Recital Study of Various Congestion Control Protocols in wireless network

IOSR Journal of Computer Engineering (IOSR-JCE)
e-ISSN: 2278-0661,p-ISSN: 2278-8727, Volume 17, Issue 2, Ver. V (Mar – Apr. 2015), PP 129-138
www.iosrjournals.org
DOI: 10.9790/0661-1725129138 www.iosrjournals.org 129 | Page
Recital Study of Various Congestion Control Protocols in wireless
network
Mehta Ishani1
, UditNarayan Kar2
, Dr. Atul Gonsai3
1
(PG Student, Computer Science, GardiVidyapith,Rajkot,Gujarat, India)
2
(Research Scholar,Computer Science, Saurastra University,Rajkot, Gujarat, India)
3
(Associate Professor, Computer Science, Saurastra University,Rajkot, Gujarat, India)
Abstract: Congestion control in packet switching networks became a high priority in network design and
research due to ever-growing network bandwidth and intensive network applications. Dozens of various
congestion control strategies have been proposed, and more are forthcoming. Based on control theory concepts
congestion control can be viewed as control policy to achieve prescribed goals (e.g., round-trip delay or
throughput) in a distributed network environment. This paper discusses the advantages, disadvantages and the
applications of various congestion control protocols for wireless networks. It explores the motivation behind the
design of congestion control protocols which is suitable to large scale wireless network and abolish drawbacks
of the most widely accepted two algorithms of TCP and RCP.
Keywords: TCP- Transmission Control Protocol, RCP- Rate Control Protocol
I. Introduction
The use and demand of wireless network has increased significantly in past years. Wireless
environment is more flexible, easy to deploy and scalable than traditional wired setup. Besides that in wireless
network air is used as access medium it is also more sensitive to inferences and to congestion. The developed
congestion control protocols like Transmission Control Protocol (TCP) and Rate Control Protocols (RCP) do
not take into account the problems and particularities of wireless network. TCP is the most widely accepted and
used congestion protocol in internet. It used window based Additive Increase Multiplicative Decrease (AIMD)
strategy for congestion control [1]. However, TCP proved to be unsuitable for dynamic environment of wireless
network where nodes are mobile and frequently participate and departure from the flow. RCP is built on rate
based congestion control strategy. It relay on network interaction modules such as routers for congestion
control. Also it is proved that RCP performs better than TCP [6]. The above described protocols are also proved
heavy for implementation into light devices with the reference of OS, Processing, and Memory Requirement and
Power consumption. They perform complex computation which is consuming power, requiring more memory
for storing per packet states, heavy for Hot Spot Tethering and might get hang with heavy processing [3]. Some
of these performance problems led to the development of new congestion control protocols. In this paper we
present performance comparison of TCP, RCP+ and our proposed model Enhanced RCP in wireless network.
The paper organizes as follows. The next section II, III and IV describes the mechanism of different congestion
control protocols, TCP, RCP and RCP+.Section V briefly introduced our proposed approach and section VI
presents implementation and evaluation results. Finally section VII provides conclusion.
II. TCP
The TCP achieves two important functions: (1) Reliable and ordered delivery and (2) Congestion
control. Congestion can be detected by Packet loss or Packet delay. Solution of this problem is limiting sender‟s
transmission rate. Now question arise, at what rate should the data be sent for the current network path? [5].
Congestion control mechanism given in [1] is shown in figure 1.
Congestion control mechanism given in [1] is shown in Fig 1. Each TCP connection initiate with a pre-
configured small congestion window no larger than 4 Maximum Segment Size (MSS).The goal of Slow-Start is
to keep a new sender from overflowing network buffers, while at the same time increasing the congestion
window fast enough and avoiding performance loss while the connection is operating with a small
window.Slow-Start increases the congestion window by one MSS for each new acknowledgment received,
which results in the window doubling after each window‟s worth of data is acknowledged. With this exponential
increase, RTT log2W (where RTT stands for round-trip time) time is required to reach a window of size W.

Recital Study of Various Congestion Control Protocols in wireless network
Figure 1: TCP‟s congestion control mechanism [1]
A connection enters Slow-Start on newly starting or on experiencing a packet retransmission timeout,
and exits Slow-Start when it detects a packet loss or when the congestion window has reached a dynamically
computed threshold, ssthresh. More specifically, ssthresh is set to half of the current congestion window when
packet loss was detected. TCP exits Slow-Start to enter the Congestion Avoidance phase, where it continues to
probe for available bandwidth, but more cautiously than in Slow-Start. During periods when no packet losses
are observed, TCP performs an Additive Increase of the window size, by 1 MSS each time a full window is
acknowledged (i.e., increases the congestion window as:
cwnd = cwnd +1 (1)
on receiving each acknowledgment packet). And when Congestion occurs it decreases the window size by half
as given in equation below
cwnd=cwnd*
1
2
(2)
III. RCP
Researchers of paper [6, 7] have proposed Rate Control Protocol (RCP). In RCP, a router assigns a
single rate, R (t), through which all flows pass. Here, the basic idea is: If there is spare capacity available, then
share it equally among all flows. Furthermore if there is a queue building up, then the link capacity is
insufficient and the flow rate is decreased evenly. they have tried to minimize Flow completion time (FCT).
FCT is time from when the first packet of a flow is sent (in TCP, this is the SYN packet) until the last packet is
received. To minimize FCT for each router a well-known method is to use processor-sharing (PS) i.e. a router
divides outgoing link bandwidth equally among ongoing flows. Also they have proposed an equation for RCP:
R(t) = R(t - d0) +
α ∁−y t −β
q t
d0
N(t)
(3)
where,
d0 = a moving average of the RTT measured across all flows,
R(t−d0) = last updated rate,
C = link capacity,
y(t) = measured input traffic rate during the last update interval (d0 in this case),
q(t) = instantaneous queue size,
N(t) = router‟s estimate of the number of ongoing flows (i.e., number of flows actively sending traffic) at time t
α, β = parameters chosen for stability and performance.
Simulation results show that RCP performs better than TCP in terms of FCT and link utilization in
wired network [9]. It is also taken account that RCP possess routing overhead on network also due to having
complex computation and involving routers. Researchers of paper [8] have introduced two limitations of RCP:
(1) RCP will need to operate alongside existing non-RCP traffic, such as TCP and UDP, without adversely
affecting or being affected by the other traffic; and (2) RCP will need to operate in a network where some
routers are not RCP-enabled.

IV. RCP+
Authors [2] described a simple congestion control algorithm called RCP+ which reduces flow
completion time for diverse flow types to large extend for broad range of traffic conditions and network
situations. In RCP+, unlike XCP, RCP they are not following feedback mechanism but they are adapting the
ancient congestion window based congestion control mechanism. The reason behind sticking to the congestion
window for congestion control is that, while each second new flow are entering and moving out of network, it is
tough to obtain exact number of flows at particular RTT. This inspired us to stay with congestion window based
mechanism instead of feedback based mechanism. We set the ‗cwnd„ value completely based on the rate
computation, but that doesn„t impose any overhead over the router. In addition, we are setting the value of
‗maxcwnd„ to obtain the better start for our optimum data rate calculation at the initial stage.
RCP+ wins upon existing congestion control algorithm with three characteristics, 1) RCP+ is more
flexible to be implemented on wireless networks in comparison of RCP. 2) It allows multiple variants of TCP to
co-exist in the same networks with RCP+ and work with each other smoothly. 3) And it performs well in Large
scale wireless networking scenario too.The equation of RCP+ can be given as:
N(t) * R(t) = (α*C - α*y(t) – (β*q(t)/d)) (4)
Where,
d = moving average of RTT per interval,
R(t) = last updated rate,
y(t) = existing traffic observed in network,
q(t) = the instantaneous queue size,
C = link capacity and
N(t) = number of flows.
α (alpha) is a stability parameter and β (beta) is a performance parameter added to the equation to make the
rate stable and not aggressive.
Thus, the equation gives us the desired aggregate rate change in presence of traffic in the next interval.
Here, Rate kept same for each flow. “cwnd‟ congestion window value is having proportionate relationship with
the rate and so, “cwnd‟ value is set on the bases of the Rate computed. RCP+ is very acute to packet loss. This
equation was obtained by manipulating Rate equation of RCP [5]
V. Proposed Approach
Our proposed approach is based Improved AIMD and RCP+ algorithm. In our proposed approach we
use congestion window mechanism of Improved AIMD algorithm to use the spare capacity of congestion
window after occurrence of congestion event. The limitation of AIMD is that the algorithm does not include the
different arrival time of flows.[1]. So we have used modified equation of RCP+ algorithm.
RCP+ algorithm is implemented based on the theory of RCP. RCP+ is having the added advantage of
coexistence with other wired and wireless TCP, XCP, RCP and DCCP protocols. RCP+ is flexible like TCP and
so is expected to have wide implementation over current demands of Internet. [2]. In the theory of RCP there is
a concept of queue. RCP was originally implemented in wired network. In wireless network the essential point
is traffic is in bursty nature. Hence it is difficult to use queue concept while going to practical simulation. So we
modified the equation of rate change.
Here we come up with a new proposal of congestion control scheme enhanced RCP. Here we use
Improved AIMD mechanism as well as modified rate change equation of RCP+. Our aim is to gain benefits of
both schemes by eliminating each other‟s demerits.
Initially we defined congestion window size 4 MSS and start data transfer. If acknowledgment is
received means no congestion and increase congestion windows size by +1 and transfer data
If acknowledgment is not received then congestion would likely to be occurred then decrease
congestion window size by equation:
Cwnd = (cwnd *1/2) + K (5)
Where, k is the increment in congestion window size (w) in t cycles or epoch
Calculate the rate change by equation:
Rate = (α*C - α*y(t) )/N(t) (6)
Again check if rate is greater then congestion window size then increase congestion window size
otherwise transfer data.

Algorithm
1) Initially set congestion window size 4 MSS
2) Data transfer
3) If acknowledgement is received increase congestion window size by cwnd = cwnd +1 and repeat step 2
4) Otherwise decrease congestion window size by equation Cwnd = (cwnd *1/2) + K
5) Calculate the rate change by equation Rate = (α*C - α*y(t) )/N(t)
6) If rate >cwnd then go to step 2
7) Otherwise transfer data
Below figure shows the flow chart for the same.
Figure2 Flow chart for proposed algorithm
VI. Implementation
We have implemented proposed model in NS-2 simulator and version is 2.35 [10].Following are the
steps for modification we have done in C++ file of TCP to implement RCP++
Step 1: Go to the location where tcp.cc is located. There, under the class definition of Tcp agent, bind two
variables for its usability under the procedure of bind.
bind("bw_",&bw_);
bind("flow_",&flow_);
Step 2: Modify TcpAgent::window procedure

if (frto_ == 2)
{
return (force_wnd(2) <wnd_ ?
force_wnd(2) : (int)wnd_+4);
}
Step 3: Modify TcpAgent::slowdown procedure
double win, halfwin, decreasewin, k;
k = (windowd() / 2) -1;
if (cwnd_ <ssthresh_)
slowstart = 1;
if (precision_reduce_)
{
halfwin = (windowd() / 2)+k;
}
else
{
int temp;
temp = (int)((window() / 2)+k);
halfwin = (double) temp;
}
Step 4: Modify TcpAgent::processQuickStart procedure
Define following variables
intapp_rate, bw_=400000, flow_=4;
float alpha = 0.1, yt = 0.005;
Step 5: Add following lines to the procedure of processQuickStart
qs_requested_ = 0;
qs_approved_ = 0;
if (qsh->flag() == QS_RESPONSE &&qsh->ttl() == ttl_diff_ &&qsh->rate() > 0) {
app_rate = (int)((alpha*TcpAgent::bw_ - alpha*yt) )/TcpAgent::flow);
Step 6: Give the computed rate to qs_cwnd defined under the procedure of processQuickStart
if (app_rate>initial_window())
{
qs_cwnd_ = app_rate;
qs_approved_ = 1;
}
else
{ // Quick Start rejected
}
Step 7: Once the changes in tcp.cc are made, save and exit the editor. Next step is to open the terminal and
under super user, type following commands one by one.
./configure
Make clean
Make
Make Install
Once you have done this your TCP agent will become RCP enabled now run the tcl scripts and
obtained the results.Following table 1.1 present the configuration details of the simulation. We have created a
wireless network and
Table 1 Configuration Table
Layer Parameter Values
Application FTP FTP over TCP agent and RCP+
Configuration No of nodes 10 to 50 incremental

Mobility Maximum Speed 10 Mbps
Pause time 2 s
Simulation time 500 s
Traffic Type TCP/CBR
Rate 4.0 Mbps
Routing Protocol AODV
MAC Mac 802_11
PHY Propagation model Two ray ground
Antenna Omni
System OS Ubuntu 12.10
Processor Intel(R) Core(TM) i5
The values of throughput, packet delivery ratio and routing load for TCP, RCP+ and proposed model
are given in tables bellow. From these values graphs are plotted and performance of each protocol can be
evaluated.
Throughput: In general terms, throughput is the rate of production or the rate at which something can be
processed. When used in the context of communication networks, throughput or network throughput is the rate
of successful message delivery over a communication channel. The data these messages belong to may be
delivered over a physical or logical link, or it can pass through a certain network node. Throughput is usually
measured in bits per second (bit/s or bps), and sometimes in data packets per second or data packets per time
slot.It is clearly shown that after 20 nodes the packer delivery ratio stated decreasing so that is the bottleneck
area and after that congestion occurs.
Table 2 Throughput Statistics
Throughput
Nodes TCP RCP+ Proposed
10 210.17 215.22 232.22
20 458.35 446.3 460.95
30 633.91 599.13 569.76
40 646.97 588.9 633.41
50 585.45 576.54 618.02
Graph 1 Throughput graph
Packet Delivery Ratio:The ratio of the number of delivered data packet to the destination. ∑ Number of packet
receive / ∑ Number of packet send. This illustrates the level of delivered data to the destination. The greater
value of packet delivery ratio means the better performance of the protocol. It is observed that proposed model
have greater values of Packet Delivery Ratio than existing protocols.

Table 3 Packet Delivery Ratio Statistics
Packet Delivery Ratio
Nodes TCP RCP+ Proposed
10 0.8858 0.908 0.9803
20 0.991 0.9636 0.9961
30 0.9386 0.8871 0.8453
40 0.7242 0.6587 0.7106
50 0.5357 0.6457 0.5654
Graph 2 Packet Delivery Ratio graph
Routing load: Routing load is the ratio of data packets or receiving packets and routing packets. Greater value
of routing load indicates more impact of router overhead on the performance of the protocol. From below
statistics Routing load of proposed model is less than other protocols.
Table 4 Routing load Statistics
Routing load
nodes TCP RCP+ Proposed
10 0.155 0.11 0.08
20 0.194 0.184 0.102
30 0.282 0.501 0.484
40 0.753 0.939 0.744
50 1.453 1.637 1.244

Graph 3 Routing load graph
Link Utilization: Link utilization is an important parameter to evaluate in congestion control protocols. It can
be given as portion of link is used in transmitting and receiving packets. We have also obtained the percentage
of link is utilized with respect to node and with respect to time.
Table 5 link utilization with respect to nodes
link utilization %
nodes TCP RCP+ Proposed
10 26.13 26.76 28.88
20 57 55.5 57.32
30 78.83 74.51 70.85
40 80.46 73.23 78.77
50 72.81 71.7 76.86
Graph 4 Link utilization with respect to nodes

Table 6 link utilization with respect to time
link utilization %
Time TCP RCP+ Proposed
0 0 0 0
50 7.28 7.17 7.69
100 14.56 14.34 15.37
150 21.84 21.51 23.06
200 29.12 28.68 30.74
250 36.4 35.85 38.43
300 43.68 43.02 46.11
350 50.96 50.19 53.8
400 58.25 57.36 61.49
450 65.53 64.53 69.17
500 72.81 71.7 76.86
Graph 5 Link utilization with respect to simulation time
VII. Conclusion
In this paper various protocols used for congestion control in wired and wireless network are presented
and discussed their pros and cons. We have proposed a model and mechanism for congestion control, Enhanced
RCP, which is window based strategy that has its fundamentals associated with RCP and TCP. We also done
simulation of proposed model and evaluate it‟s performance with existing protocols. From the results we
conclude that proposed model performs best in wireless network.
Acknowledgements
We would like to thank University of Grant Commission(UGC) for providing us the infrastructure on
which we implemented our experimental study.
References
Journal Papers:
[1]. S. Floyd, “HighSpeed TCP for Large Congestion Windows,” RFC 3649 ,http://www.icir.org/floyd/hstcp.html, December 2003.
[2]. Dr.AtulGonsai, BhargaviGoswami&UditnarayanKar, “Newly developed Algorithm RCP+ for Congestion Control on large scale
Wireless Networks”, International Journal of Innovations & Advancement in Computer Science IJIACS ISSN 2347 – 8616 Volume
3, Issue 2Apri 2014
[3]. Dr.AtulGonsai, BhargaviGoswami, UditnarayanKar, “Evolution of Congestion Control Mechanisms for TCP and Non TCP
Protocols” Matrix Academic International Online Journal Of Engineering And Technology © MAIOJET. 2013
[4]. NanditaDukkipati& Nick McKeown, “Why Flow-Completion Time is the Right Metric for Congestion Control and why this means
we need New Algorithms”, ACM SIGCOMM Computer Communication Review, Volume 36,2006
[5]. Chia-Hui Tai, Jiang Zhu, NanditaDukkipati, “Making large scale deployment of RCP practical for real networks” IEEE INFOCOM,
2008
[6]. Mehta Ishani, UditnarayanKar, Dr.AtulGonsai, “Why RCP is better than TCP” International Journal for Scientific Research &
Development (IJSRD),Vol. 2, Issue 12, 2015
Proceedings Papers:
[7]. Dr.AtulGonsai, BhargaviGoswami, UditnarayanKar, “Design of Congestion Control Protocol for Wireless Networks with Small
Flow Completion Time” Proceedings on National Conference on Emerging Trends in Information & Communication Technology.
2013

Thesis:
[8]. NanditaDukkipati, “Rate Control Protocol (RCP): Congestion Control to make Flows Complete Quickly”, Ph.D. Thesis, Stanford
University, Stanford, California. 2007
Technical Reports:
[9]. NanditaDukkipati& Nick McKeown, “Processor Sharing Flows in the Internet”, Stanford HPNG Technical Report, Stanford
University, Stanford, California. 2007.
Books:
[10]. Book “Introduction to Network Simulator NS2” springer, 2009

Recital Study of Various Congestion Control Protocols in wireless network

More Related Content

What's hot (16)

Similar to Recital Study of Various Congestion Control Protocols in wireless network (20)

More from iosrjce (20)

Recently uploaded (20)

Recital Study of Various Congestion Control Protocols in wireless network