Communications and Network, 2013, 5, 649-653
http://dx.doi.org/10.4236/cn.2013.53B2116 Published Online September 2013 (http://www.scirp.org/journal/cn)
Link Utilization Based Multicast Congestion Control
Manisha Manjul*, Rajesh Mishra, Joytsna
Department of Computer Science & Engineering, School of I.C.T., Gauatam Buddha University, Greater Noida, India
Email: *manisham@gbu.ac.in, rmishra@gbu.ac.in, jyotsna.pro@gmail.com
Received June 2013
ABSTRACT
Networking is the importan t part of th e computer networks . Multicasting is also one of the esse ntial ar eas of networking.
In multicasting, sender can send information or data to the group of end receivers in a single transmission. It is a big
issue because of more data demands of receivers which cause the congestion. Multicast is one of the crucial method to
avoid congestion and make network perform stably. In this paper, we propose, link utilization algorithm to deal with
multicast congestion control. We propose the equation to improv e the utilization of the channel or link. The simulation
results show that proposed work provide the better throughput of the network with respect to sending rate, queue size,
packet loss ratio and time.
Keywords: Multicast; Congestion; Link Utilization; Congestion Control
1. Introduction
In data networking, network congestion occurs when packet
arrival rate exceeds the outgoing link capacity, slow
processing, bursty traffic, or due to insufficient memory
to store arriving packets. The causes include for this is
queuing delay, packet loss or blocking of new connec-
tions. Modern networks use the congestion avoidance
techniques, measures, algorithms or protocols to try to
avoid congestion collapse. Another method to avoid the
negative effects of network congestion is implementing
priority schemes, so that some packets are transmitted
with higher priority than others. The priority method does
not solve the congestion in network by themselves rather
they help to alleviate the effects of congestion for some
services. The other method for congestion control is of
congestion degree concept. In this method we are defin-
ing some threshold value for the congestion occurrence
and then calculate the Congestion degree. If the calcu-
lated congestion degree smaller than the predefined thre-
shold the transmission rate is increased else decreased.
Many ways had been given to control the congestion.
Some basics are slow start, congestion avoidance, fast
retransmission and fast recovery. The focusing section in
the multicast congestion control, their approaches and
protocols of these approaches of multicast congestion con-
trol: single rate and multi-rate [1]. Various mechanisms
have been proposed to control congestion in network.
Some mechanisms like TFMCC, LI MD, AIMD, PGMCC,
proactive approach in which congestion degree concept
is used which helps in adjusting the sending rate of the
source like that we also propose the congestion control
algorithm based on link utilization. For this, propose the
equations based on utilization to improve the channel
usage. This algorithm improves the initial sending rate by
giving the new sending rate for proper utilization of the
network or link. The simulation of this algorithm is done
using NS-2 tool and we have analyzed that throughput of
the network with respect to the sending rate, queue size,
packet loss ratio and time. We have observed that the
proposed work provide the better throughput of the net-
work.
2. Related Work
There are different mechanisms have been adopted till
now to control the congestion in the network. We know
that TFMCC [2] is a steady state equation based multi-
cast technique to calculate the throughput of the network
( )
( )
( )
()
( )
( )
TCP
2
X
S /RTTSqrt2p/ 312*Sqrt3p / 8p132p= ++
(1)
Where S is Packet Size, p is Packet loss Ratio, XTCP is
Throughput, and RTT is the Round trip time [3].
But it has some problems. First, it is slow in identify-
ing the Congestion representative and therefo re it is slow
in reacting to changes in the congestion condition. Se-
condly, the CLR drag down the whole TFMCC session.
Therefore, some modifications are made to TFMCC us-
ing Additive Increase Multiplicative Decrease (AIMD)
*Corresponding a uthor.
Copyright © 2013 SciRes. CN
M. MANJUL ET AL.
650
approach by proposing a new equation to calculate the
throughput [4-6]
( )
1/1 p1/1 p
AIMD
X/p /1
αβ β
−−
= −
(2)
Where α = S/RTT, β is decreasing factor (0 < β < 1)
Now a proactive approach is used for congestion pre-
vention. In this, the congestion is calculated as follows:
Fcon=1-Ϫn/d (3)
Where Ϫn = RTTn RTTn1 and dЄ [RTTmin, RTTmax].
RTTmin and RTTmax correspond to the min. and max. RTT
experienced within the multicast ses s ion .
Then, source adjusts its rate as
n 1ncon
XX *f
+
=
(4)
Where Xn is throughput before congestion, Xn+1 is
throughput after congestion, fcon is congestion factor. If,
Ϫn > 0 then source rate is decreased else increased [4].
Other way is that, if we want to improve the through-
put of the network we use Logarithmic Increase multip-
licative Decrease (LIMD) technique instead of AIMD
technique for throughput calculations and show the better
results. The LIMD equation as follows:
( )
LIMD 2
X21 logp
β
αβ
=+−
(5)
Where α = S/RTT, L is Packet Size, RTT is roun d Trip
Time, p is Packet Loss Ratio, β is Reduction factor.
Another way to control congestion in Wireless Sensor
Network is Fuzzy approach by calculating the congestion
degree (Cd)
()
d sa
CT /T
=
(6)
Where Ts = local packet inter-service time and Ta =
local packet inter-arrival time.
Once congestion is detected , it is notified by using Im-
plicit Congestion Notification (ICN) signaling. After re-
ceiving this signal the intermediate nodes adjust its send-
ing rate so that congestion doesn’ t occur. Further cong es-
tion is implemented using Fuzzy logic controller [7,8].
Here we define some threshold value. If the congestion
degree is greater than the defined threshold value, then
congestion occurs and ICN signal is sent to intermediate
nodes to inform that they adjust their sending rates; else
there is no congestion in the network.
The other mechanism uses the Fuzzy-logic-based Rate
Adaption (FRA) Scheme for TFMCC to control the con-
gestion in order to enhance the smoothness of TFMCC.
In order to alleviate oscillations of sending rate for TFMCC
sender, FRA introduces five actions for adjusting rate
[9-11].
These five actions are as follows:
a) MD (Multiplicative Decrease)
b) AD (Additive Decrease)
c) MI (Multiplicative Increase)
d) AI (Additive Increase)
e) KP (Keep)
And uses the Fuzzy Controller to decide which action
should be taken according to the feedback information
from CLR (Current Limiting Receiver). In dynamic net-
work environment, fuzzy controller uses the difference
between expected rate and sending rate to reflect the con-
gestion degree, as well as the difference between two
latest consecutive expected rates to predict the trend of
network. Under the fuzzy controller, KP, AD and MD
actions eliminate the “sawtooth” phenomenon in TFMCC,
which is crucial for smoothing sending rate [9]. When
the available bandwidth is turning abundant, MI action
can increase sending rate rapidly, making FRA have shorter
responding time and can fully utilize the resource. In
order to be friendly to TCP flows, the fuzzy controller
has unsymmetrical membership functions and biased in-
ference rules.
3. Proposed Work
We have seen that many algorithms have been proposed
to control the conges tion in the multicast network. These
algorithms used different protocols to reduce congestion
by adjusting the sending rate of the sender and different
mechanisms has been proposed to indicate the congestion
representative. The heterogeneous behavior of the net-
work leads to the more utilization of bandwidth which
results in congestion in the network. We propose an al-
gorithm to improve the utilization by keeping the same
sending rate while congestion occurs in the network. For
this we have to first calculate the link utilization using
existing link utilization method, then we again calculate
utilization by our own proposed formula.
Proposed Algorithm
We are proposing an algorithm to improve the utilization
by keeping the same sending rate while congestion oc-
curs in the network. For this we have to first calculate the
link utilization using old link utilization method. The
proposed algorithm entitles LUMCC is given below:
Algorithm: Link Utilization Based Multicast Conges-
tion Control (LUMCC)
1) Initialize the total link capacity.
2) Initialize the initial send ing rate.
3) Initialize the queue size.
4) Initialize the packet size.
5) Set the session time.
6) Calculate the packet loss ratio on the link.
/
lsd ds
PPP P= +
Where Pls is the Packet loss observed on the link, Pd is
the number of Packets dropped, Ps is the number of
Packets sent on the link.
Copyright © 2013 SciRes. CN
M. MANJUL ET AL.
7) Calculate the link utilization, αij.
( )
*max /
f
ijfif ij
bwXCfor allfЄF
α
= Σ
Where
*
()
f
f if
bw XΣ
denotes the value of total traffic
demand for all flows fЄF that are transmitted through
link (i, j), Cij is the link capacity.
8) Setting the threshold values:
a) If (99% of link utilization < αij)
Then, Congestion is very high and we adjust
/ *2
new ij
F
αβ
=
b) If (99% αij < 90%)
Then, Congestion is high and we adjust
*2 /2
new ij
F
β
α
=
If (Fnew > 99%)
Then go to stepa”.
c) If (90% αij 50%)
Then, Congestion is medium and we adjust
* *2
new ij
F
αβ
=
If (Fnew > 90%)
Then go to stepb”.
d) If (0 < αij < 49%)
Then, Congestion is low and we adjust
2
*log
new ij
F
αα
=
Else increment Fnew till its value reaches to medium
value.
We see the example of propo se algorithm given as be-
low:
Example: Suppose the Link Capacity is 100 Mbps,
Initial Sending Rate is 80 Mbps, Packet size is 300, RTT
is 150 ms, decreasing factor, β is 0.65 (0 < β < 1), in-
creasing factor α = S/RTT is 2 and we vary the Queue
size.
Case 1: Queue size = 50 packets
Link utilization,
( )
f
ijf*if ij
bwmax X/CfF
80*50/10040%.
Є
α
= Σ
= =
The link utilization is 40% means that congestion is
low. Then we use third condition and the proposed for-
mula is:
()( )
2
22
*
40logS/ RTT40log240%
new ij
F log
αα
=
=∗ =∗=
Again, the Fnew is 40%, then we go to step “c”.
* *240*0.65*25265%
new ij
F
αβ
=== =
So, our utilization comes to 65%.
Case 2: Queue size = 80 packets
Link utilization,
( )
f
ijfif ij
bwmax X/Cfor all fF
8080 /10064%.
Є
α
=Σ∗
=∗=
The link utilization is 64% means that congestion is
medium. Then we use second condition and the proposed
formula is:
264 0.65 283%
new ij
F
αβ
=∗∗= ∗∗=
So, our utilization comes to 83%.
Case 3: Queue size = 120 packets
Link utilization,
( )
f
ijf*ifij
bwmax X/Cfor all fF
80*120 /10096%.
Є
α
= Σ
= =
As utilization is 96% which show s the high conges tion
in the network is according to set thresholds. So, we have
made the congestion medium. For we use first condition
and the proposed formula is:
0.65
2/2962 25%
new ij
F
β
α
=∗= ∗∗=
We conclude that if our link utilization is high then we
need more care about the congestion, otherwise regularly
needs to increase the flow speed according to low and
medium include with medium and high factor of speed
respectively. Therefore, utilization of link is very impor-
tant phenomenon for control the congestion.
4. Result and Discussion
4.1. Simulation Topology
The simulation is done on NS-2 tool. We created a net-
work topology s cenario to simulate our work. Our aim is
to make a simulation environment for single source mul-
ticasting. The parameters used for the multicast topology
are given be low (Table 1).
For the multicast topology, Node 0 is a single source
and box1 is a router. Where node 2, 3, 4, 5 and 6 are the
group receivers through which nodes 7, 8, 9 are attached
to group 2, nodes 10, 11, 12 are attached to group 3, 13,
Table 1. Simulation Parameters.
Parameters Value
Link Bandwidth 10 - 100 Mbps
Link Delay
20 ms
Queue Size 50 - 99 packets
Sending Rate (initial)
80 Mbps
No. of Groups
5
No. of Receivers
15
RTT
150 ms
Packet Size
300
Session Time
500ms
ThLU (Congestion Status)
99% above (very High), 99% - 90% (High),
50% - 90% (Medium), 0% - 49% (Low)
Copyright © 2013 SciRes. CN
M. MANJUL ET AL.
652
14, 15 are attached to group 4, nodes 16, 17, 18 are at-
tached to group 5 and nodes 19, 20, 21 are attached to
group 6 as an individual receiver respectively. The net-
work topology for single multicast is below of 15 receiv-
ers:
The sender and router supports 1.5 Mb, node (n1-n2,
n1-n3, n1-n4, n1-n5 and n1-n6) also supports 1.5 Mb
with delay of 10 ms and nodes (n2-n7, n2-n8, n2-n9,
n3-n10, n3-n11, n3-n12, n4-n13, n4-n14, n4-n15, n5-n16,
n5-n17, n5-n18, n6-n19, n6-n20 and n6-n21) supports 1.0
Mb with a delay of 5 ms.
4.2. Result and Comparison
Figure 2 shows the comparison of throughput for differ-
ent values of Queue Size at β = 0.65. The variation is due
to the link utilization of the network. The new proposed
sending rate (Fnew) shows the better result than the exist-
ing approach (aij) because of the less packet loss ratio. At
150 Queue size the throughput of existing approach is
more than the proposed approach because according to
our condition more than 90% link utilization shows the
high congestion. So, we reduce this congestion to me-
dium i.e. less than 90%. That’s why here the result of
proposed approach is less than existing approach.
Figure 3 shows the variation of packet loss ratio with
the throughput. As the graph shows that the throughput
and packet loss are proportional to each other. Because
Figure 1. Multicast Topology.
Figure 2. Thr o ug hput Vs Q ueue size.
there is a less packet loss than the existing approach due
to congestion control based on link utilization while oth-
er is based on normal equation which is based on packet
size and packet loss.
Figure 4 shows the variation of Initial sending rate
with the throughput. The graph shows that sending rate
and throughput is dir ectly proportion al to each other. The
throughput increases with the increment in the β value.
Figure 5 Shows the variation of throughput with the
time. It shows the maximum time needed for the multi-
cast source till reaching a steady state throughput. It is
clear that proposed approach Outperforms existing ap-
proach.
Figure 3. Va riation of P acket Loss Ra tio with th e Throughput.
Figure 4. Variation of Sending Rate with Throughput.
Figure 5. Variation of Throughput with Time.
Copyright © 2013 SciRes. CN
M. MANJUL ET AL.
5. Conclusion
In this paper, we propose a multicast congestion control
mechanism to improve the utilization of the network. The
main part of our research is an algorithm for congestion
control which based on utilization of link and taking de-
cision according to high, medium and low congestion.
The proposed equations will be defined after setting the
threshold value. The equations are based on utilization to
improve the channel usage. This algorithm improves the
initial sending rate by giving the new sending rate for
proper utilization of network. The simulation of this
work is done using the Network Simulator tool (NS-2)
and we have analyzed the throughput of the Network
with respect to sending rate, queue size, packet loss ratio
and time. We have observed that proposed work provide
the better throughput of the network.
REFERENCES
[1] S. Kumar and K. Singh, “Efficient Single Rate Based
Multicast Congestion Control,” IJCA Proceedings on Na-
tional Conference on Innovative Paradigms in Engineer-
ing and Technology, 2012.
[2] S. Shi and M. Waldvogel, “A Rate-Based End-to-End
Multicast Congestion Control Protocol,International
Conference on Convergence and Hybrid Information
Technology, 2000.
[3] W. Kammoun and H. Youssef, “Improving End to End
Multicast Rate Control in Networks,” IEEE Proceedings
of Computer Communication ISCC, 2009.
[4] W. Kammoun and H. Youssef, “Improving the Perfor-
mance of End-to-End Single Rate Multicast Congestion
Control,” Proceedings of Computer and Communication,
6-9 July 2008.
[5] G.-I. Kwon and J. W. Byers, “Leveraging Single Rate
Schemes in Multiple Rate Multicast Congestion Control
Design,IEEE Journal on Selected Areas in Communica-
tions, Vol. 22, No. 10, 2004, pp. 1975-1986.
http://dx.doi.org/10.1109/JSAC.2004.836006
[6] M. G. abowski, M. Stasiak and P. Zwierzykowski, “Mul-
ticast Connections in a Virtual Circuit Switching Node,
Proceedings of the Advanced Industrial Conference on
Telecommunications/Service Assurance with Partial and
Intermittent Resources Conference/ELearning on Tele-
communications Workshop, 2005.
[7] R. Chakravarthi and C. Gomathy, “A Fuzzy Approach To
Detect And Control Congestion In Wireless Sensor Net-
works,Indian Journal of Computer Science and Engi-
neering (IJCSE), Vol. 3, No. 3, 2011, pp. 476-483.
[8] J. Bian, Z.-Q. Wang, H.-H. Ke and G.-Z. Zhang, “Con-
gestion Control Protocol Based on Delay Parameters for
Layered Multicast Communication,ICCT International
Conference on Communication Technology, 10 April
2007.
[9] H. Ma, X. Meng, L. Zhou, H. Li and X. Zhang, “Fuzzy-
Logic-Based Adaption Scheme for TFMCC,Proceed-
ings of International Conference on Networking and Dig-
ital Society, 2009.
[10] Y.-W. Leung, “Congestion Control for Multipoint Video
Conferencing,” IEEE Transactions on Circuits and Sys-
tems for Video Technology, Vol. 10, No. 5, 2000.
[11] X. Zhang, C. R. Li, X. Li and C. X. Bao, Multicast
Congestion Control on Many-to-Many Videoconferenc-
ing,Second International Conference on Future Gener-
ation Communication and Networking, 30 December
2008.
Copyright © 2013 SciRes. CN