SCHEDULING AND QOS FOR LEAK RATE IN POLICING

MECHANISMS OVER TELECOMMUNICATIONS NETWORK

Somchai Lekcharoen *and

Chanintorn Jittawiriyanukoon **

* Faculty of Information Technology, Rangsit University

, Thailand

** Faculty of Science and Technology, Assumption University, Thailand

Keywords: VDSL, Fuzzy Control Policing Mechanisms and Queue Leak Rate (QLR).

Abstract: High-performance frame communication networks including VDSL have been conceived to carry traffic

sources and support a continuum of transport rates ranging from low bit-rate to high bit-rate traffic. As a

number of telecommunications traffic (bursty traffic) fluctuates on a certain network it results in

congestion. The traditional policing mechanisms are finite-sized buffers with queue management techniques

and fixed leak rate. Most queue management schemes employ fixed thresholds or a limited number of

arrival frames, to determine when to allow or discard the entry of frames. However, traditional policing

mechanisms have proved to be inefficient in coping with the conflicting requirements of ideal policing

mechanisms, that is, low dropping frames and high conforming frames. An alternative solution based on

artificial intelligence techniques, specifically, in the field of fuzzy systems is introduced. In this paper, a

fuzzy control queue leak rate (QLR) of the buffer prior to policing mechanism is investigated. The

performance of this alternative method is then compared with traditional policing mechanisms. Simulation

results show that over VDSL network, the fuzzy control scheme helps improve performance of QLR in

policing mechanisms. The performance of proposed method is much better than traditional policing ones.

1 INTRODUCTION

In VDSL networks, large number of traffic sources

always arises, resulting in network congestion. To

prevent this situation, some congestion control

mechanisms depending on type of traffic sources are

introduced. Also, some policing mechanisms can be

considered to help improve the main performance

measures, such as queueing delay, transmission

delay, bandwidth allocation, and throughput.

Presently, VDSL is getting very significant attention

from implementers and service providers as it

guarantees to deliver highest bandwidth data rates to

dispersed locations with little changes to the existing

infrastructure. VDSL services ensure dedicated,

point-to-point, public network access over twisted-

pair copper wire. As it is difficult to model a

complicated control system, especially at the

scheduling or at the queue leak rate (QLR), the

development of alternative modeling and control

techniques including fuzzy logic is necessary. Fuzzy

model may lead to describe the behavior of systems

well whenever the decision has to be made for the

ambiguous application in control. Thus due to the

demand for an inexpensive but reliable system, the

fuzzy approach may turn out to be a useful

mechanism to control cases with both the

complexity and uncertainty. There are many

previous studies involving traffic policing

mechanisms (Lekcharoen et al, 2004),(Pakdeepinit

et al, 2002),(Chen, 2000), however QLR is not

mentioned. In this paper, the idea of QLR control is

proposed to help improve the system performance.

The performance after being controlled by QLR is

compared to the previous one (uncontrolled by

QLR).

This paper is organized as follows. An overview

of traffic policing mechanisms is proposed in section

II. Section III defines the model of a fuzzy control

QLR in policing mechanisms. Section IV describes

the simulation model and section V discusses about

a performance evaluation of the proposed QLR and

then a comparison to traditional policing

mechanisms is produced. Section VI summarizes the

conclusion and makes recommendation for future

research.

358

Lekcharoen S. and Jittawiriyanukoon C. (2005).

SCHEDULING AND QOS FOR LEAK RATE IN POLICING MECHANISMS OVER TELECOMMUNICATIONS NETWORK.

In Proceedings of the Second International Conference on e-Business and Telecommunication Networks, pages 359-363

DOI: 10.5220/0001409203590363

 SciTePress

2 DESCRIPTION AND

MODELING OF TRAFFIC

POLICING MECHANISM

2.1 Requirement for policing

mechanism

Traffic policing allows us to control the maximum

rate of traffic sent or received on an interface during

the entire active phase and must operate in real time.

The mechanisms have been proposed which are

described in the following sections.

2.1.1 Traffic source model

In our simulation, a burst traffic stream from a

single source is modeled as a burst/silence traffic

stream. The burst-period model is single flow and

silence-period model is silent. Burst-periods and

silence periods are strictly alternating (Lekcharoen

et al, 2004).

2.1.2 Policing mechanisms

Various congestion control traffic policing

mechanisms are also introduced in (Lekcharoen et

al, 2004). In this paper, three policing mechanisms

are investigated including the Leaky Bucket(LB),

the Jumping Window Mechanism(JW) and the

triggered jumping window(TJW) for our

analysis(Pakdeepinit et. al, 2002), (Chen, 2000).

3 FUZZY CONTROL POLICING

MODEL

In this section, we will first describe a new fuzzy

control policing mechanism which meets the

requirements of performance, flexibility and cost-

effective implementation of VDSL networks.

Concepts of fuzzy sets and fuzzy logic control have

been introduced and developed by (Rose and

David,2000), (You-Chang et. al, 2001).

3.1 Regulator input fuzzification

Input variables are transformed into fuzzy set

(fuzzification) and manipulated by a collection of

IF-THEN fuzzy rules, assembled in what is known

as the fuzzy inference engine, as show in figure

below.

i. .

Figure 1: Library of fuzzy sets used in the fuzzification

process (The input variable I).

Figure 2: The term set of output variable Q.

Figure 3.The term set of output variable D.

Table 1: The fuzzy rules

IF I is Low (L) THEN Q is Low and D is Low

IF I is Medium (M) THEN Q is Medium and D is Medium

IF I is High (H) THEN Q is High and D is High

IF I is VeryHigh (VH)THEN Q is High and D is High

In our experiment, fuzzy control QLR in

policing mechanism uses a set of rules as specified

above (figure 1,2,3 and table 1). The selection of

rule base is based on our previous research

(Lekcharoen et al, 2004), (Rose and David, 2000),

(Ruy and Torsten, 2004) and assumptions of how

the system should respond. The assumptions and set

of rules may not relate to the practical point of view

but at least it could pinpoint how to improve the

performance of the system.

4 SIMULATION MODEL

The following figure 4 shows a simulation model

used in this paper.

Figure 4: Simulation model.

1 5 6 9 10 12 13 14Mbps

L M H VH

1 8 9 14 15 22 23 105 frames

Input

traffic

Fuzzy

Controll

Dropped FR

QLR

VDSL

Output

Scheduler

Policer

MQL

1 4 5 8 9 11 12 35 Mb

SCHEDULING AND QOS FOR LEAK RATE IN POLICING MECHANISMS OVER TELECOMMUNICATIONS

NETWORK

359

4.1 Input traffic

This paper confines the discussion on mainly data.

Data source are generally bursty in nature whereas

voice and video sources can be continuous or bursty,

depending on the compression and coding

techniques used.

4.2 Characteristics of queuing model

There are three components with certain

characteristics that must be examined before the

simulation models are developed.

4.2.1 Arrival characteristics

The pattern of arrivals input traffic is mostly

characterized to be Poisson arrival processes

(Pakdeepinit et al, 2002). Like many random events,

Poisson arrivals occur in such a way that for each

increment of time (T), no matter how large or small,

the probability of arrival is independent of any

previous history. These events may be individual

labels, a burst of labels, label or packet service

completions, or other arbitrary events.

4.2.2 Service facility characteristics

In this paper, service times are randomly distributed

by the exponential probability distribution. This is a

mathematically convenient assumption if arrival

rates are Poisson distributed. In order to examine the

traffic congestion at output of VDSL downstream

link with capacity 15Mbps

(www.dlink.pl/docs/datenblatt/DEV3001_datasheet

_en.pdf), the service time in the simulation model is

specified by the speed of this VDSL link, resulting

that a service time is set to be exponential

distribution with mean 216 µs ,where the frame size

is 405 bytes (www.etsi.org). The buffer size at the

entrance to VDSL network is set to be 1,024 frames

(www.dlink.pl/docs/datenblatt/DEV3001_datasheet

_en.pdf). Once it is exceeding the buffer size then it

is considered to be non-conforming frames (or

dropped frames).

4.2.3 Source traffic descriptor

The source traffic descriptor is the subset of traffic

parameters requested by the source (user), which

characterizes the traffic that will (or should) be

submitted during the connection. The relation of

each traffic parameter used in the simulation model

is defined in table 2.

Table 2: Input parameters

5 RESULTS AND ANALYSIS

The comparison of results before and after applying

QLR in policing mechanisms is shown in figure: 5,

6, 7, 8 and 9. They are listed as LB, JW, TJW and

FLB, FJW and FTJW respectively. This This section

focuses on the simulation results as of LB, JW,

TJW, FLB, FJW and FTJW. The input traffic rate

(frame rate varies from 1 Mbps to 35 Mbps) with

burst/silence ratio of 100:100 was simulated and

results were shown in figure 5. It clearly determines

that the fuzzy control QLR in policing mechanism

(that is FLB,FJW and FTJW) is much better

throughput than traditional policing mechanism (LB,

JW, TJW). Throughput is one of QoS to help

guarantee higher reliability of network performance.

In conclusion, the FLB, FJW and FTJW can handle

real time applications such as multimedia traffic.

Figure 6 demonstrates that we can help conserve the

conforming frames by simply reducing number of

dropped frames. The highest number of dropped

frames as shown in this figure is produced by

traditional policing mechanisms (LB,JW and TJW).

A regular network may cause a poor QoS by higher

non-conforming frames. Especially, a quality of

multimedia traffic such as video conference, video

on demand during the online display mode must

guarantee QoS while high dropped frames will

cause unclear pictures and retransmission.

Furthermore It will lead to higher delay time.

In figure 7, the simulation result determines

more utilization of fuzzy control prior queue length

and depleting rate in policing mechanisms compared

to traditional policing mechanisms with burst/silence

ratio of 100:100. The increment of utilization factor

does not seem to be relevant to the performance

improvement. The higher utilization may cause

approach of a bottleneck situation which can boggle

down the system. In fact the 78% in FLB, FJW and

FTJW will not affect the situation of bottleneck.

Positively this factor will rather improve the cost

effectiveness of the VDSL devices.

Figure 8 and 9 show that all fuzzy control prior

queue length and depleting rate in policing

Arrival rate

(Mbps)

Max Queue Length

(MQL in frames)

QLR

(Mbps)

1- 5 9 6

6-9 15 10

11-12 23 13

13-35 105 14

ICETE 2005 - SECURITY AND RELIABILITY IN INFORMATION SYSTEMS AND NETWORKS

360

mechanisms, all frames have to wait longer in the

buffer next to the entrance of VDSL network. The

consequence of long waiting hour is compatible to

results shown in figure 8and 9. We can observe

many more frames in average reside in the queue

prior to the entrance of VDSL network. This is trade

off fuzzy control policing mechanism. We can help

ease this congestion by increasing the data rate in

VDSL network to be higher.

6 CONCLUSIONS AND

RECOMMENDATIONS FOR

FUTURE RESEARCH

In this paper, we carried out a comprehensive study

to investigate the performance of three selected

traditional policing mechanisms; namely LB, JW,

TJW and fuzzy control QLR in policing mechanism;

namely FLB, FJW, FTJW with fixed bursty/silence

sources. This study was conducted through

simulation for which a model was developed.

We found from simulation result in general that

the fuzzy control QLR in policing mechanisms are

better than traditional policing mechanisms with

fixed bursty/silence sources. The fuzzy control QLR

in policing mechanisms will also help ease the

tremendous amount of traffic fluctuations into the

VDSL network and prevent the network from

nonconforming frames with 58% reduction of traffic

load as shown in figure 9. Fuzzy control QLR in

mechanisms can guarantee that they are better than

traditional policing mechanisms.

In the future work, we will focus on the

investigation of a network of queue (NoQ) for

central pool prior to the VDSL network.

500

1000

1500

2000

Conform

Frame (fram

)

FLB LB FJW JW FTJW TJ

Figure 5 illustrates the conforming frames by comparison

between FLB, LB, FJW, JW, FTJW and TJW at burst :

silence = 100:100.

500

1000

1500

2000

Non-confor

Frame (fram

)

FLB LB FJW JW FTJW TJ

Figure 6 illustrates the non-conforming frames by

comparison between FLB, LB, FJW, JW, FTJW and TJW

at burst : silence = 100:100.

Utiliza

FLB LB FJW JW FTJW TJ

Figure 7 illustrates the utilization by comparison between

FLB, LB, FJW,JW, FTJW and TJW at burst : silence =

100:100.

Mean qu

length ( fram

)

FLB LB FJW JW FTJW TJ

Figure 8 illustrates the mean queue length by comparison

between FLB, LB, FJW, JW, FTJW and TJW at burst :

silence = 100:100.

500

1000

1500

2000

Mean queue time (Micr os

)

FLB LB FJW JW FTJW TJ

Figure 9 illustrates the mean time in queue comparison

between FLB, LB, FJW, JW, FTJW and JW at burst :

silence = 100:100

SCHEDULING AND QOS FOR LEAK RATE IN POLICING MECHANISMS OVER TELECOMMUNICATIONS

NETWORK

361

REFERENCES

S. Lekcharoen et. al, “Performance Evaluation of VDSL

Network with Fuzzy Control Policing Mechanisms”,

Proceeding of International Conference on

Telecommunications and Computer Networks

(TCN’04), 2004, pp. 56-60.

P. Pakdeepinit et. al, “Performance Comparison of ATM

Policy Mechanisms with Telecommunications

Traffic", Proceeding of The 6th World

Multiconference on Systematics, Cybernetics and

Informatics, 2002, pp. 373-378.

Y. W. Chen, "Study on the prevention of SYN flooding by

using traffic policing", NOMS 2000 - IEEE/IFIP

Network Operations and Management Symposium,

no. 1, 2000, pp. 593-604.

Rose Qingyang Hu and David W. Peter, “A Predictive

Self-Tuning Fuzzy-Logic Feedback Rate Controller”,

IEEE/ACM transactions on networking, 2000, pp.

697-709.

Ruy de Oliveira and Torsten Braun, "A delay-based

approach using fuzzy logic to improve TCP error

detection in ad hoc networks", WCNC 2004 - IEEE

Wireless Communications and Networking

Conference, vol. 5, no. 1, 2004.

You-Chang Ko, Sun-Chun Park, Chong-Youn Chun,

Hyong-Woo Lee, Choong-Ho Cho, "An adaptive QoS

provisioning distributed call admission control using

fuzzy logic control", ICC 2001 - IEEE International

Conference on Communications, no. 1, 2001.,

pp. 356-360.

www.dlink.pl/docs/datenblatt/DEV3001_datasheet_en.pdf

www.etsi.org.

ICETE 2005 - SECURITY AND RELIABILITY IN INFORMATION SYSTEMS AND NETWORKS

362