You are on page 1of 22

Acknowledgements

First and foremost, I would like to thank my project guide, Prof. A. B PQR,
for his guidance and support. I will forever remain grateful for the constant support
and guidance extended by my guide, in making this seminar report. Through our many
discussions, he helped me to form and solidify ideas.
With a deep sense of gratitude, I wish to express my sincere thanks to, Prof.
Dr. S. S. Sane for his immense help in planning and executing the works in time. My
grateful thanks to the departmental staff members for their support.
I would also like to thank my wonderful colleagues and friends for listening my
ideas, asking questions and providing feedback and suggestions for improving my ideas.

Mr./Ms. ABC
Abstract

In recent years, explosive demand for network bandwidth has become a major
challenge. A continuous demand for networks of high capacities at low cost exists. Op-
tical data communication has been acknowledged as the best solution for meeting the
present bandwidth requirements of the users and for supporting future network services.
Contents

1 Introduction 1
1.1 Optical Burst Switching (OBS) . . . . . . . . . . . . . . . . . . . . . . . 1

2 Literature Survey 5
2.1 Burst Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 First Fit Unscheduled Channel (FFUC) Algorithm . . . . . . . . 5
2.1.2 Latest Available Unscheduled Channel Algorithm . . . . . . . . . 6
2.1.3 First Fit Unscheduled Channel with Void Filling (FFUC-VF) Al-
gorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.4 Latest Available Unused Channel with Void Filling (LAUC-VF)
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.5 Minimum Void Scheduling Algorithm . . . . . . . . . . . . . . . . 8

3 Problem Definition 12
3.1 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2.1 Reverse Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3 Timeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.4 System Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4 Conclusion 16

Bibliography 17
List of Figures

1.1 OBS Core Node . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1 Illustration of FFUC and LAUC Scheduling Algorithm. . . . . . . . . . . 6


2.2 Illustration of FFUC-VF and LAUC-VF Scheduling Algorithm. . . . . . 8
2.3 Illustration of Scheduling Algorithms . . . . . . . . . . . . . . . . . . . . 9

3.1 General Void Scheduling Approach. . . . . . . . . . . . . . . . . . . . . . 13


3.2 Reverse Scheduling Approach. . . . . . . . . . . . . . . . . . . . . . . . . 14
List of Tables

2.1 Comparison of Different Scheduling Techniques. . . . . . . . . . . . . . . 10

3.1 Timeline Chart. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15


Chapter 1

Introduction

In recent years, explosive demand for network bandwidth has become a major
challenge due to increasing global popularity of the Internet and the increased applica-
tions it affords. A continuous demand for networks of high capacities at low cost exists.
Optical data communication has been acknowledged as the best solution for meeting the
present bandwidth requirements of the users and for supporting future network services.
The capacity of optical network is realized by WDM technology. WDM evolution has
phases optical circuit switching, optical packet switching and optical burst switching net-
works which are under category All Optical Networks (AON).

1.1 Optical Burst Switching (OBS)


Traffic generated by Web browsers, variable-bit-rate (VBR) video sources are all
self-similar (or bursty at all time scales). For such traffic OBS is most suitable. For
burst switching, the basic switching entity is a burst. A burst is assembled at an ingress
edge router by aggregating a number of IP packets which are destined to the same egress
edge router. Bursts, being substantially bigger than IP packets are easier to switch with
relatively small overhead. A burst has two parts called the control (header) and data
(payload). They are referred to as control packet and data burst, respectively.
In OBS, the control and data parts of a burst are transmitted separately with

1
Seminar Title CHAPTER 1. INTRODUCTION

a time gap between the two parts, which is called the offset time. The offset time is a
variable depending on the number of hops the burst has to traverse before reaching the
destination node. This is done to ensure that the control part which carries the header
information such as the source node address, destination node address, offset time, burst
duration and quality of service (QoS) requirements among others has already reserved
resources for transmission of the data burst by the time it arrives at a node. The burst
offset time could also be adjusted to support QoS.
When a burst is ready, a control packet is sent to the core network. It is analyzed
in each core node; the routing decision is made, and sent to the next node. After an offset
time, the actual burst is also sent, without waiting for any acknowledgment. It traverses
the network entirely in the optical domain, as the path is already set according to the
information contained in the control packet. When the burst reaches its destination node
it is disassembled, and the resulting IP packets are sent to their respective destinations.
The offset time plays a key role in OBS and is determined before the burst is sent. For
a pair of source and destination nodes H hops apart, a base offset time = D * H, where
D refers to the processing time for each node, which is the time to process the control
packet, reserve the appropriate bandwidth and set up the switch. In an OBS network,
the edge routers and core routers connect with each other with WDM links.

It consist of an electronic router and an OBS interface. Edge nodes provide


burst assembly/disassembly functions. It consist of an electronic router and an OBS
interface. It collect upper layer traffic, sort it based on destination addresses and aggre-
gates it into variable sized bursts. They are also responsible for setting up connections
for a burst. This procedure contains signaling, routing and wavelength allocation. A
signaling is procedure by which services are provisioned. By using signaling protocols
a control packet can reserve resources for the corresponding data burst by guiding it
through a routing path. Routing takes fixed path routing where physical route is prede-
fined for each pair of OBS edge nodes. Wavelength allocation is done to the bursts as
per availability and scheduling. Fig ?? shows the architecture of edge/ingress node.
It consists of two layers. The upper layer is responsible for processing control
packets and configuring the switching fabric. This layer processes control packets, it first

2
Seminar Title CHAPTER 1. INTRODUCTION

O/E BP1 E/O


BHP
O/E Switch BP2 E/O
Control
Channel O/E BP3 E/O
Electronic Control Path

Optical Data Path


1 1

Optical 2
2
Cross Connect

3 3
Data
Channel

Wavelength converters
FDL Buffer pool BP: Burst Processor

Figure 1.1: OBS Core Node

converts header packets from optical to electronic domain. And then they are sent to the
BHP switch which routes that BHPs to the respective burst processors (BPs) according
to destination of data bursts as shown in Fig 1.1. Each BP is responsible for making
channel scheduling decisions for single outgoing link. BP responsible for configuring the
optical switching matrix such that the burst arrived on incoming data channels can pass
through the desired outgoing channel directly without buffering. The lower layer consists
of optical ports, wavelengths, and optical to optical connections.
A core node is mainly composed of an optical switching matrix and a switch con-
trol unit and responsible for routing and scheduling based on the burst header packets.
Depending on the switch architecture, it may or may not be equipped with optical buffer-
ing. The fiber links carry multiple wavelengths and each wavelength can be seen as a
channel.
Features of OBS:

ˆ Due to strong separation between control packet and data burst OBS need not
require tight synchronization between data and control [4]. Hence leads to high
network manageability and flexibility (for rescheduling).

3
Seminar Title CHAPTER 1. INTRODUCTION

ˆ No need of buffering.

ˆ Low setup latency.

ˆ Ensures efficient use of bandwidth as it reserves wavelength at time of actual data


transfer.

ˆ Only few control channels go through O/E/O conversion [4].

4
Chapter 2

Literature Survey

2.1 Burst Scheduling


The algorithm may need to maintain the latest available unscheduled time (LAUT)
or the horizon, gaps and voids on every outgoing data channel. Traditionally, the LAUT
of a data channel is the earliest time at which the data channel is available for an un-
scheduled data burst to be scheduled. A gap is the time difference between the arrival
of the unscheduled burst and ending time of the previously scheduled burst. A void is
the unscheduled duration (idle period) between two scheduled bursts on a data channel.
For void filling algorithms, the starting and the ending time for each burst on every data
channel must also be maintained. To find a suitable wavelength among the candidate
wavelengths for an arriving burst, several data channel scheduling algorithms, such as
First Fit Unscheduled Channel algorithm (FFUC), Latest Available Unscheduled Chan-
nel (LAUC) [4, 6, 7, 8], Latest Available Unused Channel with Void Filling (LAUC-VF)
[4, 6, 7, 8], also several variants of the LAUC-VF algorithm such as Min-SV (Starting
Void), Min-EV (Ending Void) algorithms are proposed [6].

2.1.1 First Fit Unscheduled Channel (FFUC) Algorithm


The FFUC scheduling algorithm keeps track of the LAUT on every data channel.
A wavelength is considered for each arriving burst when the unscheduled time (LAUT) of

5
Seminar Title CHAPTER 2. LITERATURE SURVEY

the data channel is less then the burst arrival time. As shown in Fig. 2.1, data channels
D1 and D2 are available for the duration of the unscheduled burst. If the channels are
ordered based on the index of the wavelengths the arriving burst is scheduled on outgoing
data channel D1 . The primary advantage of FFUC is the simplicity of the algorithm.
The algorithm has worst case time complexity of O(W). Proper link utilization is not
achieved in FFUC.

NEW

LAUT

D0

LAUT

D1

LAUT
D2

LAUT

D3

time

Figure 2.1: Illustration of FFUC and LAUC Scheduling Algorithm.

2.1.2 Latest Available Unscheduled Channel Algorithm


The LAUC or Horizon [6] scheduling algorithm keeps track of the Latest Available
Unscheduled Time (LAUT or horizon) on every data channel and assigns the data burst
to the latest available unscheduled data channel. In this algorithm, scheduler keeps track
of horizon for each data channel. Horizon is the time after which no reservation has been
made on that data channel. The scheduler assigns each arriving burst to the channel
with minimum void formed by that burst on data channel. A wavelength is considered
for each arriving burst when the unscheduled time (LAUT) of the data channel is less
than the burst arrival time. As shown in Fig. 2.1, data channels D1 and D2 are available

6
Seminar Title CHAPTER 2. LITERATURE SURVEY

for the duration of the unscheduled burst. Also, we observe that Gap1 > Gap2 thus, the
arriving burst is scheduled on outgoing data channel with the minimum gap, i.e., D2 .
Since the algorithm needs to maintain and examine only one parameter which
is the latest available unscheduled time, it is computationally simple. On the other
hand, the LAUC algorithm does not use voids on any wavelength and thus is not able to
achieve the best burst dropping performance. Its computational complexity is given by
O(W), where W is the total number of wavelengths per fiber. It result in low bandwidth
utilization and a high burst loss rate.

2.1.3 First Fit Unscheduled Channel with Void Filling (FFUC-


VF) Algorithm
The FFUC-VF scheduling algorithm maintains the starting and ending times for
each scheduled data burst on every data channel. As shown in Fig. 2.2, data channels
D0 , D1 , D2 , D3 are available for the duration of the unscheduled burst. If the channels
are ordered based on the index of the wavelengths the arriving burst is scheduled on
outgoing data channel D0 . The goal of this algorithm is to utilize voids between two
data burst assignments. FFUC-VF is similar to FFUC except that in which such voids
could be filled by new arriving bursts.

2.1.4 Latest Available Unused Channel with Void Filling (LAUC-


VF) Algorithm
In LAUC scheduling algorithms, the voids created between two data burst assign-
ments on one data channel, which is unused channel capacity, have not been utilized.
LAUC-VF is similar to LAUC except that in which such voids could be filled by new
arriving bursts. In LAUC-VF, the start time and end time of voids for each data channel
are maintained.
As shown in Fig. 2.2, all the data channels and are available for the duration
of the unscheduled burst. Also, we observe that D3 had the least gap Gap3 thus, the
arriving burst is scheduled on D3 . By selecting the latest available unused wavelength

7
Seminar Title CHAPTER 2. LITERATURE SURVEY

NEW

D0

D1

D2

D3

time

Figure 2.2: Illustration of FFUC-VF and LAUC-VF Scheduling Algorithm.

instead of the latest available unscheduled channel as in LAUC, LAUC-VF achieves high
performance with low burst dropping probability.
The drawback of LAUC-VF is the high computational complexity as this algo-
rithm needs to examine the voids between various bursts. Information of voids have to
be stored for every wavelength and to be searched through every time scheduling takes
place. The LAUC-VF algorithm therefore runs in O(W+K) time where K is the max-
imum number of voids or bursts in any wavelength assuming that voids are searched
linearly.

2.1.5 Minimum Void Scheduling Algorithm


There are several variants of the LAUC-VF algorithm including Min-SV (Starting
Void), Min-EV (Ending Void). Min-SV is functionally the same as LAUC-VF but a much
faster implementation Min-SV uses a geometric approach and organizes the voids into a
balanced binary search tree. Min-SV algorithm finds a void that minimizes the distance
between the starting time of the void and the starting time of the burst. The Min-
SV algorithm takes O(log m) time to finish, which is a significant improvement over

8
Seminar Title CHAPTER 2. LITERATURE SURVEY

LAUC-VF. To date, it is the fastest scheduling algorithm that can produce an efficient
burst schedule. On the other hand, Min-EV tries to minimize the new void generated
between the end of new reservation. In fact, one can minimize the void newly generated
by first searching for a proper void using Min-EV first, and then if (and only if) such
a proper void cannot be found, search for a horizon using Min-SV, the Min-SV/EV
algorithms are the most desirable among all void-filling algorithms. Table 2.1 shows
performance of various scheduling algorithms, which shows that LAUC-VF, Min-SV,
Min-EV, ordered scheduling have better bandwidth utilization and a lower loss rate than
the LAUC algorithm

NEW

LAUC-VF, Min-SV
t1 t1’

t2 t2’

LAUC
t3

Min-EV
t4 t4’

Best Fit
t5 t5’

ts te

Figure 2.3: Illustration of Scheduling Algorithms

In this one can minimize the void newly generated by first searching
for a proper void using Min-EV first, and then if (and only if) such a proper void cannot
be found, search for a horizon using Min-SV.
Another variant of Min-SV is Best-Fit which tries to minimize total length of

9
Seminar Title CHAPTER 2. LITERATURE SURVEY

Table 2.1: Comparison of Different Scheduling Techniques.

Algorithms Time Dropping Bandwidth


Complexity Probability Utilization
FFUC O(W) High Low
LAUC O(W) High Low
LAUC-VF O(K+W) Low High
Min-SV/EV O(log k) Low High

starting and ending void after the reservation. Table 2.1 shows performance of various
scheduling algorithms, which shows that LAUC-VF, Min-SV, Min-EV have better band-
width utilization and a lower loss rate than the LAUC algorithm.

Table 2.1 uses the following notations

ˆ W : Number of wavelengths at each output port.

ˆ K : Maximum number of data bursts (or reservations) on all channels.

The Table 2.1 shows the comparison of the different scheduling algorithms.
The primary advantage of FFUC is the simplicity of the algorithm. The algorithm has
worst case time complexity of O(W). Although proper link utilization is not achieved in
FFUC i.e. its bandwidth utilization is low. In LAUC-VF information of voids have to
be stored for every wavelength and to be searched through every time scheduling takes
place. Therefore LAUC-VF algorithm runs in O(W+K) time where K is the maximum
number of voids or bursts in any wavelength assuming that voids are searched linearly. By
selecting the latest available unused wavelength instead of the latest available unscheduled
channel as in LAUC, LAUC-VF achieves high performance with low burst dropping
probability. Min-SV algorithm finds a void that minimizes the distance between the
starting time of the void and the starting time of the burst. The Min-SV algorithm
takes O(log K) time to finish, which is a significant improvement over LAUC-VF. It has

10
Seminar Title CHAPTER 2. LITERATURE SURVEY

low burst dropping probability and high bandwidth utilization. To date, it is the fastest
scheduling algorithm that can produce an efficient burst schedule.

11
Chapter 3

Problem Definition

3.1 Problem Definition


We know that LAUC-VF is one of the scheduling method to schedule the bursts
and to reduce burst dropping probability. the incoming data bursts. This burst dropping
is harmful for voice traffic i.e. does not take the space available on data channels into
consideration. The above problem of LAUC-VF makes it unsuitable for online voice,
video traffic which occupies most of the real time Internet traffic. Hence before dropping
the burst spaces left on data channels must be considered.

3.2 Solution

3.2.1 Reverse Scheduling


We introduce a new concept of reverse scheduling in OBS which can reduce the
burst dropping probability.

In Fig. 3.1 we can see that the data bursts are scheduled using general void
scheduling algorithm but in that situation new arriving data bursts 7 and 8 will be
dropped. Because there is no space for the new arriving data bursts. In comparison to

12
Seminar Title CHAPTER 3. PROBLEM DEFINITION

Scheduled bursts

1 2
D0

Data Channels
3
D1

4
D2

5 6
D3

Arriving Bursts

7 8

Time

Figure 3.1: General Void Scheduling Approach.

this if we observe Fig. 3.2 bursts 1 and 2 are scheduled on data channel D0 . Then data
burst 3 arrives it will first check for voids. We find the void on data channel D0 and
the arriving data burst can be fit on that void so we can schedule it on data channel D0
instead of scheduling it on data channel D1 , D2 or D3 . When data burst 4 will arrive it
will be scheduled on data channel D1 because there is no void to fit that burst on any
of the data channel. When data bursts 5 and 6 will arrive similarly they are scheduled
on data channel D2 . Now when new data bursts 7 and 8 will arrive instead of dropping
them they will be scheduled on data channel D3 .
In this way the arriving bursts which will going to dropped can be scheduled
without dropping. So dropping probability can be decreased by using reverse scheduling.
Automatically performance of scheduling will be increased.

Now we will see the pseudo code for LAUC-VF and Reverse Scheduling algo-
rithms as follows.

13
Seminar Title CHAPTER 3. PROBLEM DEFINITION

Scheduled bursts

1 2
D0

Data Channels
3
D1

4
D2

5 6
D3

Arriving Bursts

7 8

Time

Figure 3.2: Reverse Scheduling Approach.

Pseudo Code for LAUC-VF Scheduling Algorithm:


Consider the new burst is arrive at time t. The latest available time of

3.3 Timeline
Table 3.1 shows the time line chart of my MTech project.

3.4 System Requirements

14
Seminar Title CHAPTER 3. PROBLEM DEFINITION

Table 3.1: Timeline Chart.

Date Work Description


9/2009-10/2009 Study of OBS basics.
11/2009-12/2009 Issues and Qos in OBS.
1/2010-2/2010 Study of scheduling methods in OBS.
3/2010-4/2010 Study of reverse scheduling in OBS.
5/2010-7/2010 Implementation of reverse scheduling in OBS.
8/2010-9/2010 Implementation of reverse scheduling
to LAUC-VF, MIN-SV, MIN-EV in OBS.

15
Chapter 4

Conclusion

An OBS can handle the explosive demand of high speed networks. An OBS is not
only a cost-effective but also a viable solution for the next generation optical Internet.
We have presented scheduling algorithms like LAUC, LAUC-VF, MIN-SV and
MIN-EV. We introduced a new concept of reverse scheduling which can minimizes the
dropping probability of new arriving bursts and it can increases the performance of
scheduling algorithm.

16
Bibliography

[1] B.Mukharjee, Optical WDM Networks, Springer Publication, 2006.

[2] Tzvetelina Battestilli and Harry Perros, “An Introduction to Optical Burst Switch-
ing”, IEEE Optical Communication, Aug. 2003.

[3] Tzvetelina Battestilli, “Optical Burst Switching: A Survey”, Tech.rep. Tr-2002-10,


NC State University, Computer Science Department, July 2002.

[4] Y.Chen, C. Qiao and X. Yu, “An Optical Burst Switching: A New Area in Optical
Networking Research”, IEEE Netwoks, Vol.18, No.5, PP. 16-23,2005.

[5] Y.Luo and S.Wang, “An FDL-based QoS scheduling algorithm in OBS networks”,
IEEE, 2005.

[6] Jinhui Xu.,Chunming Qiao and Jikai Li, Guang Xu, “Efficient Channel Scheduling
Algorithms in Optical Switched Networks”, IEEE Infocom, 2003.

[7] S.K.Tan, G.Mohan and K.K.Chua, “Burst Rescheduling with Wavelength and Last-
hop FDL Reassignment in WDM Optical Burst Switching Networks”, IEEE, 2003.

[8] S.K.Tan, G.Mohan and K.K.Chua, “Algorithms for Burst Rescheduling in WDM
Optical Burst Switching Networks”, IEEE, Computer Networks 41(2003), 41-55,
2002.

[9] B. Kim, S. Lee, Y. hoi and Y. Cho, “An efficient preemption-based channel schedul-
ing algorithm for service differentiation in OBS networks”, Computer Communica-
tion , Vol. 29, 2006.

17

You might also like