Improved EPRCA Congestion Control Scheme for ATM Networks

Int. J. Advanced Networking and Applications
Volume: 6 Issue: 4 Pages: 2409-2413 (2015) ISSN: 0975-0290
2409
Improved EPRCA Congestion Control Scheme
for ATM Networks
Dr. M.Sreenivasulu
Department of CSE, K.S.R.M.College of Engineering, Kadapa, Andhra Pradesh, India.
Email: mesrinu@rediffmail.com
-------------------------------------------------------------------ABSTRACT---------------------------------------------------------------
Traffic management and congestion control are major issues in Asynchronous Transfer Mode(ATM) networks.
Congestion arises when traffic in the network is more than offered load. The primary function of congestion
control is to ensure good throughput and low delay performance while maintaining a fair allocation of network
resources to users. In this paper, Enhanced Proportional Rate based Congestion Avoidance (EPRCA) scheme
proposed by ATM forum has been considered. But this scheme has limitation of higher cell drop problem for the
bursty traffic. Improvements to EPRCA scheme have been proposed to reduce cell drop problem and results of
improved EPRCA schemes were analyzed with basic EPRCA scheme.
Keywords: ATM networks, EPRCA, IEPRCA, NIST ATM Network Simulator.
--------------------------------------------------------------------------------------------------------------------------------------------------
Date of Submission: November 27, 2014 Date of Acceptance: January 7, 2015
---------------------------------------------------------------------------------------------------------------------------------------------------
1. Introduction
Broadband Integrated Services Digital Network (B-
ISDN) can offer high speed data transport, multimedia
communication, high quality video conferences, video
telephony, etc [17]. The Asynchronous Transfer Mode
(ATM) has emerged as a standard technology for
supporting gigabit BISDN services [18-20]. Using ATM,
the information is usually transmitted in fixed size units
called cells. The ATM technique is designed to provide
fast packet switching at variable rates from 1.54 Mbps to
2 Gbps and even beyond.
Traffic Management and congestion control are major
issues in ATM networks [1-6]. Congestion arises when
incoming traffic to specific link is more than outgoing
link capacity. When congestion problem arises, cells are
discarded without being processed. The design of
congestion control mechanism is essential not only for
regulating traffic to prevent congestion but also for
providing efficient and fair bandwidth allocation [9,15].
Among various congestion control mechanisms, rate base
mechanisms are proved efficient than other mechanisms.
Rate based congestion control schemes are end-to-end
feedback mechanisms, and the cell transmission rate of
each source end system is regulated according to
congestion feedback information returned by the network
[8,12,13].
In this paper, Enhanced Proportional Rate Control
Algorithm (EPRCA), which is a rate based scheme for
controlling congestion in ATM networks, has been
considered. The improvements to EPRCA scheme is also
presented in this paper. The basic and improved schemes
were implemented using NIST ATM network simulator.
This paper is organized as follows: The importance of
Resource Management (RM) cell and its fields are
presented in Section 2. The basic EPRCA mechanism, the
simulation environment and results analysis of EPRCA
are presented in section 3. The improvements to EPRCA
schemes, with results are presented in section 4. Finally
the conclusion of paper is presented in section 5.
2. Resource Management (RM) Cell
In rate based congestion control schemes, source end
system periodically sends a Resource Management (RM)
cell for every 31 data cells. The RM cell is used to carry
control and flow information over the connection between
source and destination end systems [11]. The destination
end system sends an RM cell with an indicator showing
the status of traffic back to the source end system. The
intermediate switches may simply forward these RM cells
or specify the explicit rate in RM cell based on
functionality of different rate based congestion control
schemes. The source end system then uses the
information in the RM cell for specifying the transmission
rate until a new RM cell is received.
The RM cell is used for finding status of congestion, if
any, in the network. Every RM cell has the header of five
bytes width. The payload type indicator (PTI) in cell
header is set to a 3 bit binary, 110, as the default value, to
indicate a RM cell. The protocol ID field, which is one
byte long, is set to one for Available Bit Rate (ABR)
application connections. The direction (DIR) bit
distinguishes between forward and backward RM cells.
The backward notification (BN) bit is set only in switch
generated RM cells. The congestion indication (CI) bit is
set by a switch to indicate presence of congestion. The no
increase (NI) bit, is set by a switch to indicate moderate
congestion in the network. The sequence number field
value is set by source end system to indicate the current
sequence number for the RM cell and the queue length
field value is set by a switch in the network. The Explicit
Rate (ER) field indicates the maximum rate allowed to the
source and this field value is set by the switch. When the
source receives the backward RM cell from the network,

2410
it adjusts its transmission rate using the ER, CI, and NI
values in the received RM Cell.
3. Basic EPRCA Scheme
The Enhanced Proportional Rate Control Algorithm
(EPRCA) is a rate based scheme, in which switch
explicitly specifies the rate at which transmission should
be made at the source [14,16]. In this scheme, each
switch in the path between source and destination end
systems maintains a queue or buffer, with two congestion
thresholds namely, thresholds QT (i.e. to indicate normal
congestion) and DQT (i.e. to indicate severe congestion).
Each switch in the network periodically computes mean
allowed cell rate (MACR) that should ideally be the mean
of all source end systems ACRs. The MACR value is
used for computing explicit rate in RM cell.
3.1. EPRCA Algorithm
Before transmission of data by the source end system, the
connection or route to be established between source and
destination end systems. The route comprising of source,
destination end systems, intermediate systems, and
switches. The behaviour of source, destination end
systems, and intermediate switch are mentioned below
[10]:
Source End System behaviour
1. At the beginning of transmission, the source transmits
an RM cell. The allowed cell rate (ACR) or current cell
rate (CCR) is initially set to peak cell rate.
2. The source sends data cells and after every 31 data
cells transmitted, one RM cell is followed.
3. The source continuously decreases its ACR until it
receives an RM cell returned from the destination. Upon
receipt of every RM cell, the source increases its ACR by
no greater than ER.
Destination End System behaviour
1. The destination returns an RM cell upon receiving
each RM cell.
Switch behaviour
1. In the network, each switch periodically computes a
mean allowed cell rate (MACR) using the following
formula:
MACR = (1 – AV) * MACR + AV * ACR,
where AV is the Average Factor, which is set to a default
value of 0.0625.
If the queue length is less than QT, then there is no
congestion in the network. When non-congested switch
receives a RM cell from a source end system and if ACR
is less than MACR, then there is no need to adjust its
MACR. If ACR is greater than or equal to MACR, it
means that MACR is not adequately large enough, then
MACR needs to be increased by using the above formula.
2. If the queue length exceeds DQT, indicating that the
switch is said to be in severe congestion state, then ER
needs to be replaced in RM suitably, as follows:
ER = min (ER, MACR * MRF)
Where MRF is Major Reduction Factor, set to a default
value of 0.95
If queue length exceeds QT, the switch is said to be in
congestion state, then only those source end systems
whose ACR is more than MACR is need to reduce their
ACR value by replacing ER in RM cell given by the
formula:
ER = min (ER, MACR*ERF)
Where ERF is Explicit Reduction Factor, set to a default
value of 0.9375.
3.2 Simulation Environment
The basic EPRCA scheme described in previous section
was implemented using NIST ATM Network Simulator
for a sample network configuration shown in fig 1 [7].
Fig.1: A Sample Network Configuration
The network shown in fig.1, consist of four source end
systems (BTE1 to BTE 4 – Broadband Terminal
Equipment), four destination end systems (BTE5 to
BTE8), nine communication links (link1 to link 9), two
switches (switch1 and switch2), four applications
(Available Bit Rate applications ABR1 to ABR4) running
on source end systems and another four applications
(ABR5 to ABR8) running on destination end systems.
The ABR applications are attached to source and
destination terminals for sending and receiving data
respectively. For example, in given network
configuration, four virtual channels have been considered
for explanation. The first Virtual Channel (VC1) can be
established is through BTE1, Switch 1, Switch 2, and BTE
5. The Second Virtual Channel (VC2) is through BTE 2,
Switch 1, Switch 2, and BTE 6. The Third Virtual Channel
(VC3) is through BTE 3, Switch 1, Switch 2, and BTE 7.
The Fourth Virtual Channel (VC4) is through BTE 4,
Switch 1, Switch 2, and BTE 8.
In our simulation, the source end systems transmit cells at
its given Allowed Cell Rate which is initially set to PCR
with default value of 149 Mbps. The bandwidths of all
links (i.e link 1 to link 9) were set to 155 Mbps. The
default values for various control parameters of switches
and source end systems are also set, as mentioned in Table
1 and Table 2 respectively.

2411
Table 1: Control parameters for Switch1 & Switch 2
for EPRCA scheme
Parameter Value
Queue length (in cells)
Low threshold (in cells)
Very Congested Queue Threshold (in cells)
Explicit Reduction Factor (ERF)
Major Reduction Factor (MRF)
Average Factor (AV)
1000
600
900
0.9375
0.95
0.0625
Table 2: Control parameters of source end systems for
EPRCA scheme
Parameter Value
Initial Cell Rate (ICR)
Minimum Cell Rate (MCR)
Peak Cell Rate (PCR)
7.49 Mbps
1. 49 Mbps
149.76 Mbps
3.3 Results Analysis
The performance of EPRCA Scheme is measured based on
cell drop percentage during transmission period of data
cells. The graph showing cell drop percentage over
transmission time is shown in figure 2.
Fig: 2 Percentage of Cell drop Vs Time at Switch 1
for basic EPRCA Scheme
From the above graph, higher percentage of cell drop is
noticed at the beginning of transmission, because ACR
was set to PCR. The cell drop at switch 1 is approximately
33 % and need a mechanism to reduce.
4. Improved EPRCA Schemes
Improvements to EPRCA scheme have been proposed to
reduce the cell drop problem by considering the
following modifications to basic EPRCA scheme.
4.1. Improved EPRCA (IEPRCA) Scheme1
In this, the behavior of source end systems are modified
by changing value of PCR to get reduced cell drops as
well as optimal link utilization. In basic EPRCA scheme,
a default value of 149.7 Mbps is set to PCR. This PCR
value is assigned to Initial Cell Rate (ICR) and all source
end systems start transmitting cells at higher rates (i.e.
with PCR) initially, resulting high cell drops at switch1.
The experimental tests have been conducted by varying
values for PCR and able to notice that the maximum
network efficiency can be reached by selecting PCR
value at an optimal value equal to 100 Mbps.
The improved EPRCA with optimal PCR value (i.e.
modified source end system behavior) was implemented
and graph for percentage of cell drop at switch1 over
transmission time is Fig. 3.
Fig.3: Percentage of Cell drop Vs Time at Switch1 for
IEPRCA sheme1
From the graph, it has been noticed that, the IEPRCA
scheme1 gives 21% cell drop, where as in the case of
basic EPRCA scheme it is approximately 33%. So the
IEPRCA scheme1 effectively reduces cell drop problem
by selecting PCR value at optimum and hence better than
basic EPRCA scheme.
4.2 Improved EPRCA (IEPRCA) Scheme2
The modification of switch behavior is also considered to
improve the performance of EPRCA scheme. The
modified switch behavior is given below.
1. When non congested switch receives forward RM cell
from source end system, it computes mean ACR
(MACR) using the following formula:
MACR = (1 – AV) * MACR + AV * CCR
Time in microseconds
%ofCelldrop
a
%ofCelldrop
a

2412
The mean ACR is used for setting ER field of the RM
cell and CCR is the Current Cell Rate.
2. When switch receives backward RM cell from the
destination end system, it checks, if there is congestion at
the switch. It compares the queue length with threshold
value to detect congestion.
(i) If queue length exceeds threshold DQT, the switch
is said to be in severe congestion state. Then it replaces
ER in RM cell by min (ER, MACR * MRF) and
reduced ER communicated to all source end systems
via backward RM cell.
(ii) If queue length exceeds QT, the switch is said to be
in congestion state. To relieve congestion in network,
the switch reduces ER by replacing ER with min(ER,
MACR * ERF) and this reduced rate is communicated
by sending backward RM cell to all source end
systems.
The IEPRCA scheme2 was implemented with modified
switch behavior and graph showing percentage of cell
drop at switch1 over transmission time is shown in Fig.
3.
From this graph, it can be observed that switch reduces
explicit rate in backward RM cell, whenever it finds the
queue length exceeds threshold value.
The source end system decreases allowed cell rate when it
receives reduced ER in backward RM cell. When the
switch1 gets relief from congestion, it increases ER field
value in backward RM cell to allow source end systems to
increase their allowed cell rate.
It can also easily noticed from the graph that, % cell drop
fairly depends upon the cell transmission rate, and
increases with the rate of transmission The % cell drop
with IEPRCA scheme2 is 27%, as against 33% with basic
EPRCA scheme. Hence IEPRCA scheme2 is proved to be
better than basic EPRCA scheme.
IEPRCA scheme2
4.3 Improved EPRCA (IEPRCA) Scheme3
To reduce the cell drop still further, the effects of
method 1 and method 2 have been combined to achieve
yet another improved scheme, called, IEPRCA scheme3.
In this scheme3, the optimal value of 100 Mbps is set to
PCR for source end system and also switch behavior is
modified as in IEPRCA scheme2. The IEPRCA scheme3
was implemented and results are shown in form of graph
for % cell drop vs time at intermediate switch1 in shown
in Fig 5.
IEPRCA scheme3
From this graph, % cell drop is 17% noticed at switch1,
which is considerably less than that of basic EPRCA
scheme as well as IEPRCA scheme1 and scheme2.
Hence it can be concluded that IEPRCA scheme3 is far
better than basic EPRCA scheme.
4.4 Performance Comparison of EPRCA and improved
EPRCA schemes
The performance of IEPRCA schemes and basic EPRCA
can be compared by considering the amount of % cell
drop at intermediate switch1 and it was tabulated in table
3.
Table 3: Percentage of cell drop for EPRCA and
IEPRCA schemes
Sl.
No.
Scheme % cell drop
1. EPRCA 33
2. IEPRCA Scheme1 21
From the above the table, it has been noticed that,
IEPRCA shceme3 results with 17% cell drop which is
considerably less than that of basic EPRCA scheme as
well as IEPRCA scheme1 and scheme2. Hence it can be
%ofCelldrop
a
%ofCelldrop
a

2413
concluded that IEPRCA scheme3 is far better than the
basic IEPRCA scheme.
5. Conclusion
In this paper, 3 methods have been proposed to enhance the
performance of basic EPRCA scheme, (i) by setting PCR to
optimal value (IEPRCA scheme1), (ii) by modifying the
switch behavior (IEPRCA scheme2), and (iii) by
setting optimal PCR value and modified switch
behavior (IEPRCA scheme3). The IEPRCA scheme3 found
to achieve the best results when compared to basic EPRCA
scheme as well as over other improved schemes IEPRCA
scheme1 and IEPRCA scheme2.
References
[1] R.S.Deshpande, Dr. P.D. Vyavahare, “Recent
Advances and a survey of congestion control
mechanisms in ATM networks”, IE(I) Journal, Vol.
88, pp. 47-54, , 2007 .
[2] W Li, Z Che, Y Li, “Research on the congestion
control of Broadband Integrated Service Digital
Network based on ATM”, Proceedings of the fifth
international conference on Machine Learning and
Cybernetics, Daliaan, pp 2510-2512, 2006.
[3] E Al-Hammadi and M M Shasavari, “Engineering
ATM networks for congestion avoidance”, Mobile
Networks and Application, Vol. 5, pp.157 – 163,
2000.
[4] Ching-Fong Su, Gustavo Do Vacianna, Jean
Warland, “Explicit Rate Flow Control for ABR
Services in ATM Networks”, IEEE/ACM
Transactions on Networking, Vol. 8, Issue 3, pp. 350-
361, 2000.
[5] A Hac, H.Lin, “Congestion Control for ABR traffic
in an ATM network”, International Journal of
Network Management, Vol. 9, pp. 249-264, 1999.
[6] K Siu, H Tzeng, “Intelligent Congestion Control for
ABR service in ATM networks”,ACM SIGCOMM
Computer Communication Review, Vol. 25, Issue 2,
pp. 81-106, 1994.
[7] N Golmie, F Mouveaux, L Hester, Y Saintillan, A
Koenig, D.Su, “The NIST ATM/HFC Network
Simulator Operation and Programming Guide”, Dec
1998.
[8] H Ohsaki, M Murata, H Suzuki et al, “Rate based
Congestion Control for ATM networks”, ACM
SIGCOMM, Computer Communication Review, Vol.
25, Issue 2, pp. 60-72, 1995.
[9] D Gaiti, G. Pujolle, “Performance management issues
in ATM networks: traffic and congestion control”,
IEEE/ACM transactions on networking, Vol. 4, No.2,
pp.249-257, April 1996.
[10] Raj Jain, Shivkumar Kalyanraman, Sonia Fahmy,
Rohit Goyal, “Source Behavior for ATM ABR
Traffic Management: An Explanation”, IEEE
Communication Magzine, pp. 50-55, 1996.
[11] Thomos M.Chen, Steve S. Lin, David Wang, Vijay
K.Samalam, Michael J. Procanik, and Dinyar
Kavouspour, “Monitoring and control of ATM
networks using special cells”, IEEE network
magazine, pp.28-38, Sept/oct 1996.
[12] Anna Hac, Yingjun Ma, “A Rate based Congestion
Control Scheme for ABR service in ATM networks”,
International Journal of Network Management,
Vol.8, pp. 292-317, 1998.
[13] Lalita A. Kulakarni, San-qi Li, “Performance
Analysis of a Rate-based Feedback Control
Scheme”, IEEE/ACM Transactions on Networking,
Vol. 6, No. 6, Decemeber 1998.
[14] L Roberts, “Enhanced Proportional Rate Control
Algorithm”, ATM forum contribution, April 1994.
[15] Rama Krishnan, “Rate based Control Schemes for
ABR Traffic – Design Principles and performance
comparison”,Global Communication Conference
Globecom96, pp.1231-1235, 1996.
[16] Xinomel Yu, Doan B. Hoang, David D.Feng, “A
Simulation Study of using ER feedback control to
transport compressed video over ATM network”,
Proceedings Workshop on Visualization, 2001.
[17] William Stallings, “ISDN and Broadband ISDN
with Frame Relay and ATM”, (Fouth Edition,
Pearson Education Asia, 2007).
[18] A Arulambalam, Xiaoqiang Chen, N. Ansari,
“Allocating Fair Rates for Available Bit Rate Service
in ATM Networks”, IEEE Communication
Magazine, Vol. 34, No. 11, pp. 92-100, 1996.
[19] Rainer Handel,Manfred N Huber,Stefan Schroder,
”ATM Networks Concepts, Protocols,Applications’
(Third Edition, Addison,Wesly,1999).
[20] J Y. Le Boudec, “The Asynchronous Transfer
Mode: A Tutorial”, Computer Networks & ISDN
Systems, Vol.24, pp. 279-302, 1992.
Author’s Biography
M. Sreenivasunlu received B.Tech
degree in Computer Science and
Engineering, from K.L. College of
Engineering (K.L.University),
Vaddeswarm, A.P. India, in 1990. He completed
M.E. in Computer Science & Engineering, from
Jadavpur University, Calcutta, India, in 1999. He
received Ph.D. in Computer Science and
Engineering, from J.N.T.University, Anantapur, A.P.
India, in the year 2013. He is having 21 years of
teaching experience. He is currently working as
Professor & Head of Computer Science and
Engineering Department, K.S.R.M.College of
Engineering, Kadapa. He is member of IEEE, life
member of IE(I), and ISTE. He published 15 papers
in international journals and in the proceedings of
National and International Conferences.

Improved EPRCA Congestion Control Scheme for ATM Networks

More Related Content

What's hot (20)

Viewers also liked (13)

Similar to Improved EPRCA Congestion Control Scheme for ATM Networks (20)

More from Eswar Publications (20)

Recently uploaded (20)

Improved EPRCA Congestion Control Scheme for ATM Networks