ArticlePDF Available

Proportional Study of TCP Variants over Heterogeneous Wireless Networks

Authors:

Abstract

Transmission control protocol (TCP) was initially designed for static networks to provide the consistent data delivery. The enhancement of TCP performance was also achieved with different types of networks with the introduction of new TCP deviations. However, there are still many factors that affect the performance of TCP. Mobility is one of the primary performance-affecting drivers in heterogeneous network. The research of state in this paper is to determine the best TCP variant from the mobility point of view. We simulate some TCP variants such as TCP-Tahoe, TCP-Reno, TCP-New Reno, TCP- Vegas, TCP-SACK and TCP-Westwood from the mobility point of view. The scenarios presented in this paper are supported by on-demand link weight (ODLW) routing protocol that helps find the efficient route with Quality of Service (QoS) provisioning. The scenarios are designed for the walking persons to vehicles, and particularly suited for rocky and deserted areas. To demonstrate the strength of these TCP variants, the scenarios are simulated and evaluated the QoS parameters such as round trip time (RTT) fairness, end-to-end delay, and the number of broken links. Finally, based on the outcomes, we identify the best TCP variant that could be used for several applications in the future over heterogeneous wireless networks.
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
24
Proportional Study of TCP Variants over Heterogeneous
Wireless Networks
Abdul Razaque
Computer Science Department
University of Bridgeport
CT-06604, USA
Khaled Elleithy
Computer Science Department
University of Bridgeport
CT-06604, USA
ABSTRACT
Transmission control protocol (TCP) was initially designed
for static networks to provide the consistent data delivery. The
enhancement of TCP performance was also achieved with
different types of networks with the introduction of new TCP
deviations. However, there are still many factors that affect
the performance of TCP. Mobility is one of the primary
performance-affecting drivers in heterogeneous network. The
research of state in this paper is to determine the best TCP
variant from the mobility point of view. We simulate some
TCP variants such as TCP-Tahoe, TCP-Reno, TCP-New
Reno, TCP- Vegas, TCP-SACK and TCP-Westwood from the
mobility point of view.
The scenarios presented in this paper are supported by on-
demand link weight (ODLW) routing protocol that helps find
the efficient route with Quality of Service (QoS) provisioning.
The scenarios are designed for the walking persons to
vehicles, and particularly suited for rocky and deserted areas.
To demonstrate the strength of these TCP variants, the
scenarios are simulated and evaluated the QoS parameters
such as round trip time (RTT) fairness, end-to-end delay, and
the number of broken links. Finally, based on the outcomes,
we identify the best TCP variant that could be used for several
applications in the future over heterogeneous wireless
networks.
General Terms
Wireless communication networks, experiments and
simulation.
Keywords
TCP variants, heterogeneous network, RTT, throughput,
broken links, end-to-end delay
1. INTRODUCTION
The fact is that the TCP was originally designed for the wired
network, and even its good performance is in the wired
network[1].The researchers have put their efforts for
improving the performance of TCP in wireless networks, and
all this work confirmed that TCP in its present structure is not
appropriate for MANET[2]. TCP has high overhead due to
packet loss that may cause the buffer congestion because the
MANET’s losses is due to an error or the frequently of the
mobility [3]. So, the TCP cannot be efficient in current
form[4]. Furthermore, the main reason of the weak
performance of the standard TCP in the wireless network is
the inability of discovering the lost packets[5]. So, we can
realize that both characteristics have the same reason of
packet loss, which is the based on the network congestion[6].
However, some new protocols have been introduced and
implemented to avoid packet loss due to congestion. We
evaluate some of the existing TCP variants, and then
demonstrate how they in a heterogeneous network particularly
MANET.
The MANET network is a particular network due to the
ability of its nodes to communicate with each other via
intermediate nodes. So, it can be set up in any remote areas
without infrastructure-based support. The nodes, which are
the part of MANET require the data and information from the
database. However, the database is available in a wired
network. Therefore, MANET can be integrated with wired
network to obtain the required data and information [7].
Some applications run over the database, which are
supported by TCP variants. This paper aims to exhibit the
weaknesses and strengths of TCP variants in the
heterogeneous network especially MANET. Simulation is
conducted using NS2. TCP variants use different congestion
control mechanisms, which include congestion window,
recovery mechanism, retry limit, maximum packet size and
backup mechanism[8]. To determine the performance for
each TCP variants, as new architecture and approaches are
required to find out complete behavior of the variants. This
motivation results to introduce such network to analyze the
effectiveness for each TCP Variants. The random waypoint
mobility model is also used for controlling the moments of
nodes. In order to analyze the impact of mobility in the
heterogeneous network, it is essential that underlying
mobility model should attain the realistic scenario or at least
significant features. To this conclusion, we deem that this
paper makes the reasonable contribution. The scope of this
paper is to analyze some existing TCP variants over
heterogeneous network. The focus of study is particularly
around the performance metrics such as throughput, RTT
fairness, end-to-end delay, and broken links in presence of
the mobility.
The remaining paper is organized as follows. In Section 2, we
present related work. In Section 3, we present heterogeneous
network design. In Section 4, we explain handover process. In
Section 5, we present simulation setup and analysis of the
result. In Section 6, we present the discussion of the results
and finally section 7 concludes the work and future
directions.
2. RELATED WORK
In this section, we present the salient features of existing TCP
variants. The performance of TCP Reno, New Reno, SACK
and Tahoe was evaluated in [9]with respect to goodput under
three routing protocols over static multi-hop network. The
authors demonstrated that Reno had edge over other TCP
variants. However, the scenario was not fully explained. Our
work is entirely different from their work because we use
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
25
ODLW protocol in heterogeneous network with the inclusion
of additional TCP variants; Vegas and Westwood.
Furthermore, our work discusses relatively all QoS
parameters. In our previous work [3], the performance of TCP
variants was compared deploying anchor point node (APN)
heterogeneous network using dynamic source routing
protocol. However, the previous work focused on the
throughput, and packet delivery ratio. The finding of this
paper gives complete knowledge about the behavior of each
TCP Variants in a heterogeneous environment. We narrowly
analyze all issues of MANET in presence of mobility and
showed their impact on the performance of each TCP
Variants. The static scenario of heterogeneous network is
simulated using the TCP Reno, TCP SACK and TCP
Westwood in [8]. The authors claimed that TCP Westwood
was better performer, whereas our work completely depends
on movement and speed of the nodes with Random waypoint
model. Moreover, we have thoroughly studied the behavior of
TCP Variants.
3. HETEROGENEOUS NETWORK
DESIGN
We have designed mobility-based scenarios in heterogeneous
network by combining the features of wired with wireless and
MANET networks to make reasonable communication even
in remote areas. The nodes, which make the possible
communication between different segments of the network are
called gateway (APN). The APN can play a role as
coordinator in the network. Three segments of networks are
jointly connected to make the heterogeneous network. The
APNs are located on different positions.
FA-1
HA FA-2
WIRELESS AREA
APN APN
MANET NETWORK
WIRED AREA
Fig 1: Heterogeneous Network Design
The APN of MANET has information about the nodes, and
these nodes are assigned the IPs locally through Dynamic
Host Configuration Protocol Server (DHCP). The APN that
is part of MANET is said as MANET anchor point node
(MAPN). Similarly, the node that is located at the area where
wireless range becomes weak is called Infrastructure-based
anchor point node (IBAPN). Both APNs can play a role as
coordinators and formulate possible communication for the
rest of nodes in fixed and MANET segment of the network.
The wired and wireless segments of network cover the urban
and suburbs areas of the urban environment and MANET
portion of the network covers the remote areas. The MANET
network is routed using ODLW protocol. We have created
twenty five traffic flows simultaneously to analyze the
performance of TCP variants. This heterogeneous network
could be suited for providing the healthy communication for
urban as well as remote areas depicted in Figure1.
4. HANDOVER PROCESS
This section presents an outline of initial connection setup
and handover process for MANET mobile node (MMN).
Figure 2 shows the timing diagram and describes the signals
involved in it. Initial connection setup and MMN handover
process can be defined by using following steps. Initially
the nodes, which are the part of MANET, establish the
communication with the corresponding node (CN). The
nodes establish the initial connection-setup and also send a
message through Current MANET anchor point node
(CMAPN). The process of connection-setup is described in
following steps:
a. Once CMAPN obtains the request for connection-
setup from MMN, then it forwards the message
“coordination request for connection setup” to
infrastructure-based anchor point node (IBAPN).
In response CMAPN also sends back message
“Reply for connection setup with CN” to MMN.
When MMN obtains the message from CMAPN,
then it waits until the initial connection is
established.
b. IBAPN forwards the message “forwarding
coordination request for connection setup” to
respective HA/FA within wired area. HA, /FA
informs the IBAPN with message “accept
coordination request” to CMAPN.
c. HA, /FA forwards message with “forwarding
initial connection setup” to CN and sends back
response to IBAPN “accept forwarding
coordination request”. When CN receives the
message, then it informs the HA/FA “accept initial
connection setup” with (MMN). With the
establishment of initial connection setup between
CN and MMN then, data exchange process is
started.
d. When MMN changes the location and moves to
another MANET then, it sends a request for
handover to new MANET anchor point node
(NMAPN) with th e message “request for joining”.
e. NMAPN sends the message “location change
forwarding message” (LCFM) to IBAPN for
informing the handover process and similar
message is forwarded to HA/FA, and finally to CN
for location update.
f. NMAPN forwards the LCFM to IBAPN and also
“update” to CMAPN. In response, CMAPN sends
acknowledgment (ACK) to NMAPN for the location
update.
g. When CN gets the message LCFM then, it sets the
connection again with MMN and message is
forwarded with “new connection setup in change of
location”.
h. With the establishment of the new connection, the
data exchange process is initiated.
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
26
CN HA/FA WIRELESS APN CURRENT MANET
APN NEW APN MANET NODE
ACCEPT INITIAL CONNECTION SETUP
COMMUNICATION PROCESS MANET
NODE
LOCATION
CHANGE
UPDATE
ACK
NEW CONNECTION SETUP IN CHANGED LOCATION
COMMUNICATION PROCESS
Forwarding Initial
Coordinating Request
for Initial Connection
Setup Process
Forwarding Initial
Coordinating Request
for Connection Setup
Process
COORDINATION
REQUEST FOR
CONNECT SETUP
REQUEST FOR
CONNECTION SETUP
WITH CN
REPLAY FOR
CONNECTION SETUP
WITH CN
WAITING FOR
CONNECTION
ACCEPT
FORWARD
COORDINATION
REQUEST
ACCEPT
FORWARD
COORDINATION
REQUEST
REQUEST FOR
JOINING
LOCATION
CHANGE
FORWARDING
MESSAGE
LOCATION
CHANGE
FORWARDING
MESSAGE
LOCATION
CHANGE
FORWARDING
MESSAGE
Fig 2: Handover timing process
5. SIMULATION SETUP AND
ANALYSIS OF RESULT
The purpose of this research is to investigate the results of
TCP variants under mobility-based heterogeneous network.
In this paper, we have critically evaluated the Performance of
TCP variants with respect to different mobility ratios and
analyzed the behavior of each TCP variant. The parameters
of interest include throughput, RTT fairness, in-order data
delivery, and control overhead. NS2.35 on redhat-10 is used
for simulation. The random waypoint mobile scenario is
generated. The simulator gives a proper model for signal
propagation and transmission range is 250 meter [11]. The
sensing and interference range is 550 meters. TCP New
Reno, Reno, Tahoe, SACK, Westwood and Vegas are
simulated and investigated on the same network so as to
ensure fairness and behavior of the TCP variants. The
moving speed of a node is randomly obtained through
uniform division [Vmin, Vmax].
The percentage of mobility means how many mobile nodes
move and resulting how many links break in the MANET.
ODLW routing protocol is used to route the data, which gives
better performance for routing in MANET and produces
minimum routing overhead. The rest of simulated parameters
are shown in Table 1.
Table 1: Simulated parameters in scenario
Name of parameters
Description
Transmission Range
250 meters
Sensing and interference
range
550 meters
Bandwidth of node
400 Kb/Sec
Number of nodes
225
Size of network
2400 * 1800 square meters
Size of MANET
400 * 400 square meters
Buffering capacity
100 Packets buffering capacity
at each node
Packet generating rate
4 packets/second
Data Packet size
1024 bytes
Simulation time
18 minutes
Initial pause time
30 Seconds
MAC protocol
IEEE802.11
Protocols
TCP Vegas, TCP SACK, TCP
NEW RENO, TCP RENO and
TCP TAHOE
Mobility model
Random waypoint mobility
model
Mobility (Speed of the
nodes)
10 m/sec, 20 m/sec
Routing Protocol
ODLW
Mobility
50%
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
27
5.1 Throughput at different mobility rates
We have simulated the scenarios over heterogeneous network
and analyzed the performance of TCP variants (Reno, Vegas,
New Reno, Tahoe, Westwood and SACK). We have counted
the acknowledged packets for each TCP variant and
analyzed the throughput performance. The Figure 3 shows
the throughput performance for each TCP Variant at the
maximum speed of 10 m/sec with random waypoint mobility
model. The performance gradually decreases of each TCP
variant from 5% to 50% moving nodes. The reasons for
decreasing the rate of delivered packet is the mobility. Since
two nodes move in the network that fewer links break and take
less time to recover, whereas 50% moving nodes cause more
time to recover from broken links. The mobility also degrades
the performance of TCP variants because mobility causes the
change of routing information in the network.
As a result, it causes long RTT and repeated timeouts. Due
to mobility, the receiver produces out-of-order segments
and in resulting, the receiver generates acknowledgments
(Ack) only for highest in-order-packets. This causes the
duplicate Acks. The reason of delivering the more packets
for TCP Vegas is that TCP Vegas retransmits the lost packets
after receiving the two duplicate acknowledgments. Whereas
competing TCP variants retransmit the segments after
receiving the three duplicate acknowledgments but in some
cases third (dupack) takes either long time or does not receive
third dupack and timeout expires. It is an advantage of Vegas
over competing TCP variants. Furthermore, TCP Vegas mostly
retransmits the lost segments before retransmission timeout
(RTO). The other reason is that TCP Vegas does not wait for
the loss to trigger congestion window (cwnd) reduction.
5 10 15 20 25 30 35 40 45 50
12100
12200
12300
12400
12500
12600
12700
12800
12900
13000
Number of Delivered Packets
Mobility rates at 10 m/sec
12000
TCP VEGAS
TCP N-RENO
TCP RENO
TCP TAHOE
TCP SACK
WESTWOOD
Fig 3: Number of delivered packets at different mobility
rates with 10 m/sec speed
Vegas possesses interesting approach regarding the congestion
because it estimates the level of congestion before it occurs
rather than attempting to avoid it. The level of congestion is
measured based onm the sample RTT and size of sending
window that is also the reason; the sender estimates the current
throughput against every RTT. TCP Tahoe and Reno have
delivered less packets than rest of TCP variants. TCP Tahoe
faces the problem due to repeatation of the slow-start phases
on each dropped segment, particularly when error is
transient and not constant. In this case congestion window
shrinks and bandwidth cannot fully be utilized. Fast
Recovery algorithm for TCP New Reno can degrade the
performance due to multiple losses of packets during the
single window because fast recovery algorithm can
manage only single loss per RTT[12].
TCP Tahoe and Reno variants have more difficulty to
differentiate between loss and congestion in the wireless
environment in IEEE 802.11.The performance degrading
factor for TCP Reno and Tahoe is also the size of the
congestion window because these variants cannot send data
during the timeout period, if mainly packets loss occurs[13].
TCP Reno and Tahoe avoid timeouts in case of multiple
consecutive losses occur. The major factor of degrading the
performance for TCP Tahoe has no support of fast recovery
algorithm. Figure 4 shows the throughput performance of TCP
Variants at the maximum speed of 20 m/sec. The performance
is affected by increasing the speed, and TCP New Reno is
severely affected.The reason of the weak performance for TCP
New Reno is also aggressive behavior of fast retransmission
algorithm whenever duplicate acknowledgment (dupack) are
received, and high mobility of nodes is available. Due to
aggressive behavior of fast retransmission algorithm, it is
difficult to deliver the packets even partial Acks are received to
sender. Multiple losses due to high mobility make the weak
performance of TCP New Reno because multiple losses cannot
be handled properly. As a result, the network becomes more
congested, and packets start to drop quickly and this claim is
already verified in [14]. Another performance degrading factor
relates to TCP New Reno is to take one RTT to perceive each
packet loss.
5 10 15 20 25 30 35 40 45 50
12100
12200
12300
12400
12500
12600
12700
12800
12900
13000
Number of Delivered Packets
Mobility rates at 20 m/sec
12000
TCP VEGAS
TCP N-RENO
TCP RENO
TCP TAHOE
TCP SACK
WESTWOOD
Fig 4: Number of delivered packets at different mobility
rates with 20 m/sec speed
5.2 Round Trip Time (RTT) Fairness of
TCP Variants
We show fairness by sharing the bandwidth among different
TCP variants based on the round trip time (RTT) depicted in
Figure 5. There are numerous reasons for RTT fairness as
one reason is to attain the equal bandwidth allocation where
the different competing flows may allocate similar
bottleneck. Long RTT consumes more resources than short
RTT in consequence long RTT produces discouraging
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
28
throughput. TCP Vegas gets higher throughput than other
TCP variants because slow start and congestion recovery
algorithms mostly influence the throughput[15].
0 50 100 150 200 250 300 350 400 450
80
160
240
320
400
480
560
640
720
800
Throughput (Kb/Sec)
Time (msec)
0
TCP VEGAS
TCP N-RENO
TCP RENO
TCP TAHOE
TCP SACK
WESTWOOD
Fig 5: RTT fairness of TCP variants
Hence, slow start and congestion recovery mechanism work in
a different way for each TCP variant because TCP Vegas
depends on the difference of expected and actual throughput.
The multiple losses can be retained by avoiding timeouts
because the TCP Vegas retransmits the lost segments after
receiving 2 (dupack). As, this is the reason before timeouts,
dropped segments are retransmitted, and better throughput is
obtained. Original feature for TCP Vegas is its congestion
detection mechanism because it shows the problems
concerning to fairness.
In congestion avoidance, the congestion detection algorithm
of TCP Vegas verifies every RTT that is a benefit of
TCP Vegas over the rest of TCP variants. Moreover, TCP
Tahoe, TCP Reno, TCP New -Reno and SACK reduce the
congestion windows more than once during the single RTT.
This is also the reason for unfairness and producing minimum
throughput, whereas RTT of Vegas reduces only once during
the RTT.
5.3 ENE TO-END DELAY at 10 m/sec and
20 m/sec
End-to-End delay is the average elapsed time for delivery of
individual data packets. All possible delays are included and
caused by routing discovery, transmission at the MAC layer
and queuing at the interface queue, etc. But successfully
delivered packets are calculated. We show trend for each TCP
variant in Figures 6 and 7. Vegas has minimum end-to-end
delay at the speed of 10 m/sec and 20 m/sec; whereas TCP
Reno and TCP Tahoe have almost similar maximum end-to-
end delay at 10 m/sec. As, at the speed of 20 m/sec, the
maximum delay has been noticed for TCP New Reno.
The reason for maximum end-to-end delay for TCP Reno and
TCP Tahoe at the speed of 10 m/sec is weakness of faster
transmission algorithms. Since, TCP Tahoe does not send
instant ACKs and depends on commutative ACKs. Therefore,
when packet is lost then it waits for timeout or pipeline is
emptied. It causes the high bandwidth -delay.
0 2 4 68 10 12 14 16 18
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
0.018
0.020
End-to-End Delay (Seconds)
Time (Minutes)
0
TCP VEGAS
TCP N-RENO
TCP RENO
TCP TAHOE
TCP SACK
WESTWOOD
Fig 6: Average End-to-End delay of each TCP Variants
at the maximum speed of 10 m/sec
TCP Reno behaves like TCP Tahoe whenever multiple
losses occur and multiple losses are perceived as single
segment losses. Another problem occurs with TCP Reno
when the size of the window is small; number of duplicates
ACKs are not detected for fast retransmission and to wait for
coarse-grained timeout. On the other hand, TCP New Reno
performs poorly by increasing the number of moving nodes,
and it takes long end-to-end delay. Reason is limitation of
retransmitting the single lost segment against per RTT, as a
result a large delay occurs in retransmitting the later lost
packets in the window. From other side, if the sender is
restricted by the receiver’s advertised window during
recovery time, then the sender is unable to utilize the
existing bandwidth successfully that takes long end-to-end
delay [16]. Minimum end-to-end delay for TCP Vegas is
fairness of retransmission algorithm when segment is lost
that TCP Vegas waits for 2 dupack and retransmit the lost
segments before expiry of timeouts. Furthermore, the RTT
of Vegas shrinks only once during the RTT.
0 2 4 68 10 12 14 16 18
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
0.018
0.020
End-to-End Delay (Seconds)
Time (Minutes)
0
TCP VEGAS
TCP N-RENO
TCP RENO
TCP TAHOE
TCP SACK
WESTWOOD
Fig 7: Average End-to-End delay of each TCP variants at
the maximum speed of 20 m/sec
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
29
5.4 Broken links at mobile node-speed
We show an average broken links for each TCP variant at the
speed of 10 m/sec and 20 m/sec given in Figure 8. These
broken links are calculated at the 50% moving nodes. When
MANET nodes want to establish the sessions to obtain
internet services from wired segment of the network then
routing protocols start route discovery process. Route
Request packet (RREQ) is broadcasted into network to
obtain any single appropriate route to the destination.
5 10 15 20 25 30 35 40 45 50
8
16
24
32
40
48
56
64
72
80
Number of Broken Links
Mobility rates of nodes
0
10 m/sec
20 m/sec
Fig 8: Broken links at 10 m/sec and 20 m/sec
When route request packet is reached destination, in response
route reply packet (RREP) is sent to originator RREQ. If a
link is broken due to mobility and speed of intermediate
nodes, a route error packet is sent to the destination.
Meanwhile, destination finds another route. The process is
repeated until the reply reaches to the targeted node.
Therefore, destination finds another route if any error
occurs in the current route.
This process causes delay in packet delivery. The high
mobility and speed makes more broken links due to the
discovery of the route. The high mobility and speed
continuously change the direction of the node, in consequence
more links break. Due to increase of speed, topology changes
rapidly and more links are broken particularity in ODLW
when more connections are established among the nodes.
6. DISCUSSION OF RESULTS
We have simulated TCP Variants in NS2 over heterogeneous
network, which consists of MANET, wired & wireless
segments of the network. We have increased the mobility in
MANET segment of the network by using ODLW routing
protocol. Based on the simulation results, we analyzed that by
increasing the mobility and speed of the nodes, the
performance of each TCP variant gradually decreases. Multiple
routes obviously give benefits, but creates disadvantage due to
high mobility. In larger networks, the source routing principle
can also generate a trouble. It has been observed that TCP
Vegas performed better than other TCP variants. TCP Vegas
produces healthy throughput, better in-order-delivery of data,
minimum end- to-end delay, and good RTT at different
mobility ratios and speeds. The major reasons for degraded
performance of TCP Reno, Tahoe and New Reno are timeouts.
As, during the timeouts period, they cannot resend the lost
segments, whereas TCP Vegas does not wait for the loss to
trigger [17].
Westwood greatly miscalculates the existing bandwidth, which
is potentially troublesome for fairness. As a result, it can lead
to starvation of simultaneous connections that is the reason to
produce lesser throughput than TCP Vegas and also validated
the reason in the work presented in [18]. Hence, Losses in
wired network are due to overflow of the buffer at routers.
TCP Reno, Tahoe and New Reno have been designed
particularly for wired network and meet the requirement of
IEEE802.11, but their performance become weaker in
heterogeneous network especially satellite link is involved.
Minimum mobility ratios create less control overhead, which
causes the better performance for each TCP Variants, which is
also proved in our simulation. TCP variants experience the
most losses in multi-hop wireless networks, which are caused
by packet drop at wireless link layer IEEE 802.11. To improve
the performance, new congestion control flavors have been
introduced, and various schemes are included.
Explicit congestion Notification (ECN) has been incorporated
to improve the congestion control. If the congestion occurs in
the network, then the intermediate routers will mark the
congestion experience (CE) code point in the header of TCP
[19]. The comparison of TCP variants is given in Table-2 that
shows the strengths and weaknesses.
7. CONCLUSION AND FUTURE WORK
Mobile Adhoc networks can be deployed to different locations
without the use of infrastructure support. In military
environment, disaster situation, scattered educational
institutions need such networks to route data packets through
dynamically mobile nodes. MANET is a better choice for
these extremely mobile and dynamic applications, which are
not supported by centralized administration. If internet
services are required that MANET is a better solution in
anywhere to integrate with a wired network to construct as
heterogeneous network to obtain an internet facility.
To investigate the performance of different transmission
control protocols, we have conducted the simulation in NS-2
by using random waypoint mobility model and analyzed
different metrics. We have particularly focused on MANET &
wired portions of the network to investigate the performance of
TCP variants. The minimum effect of mobility has been
analyzed on TCP Westwood and reasons are already
discussed in detail but it delivers lesser segments than TCP
Vegas and SACK whereas TCP Vegas has better throughput,
minimum end-to-end delay, better in-order delivery of data
and improved RTT. TCP SACK also performs better and does
not loss many segments because sender is informed, which
segment has been received. TCP SACK uses SACK blocks at
receiver side to indicate the contiguous block of data
successfully received.
The sender can find out through SACK blocks that segments
are lost as this is the reason to control the loss of segments
frequently. TCP Reno, TCP New Reno and Tahoe degrade
the throughput in high mobility ratios and take more end-
to-end delay time as compared to other TCP variants. In
future, we will analyze and evaluate TCP Variants in
heterogeneous network with respect to different mobility
models including social network model, random walk
mobility model, random direction mobility model, city section
mobility model, etc.
8. REFERENCES
[1] S. Zafar and B. B. F. Town, "A Survey of Transport
Layer Protocols for Wireless Sensor Networks,"
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
30
International Journal of Computer Applications (0975-
8887) Volume, vol. 33, 2011.
[2] Razaque, A., S. Banuri, K. Shahzad, M. A. Qadir, and M.
Ahmad. "Analyis of TCP variants at various link error
rates under mobility aware Anchor Point Node Hybrid
Network." In Communications, Signals and Coding,
2008. MIC-CSC'08. Proceedings of the Mosharaka
International Conference on, pp. 60-65. IEEE, 2008.
[3] Razaque, Abdul, Khurram Shahzad, and M. Abdul Qadir.
"Performance Evaluation of TCP Variants in mobility
based Anchor point node Hybrid Network." International
IEEE/ACM proceedings on Communication Networks
and Services Research (CNSR): 270-277.
[4] D. Kim, J.-C. Cano, P. Manzoni, and C. Toh, "A
comparison of the performance of TCP-Reno and TCP-
Vegas over MANETs," in Wireless Communication
Systems, 2006. ISWCS'06. 3rd International Symposium
on, 2006, pp. 495-499.
[5] F. Yunus, N.-S. Ismail, S. H. Ariffin, A. Shahidan, N.
Fisal, and S. K. Syed-Yusof, "Proposed transport
protocol for reliable data transfer in wireless sensor
network (WSN)," in Modeling, Simulation and Applied
Optimization (ICMSAO), 2011 4th International
Conference on, 2011, pp. 1-7.
[6] O. Oyedapo and D. Ngwenya, "Evaluation of TCP-
variants performances in an ad-hoc mobile network," in
AFRICON 2007, 2007, pp. 1-7.
[7] S. Henna, "A throughput analysis of TCP variants in
mobile wireless networks," in Next Generation Mobile
Applications, Services and Technologies, 2009.
NGMAST'09. Third International Conference on, 2009,
pp. 279-284.
[8] A. Oluwatope, A. Obabire, G. Aderounm, and M.
Adigun, "End-to-End Performance Evaluation of
Selected TCP Variants across a Hybrid Wireless
Network," Issues in Informing Science and Information
Technology, vol. 3, 2006.
[9] S. Papanastasiou, M. Ould-Khaoua, and L. Mackenzie,
"On the Evaluation of TCP in MANETs," in
International Workshop on Wireless Ad Hoc Networking
(IWWAN’05), 2005.
[10] M. Musolesi, S. Hailes, and C. Mascolo, "An ad hoc
mobility model founded on social network theory," in
Proceedings of the 7th ACM international symposium on
Modeling, analysis and simulation of wireless and
mobile systems, 2004, pp. 20-24.
[11] A. Lo, J. Zhou, M. Jacobsson, and I. Niemegeers, "ns-2
models for simulating a novel beyond 3G cellular multi-
hop network," in Proceeding from the 2006 workshop on
ns-2: the IP network simulator, 2006, p. 7.
[12] K. Yusung, C. Kilnam, and X. Lisong, "Adjusting the
aggregate throughput of parallel TCP flows without
central coordination," IEICE transactions on
communications, vol. 91, pp. 1615-1618, 2008.
[13] D. Kim, H. Bae, J. Song, and J.-C. Cano, "Analysis of
the interaction between TCP variants and routing
protocols in MANETs," in Parallel Processing, 2005.
ICPP 2005 Workshops. International Conference
Workshops on, 2005, pp. 380-386.
[14] Z. Fu, H. Luo, P. Zerfos, S. Lu, L. Zhang, and M. Gerla,
"The impact of multihop wireless channel on TCP
performance," Mobile Computing, IEEE transactions on,
vol. 4, pp. 209-221, 2005.
[15] S. Ha, Y. Kim, L. Le, I. Rhee, and L. Xu, "A step toward
realistic performance evaluation of high-speed TCP
variants," in Fourth International Workshop on Protocols
for Fast Long-Distance Networks, 2006.
[16] H.-Y. Hsieh, K.-H. Kim, and R. Sivakumar, "An end-to-
end approach for transparent mobility across
heterogeneous wireless networks," Mobile Networks and
Applications, vol. 9, pp. 363-378, 2004.
[17] M. Hassan and R. Jain, High performance TCP/IP
networking: Prentice Hall, 2003.
[18] S. Mascolo, L. A. Grieco, R. Ferorelli, P. Camarda, and
G. Piscitelli, "Performance evaluation of Westwood+
TCP congestion control," Performance Evaluation, vol.
55, pp. 93-111, 2004.
[19] Razaque, A., K. Shahzad, and M. A. Qadir. "Analytical
comparison of TCP Variants under mobility aware
Anchor Point Node (APN) Hybrid Network." In
Communications, Signals and Coding, 2008. MIC-
CSC'08. Proceedings of the Mosharaka International
Conference on, pp. 66-71. IEEE, 2008.
BIOGRAPHY
Abdul Razaque is Editor-in-Chief for International Journal for
Engineering Technology (IJET) Singapore, He holds
fellowship form Higher Education Commission (HEC)
Pakistan, and Common Wealth, UK. He served as Head of
computer science department in Model colleges setup
Islamabad, Pakistan from 2002 to 2009. He also led several
projects as project Director for promoting the trend of
information technology (IT) in Pakistan funded by United
Nation organization (UNO) and World Bank during 2005 to
2008. He is currently active researcher of wireless and Mobile
communication (WMC) laboratory, UB, USA. Abdul
Razaque has also been working as Chair, Strategic Planning
Committee for IEEE SAC Region-1. USA and Relational
Officer for IEEE SAC Region-1 for Europe, Africa and
Middle-East. Abdul Razaque has chaired more than dozen of
highly reputed international conferences and also delivered
his lectures as Keynote Speaker. His research interests include
the wireless sensor networks, design and development of
learning environments, TCP/IP protocols, multimedia
applications and ambient intelligence.
Dr. Khaled Elleithy is the Associate Vice President for
Graduate Studies and Associate Dean for Graduate Studies in
the School of Engineering at the University of Bridgeport. He
has research interests are in the areas of network security,
mobile communications, and formal approaches for design
and verification. He has published around three hundred
research papers in international journals and conferences in
his areas of expertise. Dr. Elleithy is the co-chair of the
International Joint Conferences on Computer, Information,
and Systems Sciences, and Engineering (CISSE). CISSE is
the first Engineering/Computing and Systems Research E-
Conference in the world to be completely conducted online in
real-time via the internet and was successfully running for six
years. Dr. Elleithy is the editor or co-editor of 12 books
published by Springer for advances on Innovations and
Advanced Techniques in Systems, Computing Sciences and
Software.
International Journal of Computer Applications (0975 8887)
Volume 108 No 20, December 2014
31
APPENDIX:
Table 2: Benchmark of TCP Variants: Tahoe, Reno, New-Reno, SACK, Vegas, Westwoo
TCP
Variants
RTT
Throughput
Average
Broken Links
with 50%
mobile nodes
End-to-End
Delay
Mobility-
aware
Packet Drop %
Efficiency
Tahoe
0.074
335 KB/Sec
34
Moderate
Moderate
0.0013
0.9987
Reno
0.072
334.6 KB/Sec
35
Moderate
Moderate
0.0009
0.9991
New Reno
0.069
336.1 KB/Sec
31
Moderate
Moderate
0.0008
0.9992
SACK
0.070
332.9 KB/Sec
25
Moderate
Moderate
0.0011
0.9989
Vegas
0.062
344.1 KB/Sec
14
Low
High
0.0006
0.9994
Westwood
0.068
337.2 KB/Sec
19
Low
High
0.0007
0.9993
IJCATM : www.ijcaonline.org
... They have shown the fairness of RTT and identified that increased value of RTT mean it uses more resources and this gives degradation in throughput. They have also analyzed that with the increasing speed TCP's performance degrades [9]. ...
... Their mechanism doesn't depend on RTT value. While with proportional study of TCP's different variants, authors Razaque and Elleithy, 2014 have studied about the fairness of RTT [9]. They identified that increased value in RTT means, more resources are utilized and hence it causes down performance. ...
Article
Full-text available
Transmission control protocol provides congestion control mechanisms. These mechanisms work well for mobile adhoc network and wireless networks. TCP provides different mechanisms for both networks. TCP suffers from the packet loss problem in both networks. TCP controls congestion as per type of loss. The mobile adhoc network faces two types of packet loss, 1) loss due to congestion 2) loss due to link failure. TCP can not identify these loss types and it assumes loss as congestion loss. It has existing variants which perform congestion control. One of TCP variant TCP westwood works for congestion control with the help of bandwidth estimation. It works well but has not taken care the value of last Round trip time and estimated round trip time which are the important parameters to analyze network status. The proposed westwood has taken care of these two parameters and controls congestion window based on them. The work proposed in this paper focuses utilization of round trip time for throughput improvement. The proposed algorithm is tested on NS 2.35 and compared with existing variants like TCP Reno, TCP Vegas and TCP westwood and result reveals certain improvements. Keywords: MANET (mobile adhoc network), cwnd (congestion window), ssthresh (slow start threshold), RTT (round trip time) ,RTO (retransmission time out) ,DCM (dynamic congestion control for MANET), ERTT (estimated round trip time)
Article
Full-text available
With breakthrough of technological advancement, the significance of data transmission has been in highly demanding. On the other hand, limited buffering capacity has been great challenge that limits the Quality of Service (QoS) and degrade the performance of the network particularly in Wireless Mesh Networks (WMNs). Thus, it is important to provide an efficient utilization bandwidth and buffer management. Further- more, the QoS in WMNs depends immensely on intelli- gent buffer management to avoid unexpected congestion and data loss. Some algorithms have been introduced to improve the buffer capacity and management. However, these suffer from high latency, even the loss of data because of congestion in the buffers, eventually resulting in low throughput. To address these issues we introduce the scheme “Smart Bandwidth Friendly Buffers (SBFB) for Wireless Mesh Networks. The SBFB is inspired by the features of existing schemes like the EZ Flow, WRED (Weighted Random Early Detection, and Back-off mecha- nism algorithms. The SBFB scheme contributes to prior- itize the packets, allowing the down link nodes to perceive the buffer capacity prior to transmission and it also hunts for alternative routes in case of buffer buildup. Our proposed algorithm is validated using Network simula- tor-3 (NS3). Based on the experimental results, we have determined that SBFB has lesser congestion and packet loss probability when compared to other known contem- porary schemes.
Article
Full-text available
The explosion of new promising technology and their requirement to communicate transparently has signified the need of efficient mobility management in the communication. The emerging networks have not controlled the communication gap between urban and rural territories. To remove the gap of data communication in both territories, we have introduced Hybrid Network with Anchor Point Node (APN), which is particularly effective for rural areas. The proposed network has been simulated with different static & mobility based scenarios with some existing TCP flavors. Therefore, to analyze the taxonomy & behaviors of TCP variants, as various link error rates are created in APN Hybrid Network. The main focus of the paper is to report the good put performance of TCP flavors at different link error rate and effect of link error rate in loss of good put.
Article
Full-text available
High Performance TCP/IP Networking: Concepts, Issues, and Solutions is a comprehensive guide to the study of its topic. Our book provides an in-depth coverage of (1) tools and techniques for the performance evaluation of TCP/IP networks, (2) performance concepts and issues for running TCP/IP over wireless, mobile, optical, and satellite networks, (3) congestion-control algorithms in hosts and routers, and (4) high performance implementation of TCP/IP protocol stack. This text has been created with an emphasis on fundamental concepts, such as network measurement and simulation techniques, mathematical modeling of TCP dynamics, and management of implementation overhead, which will continue to guide new developments in TCP/IP. Although many specific networks, tools, and protocols are discussed in the text, a continuous effort has been made to emphasize the underlying performance issues and concepts.
Article
Full-text available
Reliable transport protocols such as TCP are tuned to perform well in traditional networks where packet losses occur mostly because of congestion. TCP is intended for use as a highly reliable end-to-end transport protocol for transporting applications such as World-Wide Web (WWW) between hosts in packet-switched computer communication networks. TCP was originally de- signed for wired links where the error rate is really low and actually assumed that packet losses are due to congestion in the network. However, the increasing popularity of wireless networks indicates that wireless links will play more important role in future internetworks but TCP per- formance in such networks suffers from significant throughput degradation and very high interac- tive delays. TCP responds to all losses by invoking congestion control and avoidance algorithms, resulting in degraded end-to-end performance in wireless and lossy systems. Thus, in a bid to show and determine the possibility of adapting TCP protocol for optimal performance on the wireless link, this paper reviews and models the behaviors of TCP variants with a view to evalu- ate the end-to-end performance analysis of TCP versions: TCP Reno, TCP SACK and TCP Westwood (TCPW), which are designed to improve the performance of TCP in lossy networks. A wireless network model was developed using NS-2 network simulator which and the model was simulated. The results were analyzed in MATLAB 6.5 using throughput as a metric for compari- son. The overall results indicate that TCP Westwood (TCPW) demonstrates better performance indices over other versions in a hybrid wireless network environment.
Conference Paper
Full-text available
Reliable data transport is one of the most important requirements in wireless sensor network where different applications have different reliability requirements. Additionally, the characteristic of wireless sensor network, especially dense deployment, limited processing ability, memory and power supply, provide unique design challenges at transport protocol. Therefore, assuring reliable data delivery between sensor nodes and the sink in Wireless Sensor Networks (WSN) is a challenging task. A reliable protocol in wireless sensor network is a protocol that allows data transfer reliably from source to destination with reasonable packet loss. The current issues of transport protocol are how to implement reliable data transport, congestion control and energy efficient. Most of the existing transport protocols only provide reliable data transport or congestion control. However, there are several protocols that provide both functions of the transport protocol. To overcome these issues, the transport layer protocols that provide both reliable data delivery and congestion control should be taken under consideration. Besides that, transport layer algorithm also allow maximum network lifetime due to limited operating lifetime of sensor node. Thus, to prolong the lifetime of wireless sensor network, an efficient transport protocol need to support reliable message delivery and provide congestion control in most energy efficient. This paper focuses on the existing transport protocols and the future protocol that provide the entire requirement of transport protocol.
Conference Paper
Full-text available
The priority of scientists and researchers is to provide guaranteed communication everywhere. Different strategic methodologies and schemes have been introduced to connect the entire world. The substantial efforts have been made for optimized and fast data delivery so for. TCP plays pivotal role in information-oriented era. Various TCP flavors are proposed for fast and rapid communication. We simulate TCP Variants on our proposed anchor point node (APN) hybrid network. The simulated scenarios are purely mobility-based and meet the realistic situation; as MANET portion of network covers the rural area and wired-cum-wireless cover city and its suburbs. We use DSR protocol for route discovery in MANET portion. Our main focus is to investigate the fairness of TCP Variants on basis of mobility scenarios to evaluate the throughput, good put and end-to- end delay.
Conference Paper
Full-text available
The prevalence of wireless networks has become imminent over the past few years, because of the unpredictable nature of their channels; they introduce different types of errors. The congestion control algorithm of the standard transmission control protocol (TCP) assumes packet losses are due to congestion and hence not suitable to handle such channel and transmission errors. This leads to ineffective utilization of the channel bandwidth. A number of TCP variants have been developed to cater for different types of networks. In this paper we evaluate the performances of some of these variants in ad hoc mobile network by simulations using NS-2. These variants are then evaluated in terms of the congestion window evolution and throughput. We showed by the results of our simulations that the BITCP is the most stable with the packet error rates introduced, followed by the TCP-Jersey.
Article
The reliability of data transfer is vital for commercial and enterprise applications of Wireless Sensor Networks (WSN). Likewise, mission-oriented and critical military applications of these networks demand dependable and timely data transport. This reliability is required for in-bound data, from Internet node to sensor nodes which comprises of code updates, as well as for out-bound data from sensor nodes to base station or gateway which comprises of important data reported by sensor nodes. Although TCP is a time-tested transport layer protocol of Internet that ensures reliability, flow control and congestion control, being a heavy protocol, it is considered unsuitable for resource constrained sensor networks. As a result new transport layer protocols have been developed for these networks. Nonetheless efforts are directed towards making TCP suitable for sensor networks. This paper presents a survey of transport layer protocols and approaches to achieve reliable data communication in general wired-cum-wireless networks and particularly in WSN.
Article
Westwood+ TCP is a sender-side only modification of the classic Tahoe/Reno TCP that has been recently proposed to improve fairness and efficiency of TCP. The key idea of Westwood+ TCP is to perform an end-to-end estimate of the bandwidth available for a TCP connection by properly counting and filtering the stream of ACK packets. This estimate is used to adaptively decrease the congestion window and slow-start threshold after a congestion episode. In this way, Westwood+ TCP substitutes the classic multiplicative decrease paradigm with the adaptive decrease paradigm. In this paper we report experimental results that have been obtained running Linux 2.2.20 implementations of Westwood+, Westwood and Reno TCP to ftp data over an emulated WAN and over Internet connections spanning continental and intercontinental distances. In particular, collected measurements show that the bandwidth estimation algorithm employed by Westwood+ nicely tracks the available bandwidth, whereas the TCP Westwood bandwidth estimation algorithm greatly overestimates the available bandwidth because of ACK compression. Live Internet measurements also show that Westwood+ TCP improves the goodput w.r.t. TCP Reno. Finally, computer simulations using ns-2 have been developed to test Westwood, Westwood+ and Reno in controlled scenarios. These simulations show that Westwood+ improves fairness and goodput w.r.t. Reno.
Conference Paper
Almost all work on mobile ad hoc networks relies on simulations, which, in turn, rely on realistic movement models for their credibility. Since there is a total absence of realistic data in the public domain, synthetic models for movement pattern generation must be used and the most widely used models are currently very simplistic, the focus being ease of implementation rather than soundness of foundation. Whilst it would be preferable to have models that better reflect the movement of real users, it is currently impossible to validate any movement model against real data. However, it is lazy to conclude from this that all models are equally likely to be invalid so any will do.We note that movement is strongly affected by the needs of humans to socialise in one form or another. Fortunately, humans are known to associate in particular ways that can be mathematically modelled, and that are likely to bias their movement patterns. Thus, we propose a new mobility model that is founded on social network theory, because this has empirically been shown to be useful as a means of describing human relationships. In particular, the model allows collections of hosts to be grouped together in a way that is based on social relationships among the individuals. This grouping is only then mapped to a topographical space, with topography biased by the strength of social tie.We discuss the implementation of this mobility model and we evaluate emergent properties of the generated networks. In particular, we show that grouping mechanism strongly influences the probability distribution of the average degree (i.e., the average number of neighbours of a host) in the simulated network.
Article
We propose an Adjustable Parallel TCP (AP-TCP) which is a new scheme to control the aggregate throughput of parallel TCP flows. The AP-TCP can adjust the aggregate throughput to be any desired level irrespective of the parallel size (the number of parallel TCP flows). To adjust the aggregate throughput, we modify the increment factor of each parallel TCP flow to K2/N2 where N is the number of parallel TCP flows and K is a value equivalent to any desired level for the aggregate throughput. Once K is given, the AP-TCP attempts to have K times more bandwidth than a single TCP flow when they are competing on the same network path. Another feature of the AP-TCP is its self-adjustment scheme. There is no central coordination or control overhead for parallel TCP flows. We analyze the model of the AP-TCP theoretically and evaluate it by using NS-2 simulation.