectively.
His research interests are in computer networks, mobile
computing, wireless networks, distributed and enterprise
systems. He is a Professor of Computer Science at Islamic
University of Madina, Saudi Arabia.
Prof. Arshad Shaikh, is a computer science professional
with over 20 years of experience in teaching and R&D. In
addition to the core computer science, his areas of interest
include topics from basic sciences and applied engineering.
He obtained his Bachelors of Computer Science degree
from University of Karachi with a Gold Medal. He did his
MS (Computer Science) from Rutgers, the state university
of NJ, USA. He is presently associated with Isra
University, Hyderabad.
Ms. Nazish Nawaz received the MS in IT. She is currently
pursuing the Ph.D. in computer science at the Isra
university Hyderabad, Pakistan. She is also assistant
professor in department of computer science at university
of Sindh, Jamshoro, Pakistan. Her current research involves
in multicast routing protocols for MANET.
Rahat Ali Khan has about 10 years’ experience of
teaching of electronics and telecommunication subjects. He
is currently research and teaching Assistant at IICT
University of Sindh, Jamshoro, Sindh Pakistan. He received
M.Phil. in Telecommunication from University of Sindh,
Jamshoro, Sindh, Pakistan. He also received BS (hons) in
Electronics from University of Sindh, Jamshoro. His
research interest in WBAN and wireless sensor network.
top related