ArticlePDF Available

Abstract and Figures

Maintaining the quality of service (QOS) and controlling the network congestion are quite complicated tasks. They cause degrading the performance of the network, and disturbing the continuous communication process. To overcome these issues, one step towards this dilemma has been taken in form of Pre-congestion notification (PCN) technique. PCN uses a packet marking technique within a PCN domain over IP networks. It is notified by egress node that works as guard at entry point of network. Egress node gives feedback to communicating servers whether rate on the link is exceeded than configured admissible threshold or within the limit. Based on this feedback, admission decisions are taken to determine whether to allow/block new coming flows or terminate already accepted. The actual question is about selection of right algorithm for PCN domain. In this paper, we investigate the analytical behavior of some known PCN algorithms. We make slide modifications in originality of PCN algorithms without disquieting working process in order to employ those within similar types of scenarios. Our goal is to simulate them either in highly congested or less congested realistic scenarios. On the basis of simulation done in ns2, we are able to recommend each PCN algorithm for specific conditions. Finally, we develop a benchmark that helps researchers and scientific communities to pick the right algorithm. Furthermore, the benchmark is designed to achieve specific objectives according to the users' requirements without congesting the network.
Content may be subject to copyright.
ANALYTICAL STUDY OF PRE-CONGESTION
NOTIFICATION (PCN) TECHNIQUES
Marwah Almasri
1
, Khaled Elleithy
2
, and Abdul Razaque
3
1
Department of Computer Science and Engineering, University of Bridgeport,
Bridgeport, CT 06604
maalmasr@bridgeport.edu
2
Department of Computer Science and Engineering, University of Bridgeport,
Bridgeport, CT 06604
elleithy@bridgeport.edu
3
Department of Computer Science and Engineering, University of Bridgeport,
Bridgeport, CT 06604
arazaque@bridgeport.edu
ABSTRACT
Maintaining the quality of service (QOS) and controlling the network congestion are quite complicated
tasks. They cause degrading the performance of the network, and disturbing the continuous
communication process. To overcome these issues, one step towards this dilemma has been taken in form
of Pre-congestion notification (PCN) technique. PCN uses a packet marking technique within a PCN
domain over IP networks. It is notified by egress node that works as guard at entry point of network.
Egress node gives feedback to communicating servers whether rate on the link is exceeded than
configured admissible threshold or within the limit. Based on this feedback, admission decisions are
taken to determine whether to allow/block new coming flows or terminate already accepted. The actual
question is about selection of right algorithm for PCN domain. In this paper, we investigate the analytical
behavior of some known PCN algorithms. We make slide modifications in originality of PCN algorithms
without disquieting working process in order to employ those within similar types of scenarios. Our goal
is to simulate them either in highly congested or less congested realistic scenarios. On the basis of
simulation done in ns2, we are able to recommend each PCN algorithm for specific conditions. Finally,
we develop a benchmark that helps researchers and scientific communities to pick the right algorithm.
Furthermore, the benchmark is designed to achieve specific objectives according to the users’
requirements without congesting the network.
KEYWORDS
Pre-congestion notification (PCN) technique, Random Early Detection (RED), Explicit Congestion
Notification (ECN), Token bucket (TB), Bandwidth Metering (BM), Additional Buffer Technique (AB).
1. INTRODUCTION
With the revolution of technology, the numerous users of the Internet face many challenging
issues that highly affect the quality of service (QoS). However, the Internet Engineering Task
Force (IETF) has come up with the idea of pre-congestion notification (PCN) in order to avoid
congestion of highly loaded network to assure the quality of service (QoS) within a Diffserv
domain [1]. Excessive network load causes packet loss in the network. Furthermore, PCN
maximizes the use of recourses over the link. PCN has three types of nodes, which are ingress,
interior, and egress nodes. To avoid congestion, PCN uses admission control mechanism (AC)
to limit the number of flows, and flow termination (FT) mechanism to remove some already
accepted flows. The network’s load is measured inside the PCN domain and packets are marked
according to the load condition [6].
Many algorithms have been introduced to avoid the congestion and to measure the network’s
load. Active queue management mechanisms have positive impacts on the performance of the
Internet [2]. Random early detection (RED) algorithm is one of the active queue management
mechanisms to avoid congestion [2]. It is deployed in Internet [2]. RED decreases the number of
dropped packets in routers. Also, it reduces the delay especially in interactive services by
providing smaller average queue size. Another benefit of using this kind of algorithm is to
ensure availability of the buffer for all arriving packets to control the lock-out behavior [2].
Explicit congestion notification (ECN) mechanism is another approach to avoid unnecessary
dropped packets and delay in low-bandwidth TCP connections [13]. Token bucket (TB) and
bandwidth metering (BM) algorithms measure the network load and signal the PCN egress node
in case of congestion [14]. These both later algorithms are varied in their algorithmic
complexities. Token bucket algorithm is basically a bit counter, which is updated only when a
packet arrives. As a result, token bucket algorithm has a limited complexity; whereas the
bandwidth metering algorithm needs more memory requirements to store the arriving packets.
Bandwidth metering (BM) algorithm is helpful especially in highly congested networks [14].
The objective of this paper is to determine the weakness and strength of five algorithms
supporting to PCN domain.
The paper is structured as follows. Section 2 discusses the related work regarding the PCN and
the various techniques used for measuring the network load. Section 3 shows an overview of the
PCN architecture and summaries existing well known techniques. Section 4 describes the
simulation setup. Section 5 provides analysis of simulation results and section 6 discusses the
results. Finally, section 7 concludes the paper on the basis of findings.
2. RELATED WORK
In this section, we discuss some salient features of PCN techniques. New bandwidth
measurement technique is introduced in [3]. The paper discusses about based admission control
algorithm and the performance of PCN on VBR video services. Also, it studies and analyzes the
benefit of using PCN mechanism. An additional buffering technique is proposed and
implemented using NS-2 based simulator. Authors validate the findings on the basis of
simulation. 26.5% network utilization has been increased through this technique [4]. Different
admission control (AC) methods in PCN have been investigated [5]. Authors discuss over
admission flows in PCN based on excess traffic. The paper discusses that marking is occurred
owing to weak pre-congestion signals. It also studies the performance of probe-based AC
(PBAC) and congestion-level estimate based AC (CLEBAC) through simulation and
mathematical modeling to deploy the results. Furthermore, it is observed that it is more
influential in challenging conditions such as on/off traffic, low traffic aggregation, and delayed
media [5].
The paper [6] discusses the encoding in IPv4 header through PCN marking mechanism. It
concludes the difference between existing approaches, and suggests the enhancement of PCN
design. Many algorithms have been proposed to configure the PCN threshold rate in single and
dual marking PCN domain [7]. Authors suggest more requirements to incorporate in single
marking than dual marking. The paper discusses that dual marking has higher resource
efficiency as compared with a single marking in PCN architecture [7]. Two-layer architecture
that uses various types of algorithms is introduced in [8]. The paper [15] introduces a new
congestion control algorithm called modified forward active network congestion control
algorithm (MFACC), which uses RED algorithm to control the queue length of the router and
avoid packets loss. It also studies the active detection and the passive indication mechanism. It
concludes on basis of simulation that MFACC algorithm resolves many problems found in
previous used techniques, enhances the quality of service (QoS), and reduces the delay and loss
packets rate [15]. RED algorithm is used with responsive and non-responsive flows [16]. The
goal of this paper is to study the network performance and efficiency using NS-2 simulator. The
simulations results prove that RED algorithm is working effectively even with non-responsive
flows. It is also proven that RED reduces the lock out phenomenon and delay that causes of
increasing the throughput [16].
We do a comprehensive analytical study of five existing algorithms using NS-2 simulator and
plot the strengths and weaknesses of each technique. These techniques are: random early
detection (RED), explicit congestion notification (ECN), token bucket (TB), bandwidth
metering (BM), and an additional buffer technique (AB). On basis of findings, we make
benchmark that helps to understand the depth of each technique in PCN domain. To validate
these strengths and weaknesses of these algorithms, the realistic scenarios have been built to
measure the behavior of each that supports to make the benchmark. Finally, we recommend
each algorithm in specific conditions to achieve more targets..
3. AN OVERVIEW OF PCN ARCHITECTURE AND EXISTING
WELL KNOWN TECHNIQUE
In this section, we demonstrate the design of PCN and present many used techniques inside the
PCN domain that have tremendous impact on avoiding the congestion in the network.
Pre-congestion notification (PCN) has gained a lot of attention especially for the needs of
emerging technology. The PCN uses two main mechanisms, which are admission control (AC)
and flow termination (FT). The admission control is used to determine whether to accept or
block new flows. On the other hand, the flow termination is used to terminate some of the
already accepted flows [1]. Within the PCN domain, there are three types of nodes as illustrated
in Figure 1: ingress node, interior node, and egress node. The ingress node and the egress node
are located at the boundaries of the PCN network [3]. The interior node is responsible for
measuring the congestion level inside the PCN domain by calculating the congestion level
estimator, (CLE) that uses an exponential weighted moving average (EWMA). The value of
(CLE) falls between one and zero. After computing the (CLE) value, the egress node sends
signal to the ingress node in order to decide whether to accept or block new flows [4]. When the
CLE is 1, there is a pre-congestion while 0 means there is no congestion. CLE can be calculated
as follows:
CLE
n
= Thr * (1 - CLE
W
) + CLE
W
* CLE
n-1
(1)
CLE
W
donates the CLE weight. Thr is a threshold value that can be either 1 or 0 depending on
the packet marking status [3].
Figure 1: Describes the PCN architecture which consists of three types of nodes: ingress, egress,
and interior nodes.
In addition, PCN has defined two rate thresholds, an admissible and a supportable rate threshold
(Ar), (Sr) which provide three types of pre-congestion. Figure 2, summaries these types as
follows: when the PCN traffic rate r < Ar, there is no congestion in the network. Hence, new
flows are accepted. In contrast, when the PCN traffic rate r > Ar, the link L’ in the network is
Ar-pre-congested and the overload value is said to be Ar-overload. As a result, new flows are
not accepted. Another issue when the PCN traffic rate r > Sr, the link L’ in the network is Sr-
pre-congested and the overload value is said to be Sr-overload. Consequently, we need to reduce
the value of rate r(L) by terminating already accepted flows [8].
Figure 2: Describes the three types of pre-congestion.
3.1 Random Early Detection (RED)
Random early detection (RED) algorithm is used to avoid the congestion in the network
especially in the PCN mechanism. It detects early congestion in order to avoid the congestion
and to enhance the TCP throughput performance. RED algorithm’s idea is basically based on the
buffering queue length [7]. It computes an average queue size (avg) by an exponential weighted
moving average. Then it compares this average (avg) with two other parameters which are a
minimum threshold (min_thr) and a maximum threshold (max_thr). If average queue size (avg)
falls below the minimum threshold (min_thr), then no packets are marked or dropped. On the
other hand, if average queue size (avg) goes above the maximum threshold (max_thr), then the
packets are marked. Another issue is when the average queue size (avg) is between the minimum
threshold (min_thr) and the maximum threshold (max_thr) values, the packets are marked
relatively to a probability P
A
[9]. The general model of RED algorithm is described in Algorithm
1. In addition, the packet-marking probability P
p
increases linearly from 0 to Max
p
along with the
average queue size avg as follows:
P
p=
Max
p
(avg−min_thr) / (max_thr − min_thr) (2)
The final packet-marking probability P
A
increases along with the increment of the counter since
the last marked packet [9] as follows:
P
A
= P
p
/(1−count * P
p
) (3)
The mechanism of this algorithm comprises of two main parts. One is to compute the average
queue size (avg), and the other to calculate the probability P
A
that the packets are marked with
[9].
Algorithm. 1: RED algorithm
3.2 Explicit Congestion Notification (ECN)
Explicit congestion notification (ECN) is another technique that is used in PCN model in order
to avoid congestion. It is based on random early detection (RED) algorithm.
It signals the incipient congestion to notify the TCP sender to reduce the sending rate window
[10]. ECN protocol uses congestion experienced (CE), that is the code point, located at the
packet header. This is useful to indicate that there is congestion rather than just dropping the
packets. As a result, using ECN in TCP/IP networks has the benefit of not dropping or delaying
the packets. In addition, the TCP and IP header need to be modified to hold extra bits for ECN
protocol as described in Figure 3, for the IP header, and Figure 4, for the TCP header.
If the value of the code point is “10”, or “01”, the packets are ECN capable. If the value of the
code point is “00”, the packets are not ECN capable. However, if the value of the code point is
“11” which is the CE code point, meaning that there is congestion and the packets are marked.
Hence, when receiving the marked packet with CE code point, TCP connection should reduce
its sending rate of packets [11].
Figure 3: ECN in IP header.
Figure 4: ECN in TCP header.
For each arriving packet
Compute the average queue size avg
If min_thr ≤ avg ≤ max_thr
{
Compute probability P
A
Mark packet according to P
A
}
Else if max_thr ≤ avg
{
Mark the arriving packet
}
3.3 Token Bucket (TB)
Token bucket is a measurement algorithm that is used at the interior node in the PCN domain.
It’s primarily goal is to limit the speed of network transmission. On the other hand, it is also
helpful to determine the bandwidth usage. The token bucket is a bit counter that shows the load
in the network [12], as given in Figure 5. These tokens are added to the bucket at a constant
token rate (R). However, when the packets are reached at interior nodes in the PCN domain,
tokens are removed from the bucket. If the aggregate bandwidth is lower than (R), the number
of tokens increases. On the other hand, if the aggregate bandwidth is greater than (R), the
number of tokens decreases. As a result, in the PCN architecture, packets are marked when the
number of tokens is fewer than the token bucket threshold [3]. Algorithm 2, shows the
algorithm for token bucket mechanism.
Figure 5: Token bucket mechanism.
Algorithm 2: Token bucket algorithm.
1. Bucket= empty, Token =
0;
2. Max_rate = maximum
output rate;
3. BC = capacity of token
bucket;
4. L = burst length;
5. G = generating rate;
6. Max_rate = BC / L + G;
7. Repeat
If (token < BC)
8. accept packets
9. If (token > 0)
10. G= dequeue (bucket,
Max_rate)
11. Display Max_rate output
12. If (t == Δt)
13. token++
End Repeat
End If
End If
End If
3.4 Bandwidth Metering (BM)
This mechanism is another way to measure the load in the network. It differs from the token
bucket because it uses a time window technique. This algorithm marks packets when the
aggregate bandwidth is greater than a predefined threshold. This algorithm is better than the
token bucket because it computes the accurate measurement of the bandwidth instead of just
comparing it with the token rate (R). In contrast, the bandwidth metering requires more memory
but its advantage outweigh this extra requirement. In addition, the base of this technique is on a
sliding window with a fixed “mi”, which is a measurement interval. During the last “mi”, the
bandwidth is measured as it receives packets and then the packets are marked if the bandwidth
is higher than a bandwidth threshold [12]. Algorithm. 3, demonstrates the bandwidth metering
algorithm.
Algorithm. 3:Bandwidth metering algorithm.
3.5 An Additional Buffer Technique (AB)
In this technique, an additional buffer is used in the PCN domain as shown in Figure 6. The
algorithm for this technique can be implemented as follows. The threshold rate (Tr) is calculated
by (4) where (Ar) is the admissible rate threshold, and (Or) is the objective rate:
Tr = Ar + Or /2 (4)
Figure 6: Describes the additional buffer technique.
1. Input: (B = bandwidth
measurement, B_Thr = bandwidth
threshold, mi = fixed measurement
interval)
2. Output: (m_packet)
3. For each packet at arrival time t
4. Compute B during the last mi
seconds.
5. If B (t) > B_Thr (t)
6. Process m_packet
7. End
The computed threshold rate (Tr) is important to construct the buffer [4]. All packets enter the
PCN domain through the ingress node which works as the decision maker. It classifies the
packets into accepted packets or dropped/marked packets and sends them to the buffer.
However, if the buffer is full, incoming packets are dropped until the buffer gets some space for
accepting new packets. After that, all packets including the dropped packets are sent to the
scheduler. The scheduler’s function is send these packets to the egress node with a priority.
Accepted packets get higher priority while dropped/marked packets get lower priority. After
transmitted the packets to the egress node, the packets are forwarded to the transport layer. This
priority depends on the weighted value of the dropped packets (Wd) and the buffer (Wb). These
two weighted values can be calculated by using (5) and (6).
Wd = 1-(Tr/Or) Wd [0, 1] (5)
Wb = Tr/Or Wb [0, 1] (6)
4. SIMULATION SETUP
Figure 7: The simulation scenario.
Our objective is to simulate all the techniques described in section 3 based on realistic scenarios
in simple and highly congested network. We make PCN domain highly congested to examine
the performance of network. We assume in our simulation scenario that there are eight
educational institutions located at two different states as shown in Figure 7. The objective of
this scenario is to maintain the quality of service and provide better data communication among
these educational institutions. On the basis of the scenario, we use parameters that help to
examine the behavior of these well know techniques with different network bandwidth. These
scenarios are simulated using CBR, FTP, and HTTP applications and supported with UDP and
TCP layer protocols. The simulation area is 500X500 m2 and the number of mobile nodes is 50
nodes. The sensing range of the node is 250 m2. The size of packets is 1040 bytes including 40
bytes header. In our scenario, the PCN ingress node is connected via 5 links. The size of
bandwidth is 300 Mbps, 400 Mbps, and 500 Mbps used in the PCN domain with these
techniques. The capacity of each link is equal. If the bandwidth of the network is 300 Mbps that
each link gets equal share 60 Mbps. Similarly, 400 Mbps and 500 Mbps bandwidths are divided
in equal five shares. The 10 connections are established at a time and the packet generation rate
is 15 packets per second. In our simulation experiment, we set 4 seconds pause time after
interval of 5 minutes of simulation.
5. ANALYSIS OF RESULTS
5.1 Throughput
We use TCP as transport protocol for sending and receiving the data. The throughput in the
simulation can be calculated by using (7):
Max_Throughput= Buf_Size / RTT (7)
Where Max_Throughput= Maximum throughput; Buf_Size= Received buffer size; RTT= Round
Trip Time.
RED algorithm gives better throughput at the 500 Mbps bandwidth. At the same time, the token
bucket algorithm has the lowest throughput rate. The average throughput is calculated 38.75 at
500 Mbps bandwidth RED algorithm whereas it is calculated 31.25 for token bucket algorithm,
that makes the RED algorithm 19.4 % better than the token bucket algorithm. In addition, RED
algorithm provides the best throughput at 400 Mbps bandwidth whereas token bucket and
additional buffer techniques have the lowest throughput rate. The average throughput at 400
Mbps bandwidth is measured 39.25 Mbps for RED algorithm, 31.25 for token bucket technique,
and 31.75 Mbps for the additional buffer technique. As a result, we have obtained 19.7 % more
throughput when using RED algorithm at 400 Mbps bandwidth than using token bucket or
additional buffer techniques. On the other hand, ECN algorithm produces the highest
throughput at 300 Mbps bandwidth while the lowest throughput is examined at the additional
buffer technique. The average throughput is calculated 34 Mbps for ECN algorithm and 29
Mbps for the additional buffer technique. Hence, ECN algorithm produces 14.7 % more
throughput than the additional buffer technique. The average throughput of all techniques used
in this paper is given in Figure 8, Figure 9, and Figure 10. Table 1 describes the abbreviations of
different techniques used in the graphs.
Table 1: describes the abbreviations of different techniques used in the graphs.
Abbreviation
Full Description
TB
Token Bucket
Technique
BM
Bandwidth Metering
Technique
RED
Random Early
Detection Technique
AB
Additional Buffer
Technique
ECN
Explicit Congestion
Notification
Technique
Figure 8: The average throughput at 500 Mbps bandwidth with different techniques.
Figure 9: The average throughput at 400 Mbps bandwidth with different techniques.
Figure 10: The average throughput at 300 Mbps bandwidth with different techniques.
5.2 Loss/Drop of Packets Ratio %
The loss/drop of packets ratio can be calculated by using (8) and (9) :
LP= TSP- TAP (8)
Where LP= total number of loss packets; TSP= total number of sent packets; TAP= total
number of acknowledged packets.
DR= LP X 100 / TSP , Where DR = Drop rate (9)
All techniques discussed in this paper are investigated with various network bandwidths in order
to examine the packet loss ratio. Token bucket technique gives lower packet loss ratio at 500
Mbps bandwidth whereas the additional buffer technique provides higher packet loss ratio at the
same bandwidth. Token bucket reduces 22.3% of average packet loss than the additional buffer
technique. In contrast, at 400 Mbps bandwidth, the lowest packet loss ratio is obtained by ECN
algorithm whereas the highest is examined at RED algorithm. Hence, 5.3% reduction in packet
loss ratio with ECN algorithm is calculated. Conversely, RED algorithm produces lower packet
loss ratio at 300 Mbps bandwidth whereas the additional buffer technique gives higher packet
loss ratio. RED algorithm reduces 18.6% the average packet loss ratio than the additional buffer
technique. The average packet loss ratio is described in Figure 11, Figure 12, and Figure 13.
Figure 11: The average packet loss rate at 500 Mbps bandwidth with different techniques.
Figure 12: The average packet loss rate at 400 Mbps bandwidth with different techniques.
Figure 13: The average packet loss rate at 300 Mbps bandwidth with different techniques.
5.3 Admitted Sessions
We have set the optimal TCP window size that is most useful to control the congestion even in
interior nodes. We apply the following formula to find the optimal window size:
Optimal window size= 2 X B X DP (10)
Where B= Bandwidth; DP= Delay of Product.
The optimal window size helps to determine the current size of window in order to establish the
session as per bandwidth capacity of network. The additional buffer technique has 71 admitted
session whereas token bucket technique has 64 admitted sessions at 500 Mbps bandwidth.
Hence, the additional buffer technique has admitted 9.86% more than the token bucket
technique. Furthermore, the additional buffer technique and the bandwidth metering techniques
admitted 63 sessions which are measured as 12.7 % more sessions than the token bucket
technique at 400 Mbps bandwidth. In addition, RED and bandwidth metering algorithms accept
58 sessions comparing to 53 secessions through token bucket technique at 300 Mbps bandwidth.
Thus, 8.6 % more sessions are gained by RED and bandwidth metering algorithms than the
token bucket technique. The averages of sessions admitted are shown in Figure 14, Figure 15,
and Figure 16.
Figure 14: The average number of sessions admitted at 500 Mbps bandwidth with different
techniques.
Figure 15: The average number of sessions admitted at 400 Mbps bandwidth with different
techniques.
Figure 16: The average number of sessions admitted at 300 Mbps bandwidth with different
techniques.
6. DISCUSSION OF RESULTS
The main objective of this work is to determine the most effective technique in PCN domain.
We consider three important parameters to justify the weaknesses and the strengths of each
algorithm. The first is an average throughput, the second is packet loss/drop ratio, and third is
number of average admitted sessions. As the total network bandwidth increases, the RED
algorithm is a better choice to enhance the throughput because of the small average queue size,
which causes high throughput and low average delay. In contrast, as the total network
bandwidth decreases, the ECN algorithm is a better choice to enhance the throughput due to the
less sensitivity to network parameters, which also improves throughput and reduces delay.
However, token bucket and additional buffer techniques always have the lower throughput.
Regarding the packet loss rate, RED algorithm is found to be the best as the network bandwidth
decreases and Token bucket technique is more reliable as the network bandwidth increases.
RED algorithm marks packets randomly with a certain probability P
A
instead of discarding them
when average queue size avg is between the values of min_thr and max_thr. As a result, it
encourages the early detection of congestion and thus adjusts the window size to avoid packet
loss. Moreover, the additional buffer technique enables to admit more sessions as the bandwidth
increases and RED algorithm enables to admit more sessions as the bandwidth decreases.
However, the Token bucket technique always admits lower number of sessions comparing with
other techniques. The additional buffer technique admits more sessions because it reduces the
admissible rate threshold (Tr) when the network is congested, so, it increases the number of
admitted sessions. To conclude, RED algorithm is the best technique that reflects more positive
results than others. RED has a higher average throughput and a lower delay time. It provides a
fair service of packet dropping due to the better loss packet rate and admits a satisfactory
number of sessions. The benchmark of our results is shown in Table 2.
Table 2: Benchmark of all techniques used.
7. CONCLUSION
In this paper, we have reported the results of simulation results and investigated the
performance of five different PCN techniques with different bandwidth. These five PCN
techniques are implicitly discussed with slide modification in their existing algorithms. This
modification helps to determine the exact behavior of each technique in congested PCN domain.
The goal of the research is to determine which technique is the most suitable in particular
scenario. RED algorithm provides better performance comapred to other techniques in terms of
throughput due to its small average queue size especially when there is enough bandwidth. RED
has a lower packet loss ratio. It has an ability to mark packets according to probability P
A
instead of discarding them in small bandwidth. The number of admitted sessions are calculated
more using the additional buffer technique in high bandwidth and maintains minimum sessions
in case of low bandwidth whereas RED technique makes more sessions in small bandwidth.
In future work, we plan to integrate the best features of all techniques and introduce new
technique in PCN domain to avoid congestion, and maintain high quality of service by
achieving maximum throughput.
FACTORS
300 Mbps
400 Mbps
500 Mbps
AVERAGE
THROUGHPUT
TECHNIQUE
AVG
AB
29
ECN
34
TB
30.25
BM
31
RED
33
TECHNIQUE
AVG
AB
31.75
ECN
33.75
TB
31.25
BM
35.25
RED
39.5
TECHNIQUE
AVG
AB
33.5
ECN
34.5
TB
31.25
BM
35.5
RED
38.75
AVERAGE
PACKET LOSS
RATE
TECHNIQUE
AVG
AB
3.38
ECN
3
TB
3.1
BM
2.95
RED
2.75
TECHNIQUE
AVG
AB
2.4
ECN
2.35
TB
2.38
BM
2.38
RED
2.48
TECHNIQUE
AVG
AB
1.85
ECN
2.18
TB
2.38
BM
2.3
RED
2.23
AVERAGE
ADMITTED
SESSIONS
TECHNIQUE
AVG
AB
55
ECN
56
TB
53
BM
58
RED
58
TECHNIQUE
AVG
AB
63
ECN
62
TB
55
BM
63
RED
62
TECHNIQUE
AVG
AB
71
ECN
66
TB
64
BM
65
RED
65
REFERENCES
[1] P. Eardley, “Pre-Congestion Notification (PCN) Architecture,” RFC 5559 (Informational), Jun. 2009.
[Online]. Available: http://www.ietf.org/rfc/rfc5559.txt
[2] B. Braden et al. RFC2309: Recommendations on Queue Management and Congestion Avoidance in
the Internet, Apr. 1998.
[3] S. Latr´e, B. De Vleeschauwer, W. Van de Meerssche, F. De Turck, P. Demeester ”Design and
Configuration of PCN Based Admission Control in Multimedia Aggregation Network”, in
Proceedings of IEEE Globalcom, 2009.
[4] K. Roobroeck, S. Latr´e, T. Wauters, F. De Turck, "Optimized network utilisation through buffering
in PCN enabled multimedia access networks," Integrated Network Management (IM), 2011
IFIP/IEEE International Symposium on , vol., no., pp.1243-1249, 23-27 May 2011.
[5] M. Menth, F. Lehrieder, "Performance of PCN-Based Admission Control Under Challenging
Conditions," Networking, IEEE/ACM Transactions on , vol.20, no.2, pp.422-435, April 2012.
[6] M. Menth et al., “A Survey of PCN-Based Admission Control and Flow Termination,” IEEE
Communications Surveys & Tutorials, vol. 12, no. 3, 2010.
[7] M. Menth, M. Hartman, ”Threshold configuration and routing optimization for PCN-based resilient
admission control”, Journal of Computer Networks, 2009.
[8] M. Menth, F. Lehrieder, ”Performance Evaluation of PCN-Based Admission Control”, in
Proceedings of Quality of Service, 2008. IWQoS 2008. 16th International Workshop on , 2008, pp
110-120.
[9] S. Floyd and V. Jacobson. Random Early Detection Gateways for Congestion Avoidance. IEEE/ACM
Transactions on Networking, 1(4):397413, Aug. 1993.
[10] K. Ramakrishnan, S. Floyd, and D. Black. RFC3168: The Addition of Explicit Congestion
Notification (ECN) to IP, Sept. 2001.
[11] R. Arora, O. Yang, H. Zhang, “TCP Networks with ECN over AQM”. OPNETWORK 2002, August
2002, Washington DC, USA.
[12] S. Latre, B. De Vleeschauwer, W. Van de Meerssche, S. Perrault, F. De Turck, P. Demeester, K. De
Schepper, C. Hublet, W. Rogiest, S. Custers, and W. Van Leekwijck, “An Autonomic PCN based
Admission Control Mechanism for Video Services in Access Networks,” in IEEE ACNM, 2009.
[13] P. S, Floyd, “TCP and explicit congestion notification”, in ACM SIGCOMM Computer
Communication Review, October 1994.
[14] K.Venkateswara Rao, E.Hemalatha, “Performance enhancement by measuring traffic at edge routers
in a high speed networks”, in International Journal of Engineering Science and Technology, vol. 3,
no. 6, June, 2011.
[15] J. Wang, X. Wang, H. Wang, M. Huang, L. Ma, Z. Pan , "The Research Of Active Network
Congestion Control Algorithm Based On Operational Data," Communications and Networking in
China, 2007. CHINACOM '07. Second International Conference on , vol., no., pp.213-217, 22-24
Aug. 2007.
[16] A. Krempa, “Analysis of RED algorithm with responsive and non-responsive flows”, Poznan
University of Technology Academic Journals, 2007.
Authors
Mrs. Marwah Almasri: is a Ph.D. student in the Computer Science &
Engineering Department at the University of Bridgeport. She received her MBA
in Management Information System (MIS) from the University of Scranton, PA,
in 2011. She received award from MIS department at the University of Scranton
for her outstanding work. She holds a bachelor degree in Computer Science &
Engineering from Taibah University in Medina, Saudi Arabia. Her research
interests include congestion mechanisms, wireless and sensor networks, computer
networks, mobile computing, and network security.
Dr. Khaled Elleithy: is the Associate Dean for Graduate Studies in the School
of Engineering at the University of Bridgeport. His research interests are in the
areas of, network security, mobile wireless communications formal approaches
for design and verification and Mobile collaborative learning. He has published
more than one two hudereds research papers in international journals and
conferences in his areas of expertise.
Dr. Elleithy is the co-chair of 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 four years. Dr. Elleithy is the editor
or co-editor of 10 books published by Springer for advances on Innovations and Advanced Techniques in
Systems, Computing Sciences and Software.
Dr. Elleithy received the B.Sc. degree in computer science and automatic control from Alexandria
University in 1983, the MS Degree in computer networks from the same university in 1986, and the MS
and Ph.D. degrees in computer science from The Center for Advanced Computer Studies in the
University of Louisiana at Lafayette in 1988 and 1990, respectively. He received the award of
"Distinguished Professor of the Year", University of Bridgeport, during the academic year 2006-2007.
Mr. Abdul Razaque is PhD student of computer science and Engineering
department in University of Bridgeport. His current research interests include
the design and development of learning environment to support the pedagogical
activities in open, large scale and heterogamous environments, collaborative
discovery learning and the development of mobile applications to support
mobile collaborative learning (MCL), the congestion mechanism of transmission
of control protocol including various existing variants, delivery of multimedia
applications. He has published over 40 research contributions in refereed
conferences, international journals and books. He has also presented his work more than 10 countries.
During the last two years he has been working as a program committee member in IEEE, IET, ICCAIE,
ICOS, ISIEA and Mosharka International conference.
Abdul Razaque is member of the IEEE, ACM and Springer. Abdul Razaque served as Assistant Professor
at federal Directorate of Education, Islamabad. He completed his Bachelor and Master degrees in
computer science from university of Sind in 2002. He obtained another Master degree in computer
Science with specialization of multimedia and communication (MC) from Mohammed Ali Jinnah
University, Pakistan in 2008. Abdul Razaque has been directly involved in design and development of
mobile applications to support learning environments to meet pedagogical needs of schools, colleges,
universities and various organizations.
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.
Conference Paper
Full-text available
The introduction of new added value services such as IPTV has introduced great challenges for today's broadband DSL access networks as these services have stringent quality demands. In an attempt to protect the quality delivery of existing sessions, operators employ admission control mechanisms that limit the amount of sessions transmitted in the network. Current admission control mechanisms require a traffic specification of each stream, in order to know beforehand how many resources need to be reserved. For variable bit rate videos, which are bursty of nature, resources are reserved using the peak rate of the video. This leads to under-utilisation of the network as the amount of resources needed is over-dimensioned. We propose an autonomic measurement based admission control algorithm, optimised for the protection of video services in multimedia access networks. The algorithm is based on the IETF precongestion notification (PCN) mechanism and autonomically adjusts its parameters to the traffic characterisation of the video. The performance of this mechanism has been extensively evaluated in a packet based network simulation environment. Tests show that the autonomic nature of the algorithm leads to a better utilisation of the network while still avoiding any congestion in the network.
Conference Paper
Full-text available
DSL aggregation networks are evolving to the standard platform for the delivery of multimedia services such as television and network based personal video recording. These multimedia services introduce large challenges for network operators as they are sensitive to packet loss. Therefore, admission control mechanisms are required to avoid congestion caused by allowing too many sessions. However, as multimedia services are often bursty it is not possible to reserve a fixed amount of bandwidth in the network since this policy will lead to either over-admittance or under-admittance. Recently, the IETF Pre-Congestion Notification (PCN) Working Group, proposed a measurement based admission control mechanism, where the network load is measured at each node and sessions are allowed or blocked at the edge of the network. In this paper, we extend and evaluate the PCN mechanism: we propose a new measurement algorithm for PCN, based on bandwidth metering, and determine the configuration guidelines for the parameters of both the original token bucket based approach and the novel algorithm for different network conditions and traffic types. More specifically, we study PCN's applicability on protecting VBR video services, which is currently not studied in the PCN Working Group. Furthermore, we characterise the gain of PCN in comparison to a centralised admission control mechanism.
Conference Paper
Full-text available
gbroadband DSL networks are facing enormous challenges. These services have strict QoS demands in terms of packet loss, jitter and delay. In an effort to meet these demands, operators introduced centralized admission control mechanisms to avoid congestion when too many session were allowed. These centralized approaches often fail to effectively manage the avail­ able resources mainly because of the bursty nature of multimedia traffic. When transmitting variable bit rate videos resources are reserved based on the peak rate of the video. This leads to under­ utilisation of the network. Measurement based admission control mechanism have been proposed such as the IETF Pre-Congestion Notification (PCN) to allow better network utilisation. Each node in the PCN domain measures the network load and admits or blocks accordingly sessions at the edges of the network. Previous research proposed bandwidth metering and an autonomic rate adaptation algorithm which led to a better utilisation of the network but still introduced unnecessary bandwidth headroom caused by the variable bit rate of videos. In this paper, we propose an additional buffering step before traffic enters the PCN domain and determine configuration guidelines for the parameters. The performance of this buffering step has been evaluated in an NS - 2 based simulator environment. The conducted tests show a 26.5% increase of network utilisation.
Article
There is an enormous increase of traffic on the network due to usage of real-time multimedia applications, which became an indispensable part of Internet traffic in the present day world. These applications include voice over IP, streaming audio and video, Internet gaming and real-time video conferencing. Such applications are multimedia information, geographical pictures, social network applications and global sharing applications. In order to reduce this traffic, high quality of service is needed which would be possible by Adaptive mechanisms. Integrated service on the network flows offer a bounded delay packet delivery to support real time applications. To provide bounded delay service, networks must use admission control to regulate their load. It is very mandatory to allocate and manage resources for multimedia traffic and various concurrent applications flows with real-time performance, in order to provide reliable quality of service (QoS). In this paper, we develop an enhanced model and an algorithm for admission control of real-time flows; comparisons among various distinguish features of admission control algorithms. In our approach, admission decision is made for each flow at the edge routers, but it is scalable because per-flow states are not maintained and the admission algorithm is simple. In the proposed admission control scheme, an admissible bandwidth, which is defined as the maximum rate of a flow that can be accommodated additionally while satisfying the delay performance requirements for both existing and new flows, is calculated based on the available bandwidth measured by edge routers. The admissible bandwidth is a threshold for admission control, and thus, it is very important to accurately estimate the admissible bandwidth. The performance of the proposed algorithm is evaluated by taking a set of simulation experiments using bursty traffic flows and the results are found to be encouraging.
Article
Precongestion notification (PCN) is a packet-marking technique for IP networks to notify egress nodes of a so-called PCN domain whether the traffic rate on some links exceeds certain configurable bounds. This feedback is used by decision points for admission control (AC) to block new flows when the traffic load is already high. PCN-based AC is simpler than other AC methods because interior routers do not need to keep per-flow states. Therefore, it is currently being standardized by the IETF. We discuss various realization options and analyze their performance in the presence of flash crowds or with multipath routing by means of simulation and mathematical modeling. Such situations can be aggravated by insufficient flow aggregation, long round-trip times, on/off traffic, delayed media, inappropriate marker configuration, and smoothed feedback.
Article
This paper presents Random Early Detection (RED) gateways for congestion avoidance in packet-switched networks. The gateway detects incipient congestion by computing the average queue size. The gateway could notify connections of congestion either by dropping packets arriving at the gateway or by setting a bit in packet headers. When the average queue size exceeds a preset threshold, the gateway drops or marks each arriving packet with a certain probability, where the exact probability is a function of the average queue size. RED gateways keep the average queue size low while allowing occasional bursts of packets in the queue. During congestion, the probability that the gateway notifies a particular connection to reduce its window is roughly proportional to that connection's share of the bandwidth through the gateway. RED gateways are designed to accompany a transport-layer congestion control protocol such as TCP. The RED gateway has no bias against bursty traffic and avoids the global synchronization of many connections decreasing their window at the same time. Simulations of a TCP/IP network are used to illustrate the performance of RED gateways.
Article
The purpose of this document is to describe a general architecturefor flow admission and termination based on aggregated (pre-) congestion information in order to protect the quality of service of established inelastic flows within a single DiffServ domain.
Article
Pre-Congestion Notification (PCN) in IP networks uses packet metering and marking within a PCN domain to notify its egress nodes whether link-specific admissible or supportable rate thresholds have been exceeded by high-priority traffic. Based on this information simple admission control and flow termination is implemented. The latter is a new flow control function and useful in case of overload through high priority traffic which can occur in spite of admission control, e.g., when traffic is rerouted in failure cases. Resilient admission control admits only so much traffic that admitted traffic can be rerouted without causing congestion on backup paths in case of a likely failures, e.g., single link failures.We propose algorithms to configure the link-specific PCN rate thresholds such that resources are utilized efficiently and fairly by competing traffic aggregates while meeting resilience constraints. This is done for the single and dual marking PCN architecture whereby the single marking case is more demanding since it requires that the supportable rate is a fixed multiple of the admissible rate on all links within a single PCN domain. Furthermore, we derive objective functions to optimize the underlying routing system for both cases. Our performance results for various network types show that the dual marking PCN architecture leads to significantly better resource efficiency than the single marking PCN architecture.