Tài liệu Distributed Computing and Networking pptx

Lecture Notes in Computer Science 5935
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David Hutchison
Lancaster University, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Alfred Kobsa
University of California, Irvine, CA, USA
Friedemann Mattern
ETH Zurich, Switzerland
John C. Mitchell
Stanford University, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
Oscar Nierstrasz
University of Bern, Switzerland
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
TU Dortmund University, Germany
Madhu Sudan
Microsoft Research, Cambridge, MA, USA
Demetri Terzopoulos
University of California, Los Angeles, CA, USA

Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max-Planck Institute of Computer Science, Saarbruecken, Germany
Krishna Kant Sriram V. Pemmaraju
Krishna M. Sivalingam Jie Wu (Eds.)
Distributed
Computing
and Networking
11th International Conference, ICDCN 2010
Kolkata, India, January 3-6, 2010
Proceedings
13
Volume Editors
Krishna Kant
National Science Foundation
Arlington VA 22130, USA
E-mail:
Sriram V. Pemmaraju
The University of Iowa
Department of Computer Science
Iowa City, IA 52242-1419, USA
E-mail:
Krishna M. Sivalingam
Indian Institute of Technology (IIT)
Department of Computer Science and Engineering
Madras, Chennai 600036, India
E-mail:
Jie Wu
Temple University

Department of Computer and Information Science
Philadelphia, PA 119122, USA
E-mail:
Library of Congress Control Number: 2009941694
CR Subject Classification (1998): C.2, E.3, C.4, D.2.8, D.2, F.2, D.1.3, H.2.8, E.1
LNCS Sublibrary: SL 1 – Theoretical Computer Science and General Issues
ISSN
0302-9743
ISBN-10
3-642-11321-4 Springer Berlin Heidelberg New York
ISBN-13
978-3-642-11321-5 Springer Berlin Heidelberg New York
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting,
reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965,
in its current version, and permission for use must always be obtained from Springer. Violations are liable
to prosecution under the German Copyright Law.
springer.com
© Springer-Verlag Berlin Heidelberg 2010
Printed in Germany
Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India
Printed on acid-free paper SPIN: 12827701 06/3180 543210
Message from the General Chairs
As General Chairs it is our pleasure to welcome you to the proceedings of ICDCN
2010, the 11th International Conference on Distributed Computing and Network-
ing. This series of events started as the International Workshop on Distributed
Computing (IWDC) in the year 2000. In view of the growing number of papers
both in distributed computing and networking, and the natural synergy between
the two areas, in 2006 the workshop series assumed its current name. Since then

the conference has grown steadily in its reach and stature. The conference has at-
tracted quality submissions and top speakers annually in the areas of distributed
computing and networking from all over the world, thereby strengthening the
connection between research in India, which has been on the rise, and the rest
of the world. After a foray into Central India in the year 2009, this year the
conference returned to the city of Kolkata.
ICDCN continues to be a top-class conference due to the dedicated and tire-
less work put in by the volunteers who organize it each year. This year again, the
General Chairs were honored to work with a truly superb team who basically
left us with very little to do!
A good conference is known by its technical program, and this year’s program
was in the able hands of a Program Committee chaired by Krishna Sivalingam
and Jie Wu (Networking Track), and Krishna Kant and Sriram Pemmaraju (Dis-
tributed Computing track). There were 169 submissions, 96 to the networking
track and 73 to the distributed computing track. After a rigorous review pro-
cess, the committee selected 23 papers for the networking track, and 21 for the
distributed computing track (16 regular, 5 short).
We would like to thank the Keynote Chair, Sajal Das, for organizing an ex-
cellent invited program. This year’s keynote speakers are Prith Banerjee, Senior
VP of Research, and Director HP Labs, Prabhakar Raghavan, Head of Yahoo!
Labs, and Manish Gupta, Associate Director, IBM India Research Labs. The
Prof. A.K. Choudhury Memorial Lecture was delivered by Sartaj Sahni, Dis-
tinguished Professor and Chair of Computer Science, University of Florida and
Ashok Jhunjhunwala, the head of the Telecommunications and Computer Net-
works group at IIT Madras gave an invited lecture.
This year’s tutorial topics included: Vehicular Communications: Standards,
Protocols, Applications and Technical Challenges, by Rajeev Shorey; Informa-
tive Labeling Schemes, by Amos Korman; Middleware for Pervasive Computing,
by Jiannong Cao; Secure Distributed Computing, by C. Pandurangan; Next Gen-
eration of Transportation Systems, Distributed Computing, and Data Mining,

by Hillol Kargupta; Peer-to-Peer Storage Systems: Crowdsourcing the Storage
Cloud, by Anwitaman Datta. We thank the Tutorial Co-chairs, Gopal Panduran-
gan, Violet R. Syrotiuk, and Samiran Chattopadhyaya, for their efforts in putting
together this excellent tutorial program.
VI Message from the General Chairs
We would like to thank Sriram Pemmaraju who, as Publication Chair, dealt
with the many details of putting the proceedings together, and the Publicity
Chair, Arobinda Gupta, for doing a good job of getting the word out about the
event this year. Our Industry Chairs, Sanjoy Paul and Rajeev Shorey, helped
keep everyone’s feet on the ground! Our congratulations to them for organizing
a “cutting-edge” industry session with a set of esteemed panelists and speakers
from the booming IT sector in India. This year, ICDCN also hosted a PhD
Forum to encourage PhD students in India and abroad to present and discuss
their research with peers in their fields. Thanks to Indranil Sengupta and Mainak
Chatterjee for making this happen. Special thanks go out to the Organizing
Co-chairs Devadatta Sinha, University of Calcutta, Nabendu Chaki, University
of Calcutta, and Chandan Bhattacharyya, Techno India, Salt Lake, and to the
Finance Chair, Sanjit Setua, University of Calcutta, for having done a marvelous
job of taking care of all the nitty-gritty details of the conference organization.
The vision of the founders of this conference series, Sajal Das and Sukumar
Ghosh, continues to play a key role in the Steering Committee, and we hope
that under their leadership the conference will continue to grow and become
one of the major international research forums in distributed computing and
networking.
We thank all the authors and delegates for their participation. The success
of any conference is measured by the quality of the technical presentations, the
discussions that ensue, and the human networking that takes place. We expect
that, given the dedication and hard work of all the organizers, the conference
did not fall short on any of these measures.
January 2010 Anurag Kumar

Michel Raynal
Message from the Technical Program Chairs
Welcome to the proceedings of the 11th International Conference on Distributed
Computing and Networking (ICDCN 2010). ICDCN enters its second decade as
an important forum for disseminating the latest research results in distributed
computing and networking.
We received 169 submissions from all over the world, including Brazil, Canada,
China, France, Germany, Hong Kong, Iran, The Netherlands, Switzerland, and
the USA, besides India, the host country. The submissions were carefully read
and evaluated by the Program Committee, which consisted of 43 members for
the Networking Track and 34 members for the Distributed Computing Track,
with the additional help of external reviewers. The Program Committee selected
39 regular papers and 5 short papers for inclusion in the proceedings and presen-
tation at the conference. The resulting technical program covers a broad swath
of both distributed computing and networking. The networking track contains
papers on wireless, sensor, mobile, and ad-hoc networks and on network proto-
cols for scheduling, coverage, routing, etc., whereas the distributed computing
track contains papers on fault-tolerance, security, distributed algorithms, and
the theory of distributed systems.
While the technical program forms the core of the conference, this year’s
ICDCN was rich with many other exciting events. We were fortunate to have
several distinguished scientists as keynote speakers and we had a strong tuto-
rial program preceding the official start of the conference. In addition, we had a
fabulous industry session that has the potential of strengthening research ties be-
tween academics and the industry. Finally, this year ICDCN hosted a PhD forum
whose aim was to connect student researchers with peers as well as experienced
researchers.
We thank all those who submitted a paper to ICDCN 2010 for their interest.
We thank the Program Committee members and external reviewers for their
careful reviews despite a tight schedule.

January 2010 Krishna Kant
Sriram V. Pemmaraju
Krishna M. Sivalingam
Jie Wu
Organization
ICDCN 2010 was organized by the University of Calcutta, Department of Com-
puter Science and Engineering in collaboration with the Techno India Group,
Salt Lake.
General Chairs
Michel Raynal Institut de Recherche en Informatique et
Syst`emes Al´eatoires (IRISA)
Anurag Kumar Indian Institute of Science (IISc), Bangalore
Program Chairs: Networking Track
Krishna M. Sivalingam Indian Institute of Technology (IIT) Madras
Jie Wu Temple University
Program Chairs: Distributed Computing Track
Krishna Kant Intel and National Science Foundation (NSF)
Sriram V. Pemmaraju The University of Iowa
Keynote Chair
Sajal K. Das University of Texas at Arlington and National
Science Foundation (NSF)
Tutorial Chairs
Gopal Pandurangan Purdue University
Violet R. Syrotiuk Arizona State University
Samiran Chattopadhyaya Jadavpur University, Kolkata, India
Publication Chair
Sriram V. Pemmaraju The University of Iowa
Publicity Chair
Arobinda Gupta Indian Institute of Technology, Kharagpur
X Organization

Industry Chairs
Sanjoy Paul Infosys, India
Rajeev Shorey NIIT University, India
Finance Chair
Sanjit Setua University of Calcutta
Organizing Committee Chairs
Devadatta Sinha University of Calcutta
Nabendu Chaki University of Calcutta
Chandan Bhattacharyya Techno India, Salt Lake
Steering Committee
Pradip K. Das Mody Institute of Technology and Science,
Jaipur, India
Sajal K. Das The University of Texas at Arlington, USA and
National Science Foundation (NSF) (Co-chair)
Vijay Garg IBM India and Univ. of Texas at Austin, USA
Sukumar Ghosh University of Iowa, USA (Co-chair)
Anurag Kumar Indian Institute of Science, Bangalore, India
David Peleg Weizman Institute of Science, Israel
Michel Raynal Institut de Recherche en Informatique et
Syst`emes Al´eatoires (IRISA), France
Indranil Sengupta Indian Inst. of Tech., Kharagpur, India
Bhabani Sinha Indian Statistical Institute, Kolkata, India
Program Committee: Networking Track
Alessandro Puiatti SUPSI-DTI, Switzerland
Anil Vullikanti Virginia Tech (VPI), USA
Arzad Kherani GM India Science Lab, India
Biplab Sikdar RPI, USA
David Kotz Dartmouth College, USA
David Simplot-Ryl INRIA Lille, France
Deep Medhi University of Missouri - Kansas City, USA

Deva Seetharam IBM, India
Falko Dressler University of Erlangen, Germany
Gaurav Raina IIT Madras, India
Guohong Cao Pennsylvania State University, USA
Organization XI
Imad Jawhar UAE University, UAE
Joy Kuri Indian Institute of Science, Bangalore, India
Koushik Kar RPI, USA
Lin Gu Hong Kong Univ. of Science and Tech., China
Mainak Chatterjee University of Central Florida, USA
Manimaran Govindarasu Iowa State University, USA
Manjunath D. IIT Bombay, India
Marco Conti IIT-CNR, Italy
Marimuthu Palaniswami University of Melbourne, Australia
Matt Mutka Michigan State University, USA
Mingming Lu Central South University, China
Prashant Krishnamurthy University of Pittsburgh, USA
Prasun Sinha Ohio State University, USA
Qin Yang HIT ShenZhen Graduate School, China
Radim Bartos University of New Hampshire, USA
Rajarshi Roy IIT Kharagpur, India
Rajeev Rastogi Yahoo Research, India
Rajesh Sundaresan Indian Institute of Science, Bangalore, India
Sanglu Lu Nanjing University, China
Sanjay Bose IIT Guwahati, India
Sanjay Jha University of New South Wales, Australia
Santosh Kumar University of Memphis, USA
Saswati Sarkar University of Pennsylvania, USA
Shivkumar Kalyanaraman IBM India and RPI, USA
Srihari Nelakudit University of South Carolina, USA

Umamaheswari Devi IBM, India
Vikram Srinivasan Alcatel-Lucent Bell Labs, India
Wei Lou Hong Kong Polytechnic University, Hong Kong
Wenjing Lou Worcester Polytechnic Institute, USA
Wenye Wang North Carolina State University, USA
Wonjun Lee Korea University, Seoul, Korea
Xu Li University of Ottawa, Canada
Program Committee: Distributed Computing Track
Ajay Kshemkalyani University of Illinois at Chicago, USA
Amos Korman CNRS, France
Arobinda Gupta IIT Kharagpur, India
Bruhadeshwar Bezawada IIIT Hyderabad, India
Gopal Pandurangan Purdue University, USA
Gregory Chokcler IBM Research, Israel
Haifeng Yu National University of Singapore. Singapore
Indranil Gupta Univ. of Illinois at Urbana-Champaign, USA
Jiannong Cao HongKong Polytech University, China
XII Organization
Kishore Kothapalli IIIT Hyderabad, India
Krishnamurthy Vidyasankar Memorial University of Newfoundland, Canada
Maria Potop-Butucaru University Pierre and Marie Curie (Paris 6),
France
Mark Tuttle Intel, USA
Neeraj Mittal The University of Texas at Dallas, USA
Philippas Tsigas Chalmers University, Sweden
Pierre Fraigniaud CNRS, France
Prasad Jayanti Dartmouth College, USA
Rajkumar Buyya The University of Melbourne, Australia
Roger Wattenhofer ETH Zurich, Switzerland
Rong Zheng University of Houston, USA

Sanjay Ranka University of Florida, USA
Sanjoy Paul InfoSys Technologies, India
Sebastien Tixeuil LIP6 & INRIA Grand Large, France
Sergio Rajsbaum UNAM, Mexico
Shlomi Dolev Ben-Gurion University, Israel
Soma Chaudhuri Iowa State University, USA
Stephan Eidenbenz Los Alamos National Labs, USA
Sukumar Ghosh The University of Iowa, USA
Tao Xie San Diego State University, USA
Thomas Moscibroda Microsoft Research, USA
Umakishore Ramachandran Georgia Tech, USA
VijayGarg UniversityofTexas,USA
Winston Seah Institute for Infocomm Research, Singapore
Yehuda Afek Tel Aviv University, Israel
Additional Referees: Networking Track
Amin Ali
Swapnil Bhatia
Debojyoti Bhattacharya
Chiara Boldrini
Swastik Brahma
Raffaele Bruno
Ning Cao
Surendar Chandra
Saptarshi Debroy
S. Sharmila Deva Selvi
Juergen Eckert
Wei Gao
Chase Gray
Santanu Guha
James Joshi

Aditya Karnik
R.M. Karthik
Kim Kyunghwi
Ming Li
Qinghu Li
Tobias Limmer
Changlei Liu
Salahuddin Masum
Somnath Mitra
Skanda Muthaiah
Andrea Passarella
Chuan Qin
Venkatesh R.
Krishna Ramachandran
Glenn Robertson
Naveen Santhapuri
Mukundan
Venkataraman
T. Venkatesh
S. Sree Vivek
Guojun Wang
Wenjing Wang
Zhenyu Yang
Eiko Yoneki
Shucheng Yu
Organization XIII
Additional Referees: Distributed Computing Track
Yaniv Altshuler
Bharath
Balasubramanian

Sumit Bose
Hana Chockler
Peter Chong
Jorge Cobb
Reetuparna Das
Atish Das Sarma
Sergei Frenkel
Nurit Galoz
David Hilley
Shiva Kasiviswanathan
Idit Keidar
Maleq Khan
Rajnish Kumar
Dave Lillethun
Thomas Locher
Remo Meier
Dushmanta Mohapatra
Yoram Moses
Rajarathnam Nallusamy
Danupon Nanongkai
Gal-Oz Nurit
Dmitri Perelman
Olivier Peres
Ravi Prakash
Frankel Sergey
Junsuk Shin
Benjamin Sigg
Vishak Sivakumar
Jasmin Smula
Arun Somasundara

Christian Sommer
Hwee-Pink Tan
Amitabh Trehan
Zigi Walter
Table of Contents
Keynotes
An Intelligent IT Infrastructure for the Future 1
Prith Banerjee
Heavy Tails and Models for the Web and Social Networks 2
Prabhakar Raghavan
Data Structures and Algorithms for Packet Forwarding and
Classification: Prof. A.K. Choudhury Memorial Lecture 3
Sartaj Sahni
Spoken Web: A Parallel Web for the Masses: Industry Keynote 4
Manish Gupta
India’s Mobile Revolution and the Unfinished Tasks: Invited Lecture 5
Ashok Jhunjhunwala
Network Protocols and Applications
Scheduling in Multi-Channel Wireless Networks 6
Vartika Bhandari and Nitin H. Vaidya
Email Shape Analysis 18
Paul Sroufe, Santi Phithakkitnukoon, Ram Dantu, and
Jo˜ao Cangussu
Maintaining Safety in Interdomain Routing with Hierarchical
Path-Categories 30
Jorge A. Cobb
Fault-tolerance and Security
On Communication Complexity of Secure Message Transmission in
Directed Networks 42
Arpita Patra, Ashish Choudhary, and C. Pandu Rangan

On Composability of Reliable Unicast and Broadcast 54
Anuj Gupta, Sandeep Hans, Kannan Srinathan, and
C. Pandu Rangan
A Leader-Free Byzantine Consensus Algorithm 67
Fatemeh Borran and Andr´e Schiper
XVI Table of Contents
Authenticated Byzantine Generals in Dual Failure Model 79
Anuj Gupta, Prasant Gopal, Piyush Bansal, and Kannan Srinathan
Sensor Networks
Mission-Oriented k-Coverage in Mobile Wireless Sensor Networks 92
Habib M. Ammari and Sajal K. Das
Lessons from the Sparse Sensor Network Deployment in Rural India 104
T.V. Prabhakar, H.S. Jamadagni, Amar Sahu, and
R. Venkatesha Prasad
A New Architecture for Hierarchical Sensor Networks with Mobile Data
Collectors 116
Ataul Bari, Ying Chen, Arunita Jaekel, and Subir Bandyopadhyay
Stability Analysis of Multi-hop Routing in Sensor Networks with
Mobile Sinks 128
Jayanthi Rao and Subir Biswas
Distributed Algorithms and Optimization
Optimizing Distributed Computing Workflows in Heterogeneous
Network Environments 142
Yi Gu and Qishi Wu
Radio Network Distributed Algorithms in the Unknown Neighborhood
Model 155
Bilel Derbel and El-Ghazali Talbi
Probabilistic Self-stabilizing Vertex Coloring in Unidirectional
Anonymous Networks 167
Samuel Bernard, St´ephane Devismes, Katy Paroux,

Maria Potop-Butucaru, and S´ebastien Tixeuil
A Token-Based Solution to the Group Mutual l-Exclusion Problem in
Message Passing Distributed Systems (Short Paper) 178
Abhishek Swaroop and Awadhesh Kumar Singh
Peer-to-Peer Networks and Network Tracing
The Weak Network Tracing Problem 184
H.B. Acharya and M.G. Gouda
Poisoning the Kad Network 195
Thomas Locher, David Mysicka, Stefan Schmid, and
Roger Wattenhofer
Table of Contents XVII
Credit Reputation Propagation: A Strategy to Curb Free-Riding in a
Large BitTorrent Swarm 207
Suman Paul, Subrata Nandi, and Ajit Pal
Formal Understanding of the Emergence of Superpeer Networks: A
Complex Network Approach 219
Bivas Mitra, Abhishek Kumar Dubey, Sujoy Ghose, and
Niloy Ganguly
Parallel and Distributed Systems
Parallelization of the Lanczos Algorithm on Multi-core Platforms 231
Souvik Bhattacherjee and Abhijit Das
Supporting Malleability in Parallel Architectures with Dynamic
CPUSETs Mapping and Dynamic MPI 242
M´arcia C. Cera, Yiannis Georgiou, Olivier Richard,
Nicolas Maillard, and Philippe O.A. Navaux
Impact of Object Operations and Relationships on Concurrency
Control in DOOS (Short Paper) 258
V. Geetha and Niladhuri Sreenath
Causal Cycle Based Communication Pattern Matching (Short Paper) 265
Himadri Sekhar Paul

Wireless Networks
Channel Assignment in Virtual Cut-through Switching Based Wireless
Mesh Networks 271
Dola Saha, Aveek Dutta, Dirk Grunwald, and Douglas Sicker
Efficient Multi-hop Broadcasting in Wireless Networks Using k-Shortest
Path Pruning 283
Michael Q. Rieck and Subhankar Dhar
Bandwidth Provisioning in Infrastructure-Based Wireless Networks
Employing Directional Antennas 295
Shiva Kasiviswanathan, Bo Zhao, Sudarshan Vasudevan, and
Bhuvan Urgaonkar
ROTIO+: A Modified ROTIO for Nested Network Mobility 307
Ansuman Sircar, Bhaskar Sardar, and Debashis Saha
Applications of Distributed Systems
VirtualConnection: Opportunistic Networking for Web on Demand 323
Lateef Yusuf and Umakishore Ramachandran
XVIII Table of Contents
Video Surveillance with PTZ Cameras: The Problem of Maximizing
Effective Monitoring Time 341
Satyajit Banerjee, Atish Datta Chowdhury, and Subhas Kumar Ghosh
DisClus: A Distributed Clustering Technique over High Resolution
Satellite Data 353
Sauravjyoti Sarmah and Dhruba Kumar Bhattacharyya
Performance Evaluation of a Wormhole-Routed Algorithm for Irregular
Mesh NoC Interconnect 365
Arshin Rezazadeh, Ladan Momeni, and Mahmood Fathy
Optical, Cellular and Mobile Ad Hoc Networks
Dynamic Multipath Bandwidth Provisioning with Jitter, Throughput,
SLA Constraints in MPLS over WDM Network 376
Palash Dey, Arkadeep Kundu, Mrinal K. Naskar,

Amitava Mukherjee, and Mita Nasipuri
Path Protection in Translucent WDM Optical Networks 392
Q. Rahman, Subir Bandyopadhyay, Ataul Bari, Arunita Jaekel, and
Y.P. Aneja
Post Deployment Planning of 3G Cellular Networks through Dual
Homing of NodeBs 404
Samir K. Sadhukhan, Swarup Mandal, Partha Bhaumik, and
Debashis Saha
K-Directory Community: Reliable Service Discovery in MANET 420
Vaskar Raychoudhury, Jiannong Cao, Weigang Wu, Yi Lai,
Canfeng Chen, and Jian Ma
Theory of Distributed Systems
An Online, Derivative-Free Optimization Approach to Auto-tuning of
Computing Systems 434
Sudheer Poojary, Ramya Raghavendra, and D. Manjunath
Consistency-Driven Probabilistic Quorum System Construction for
Improving Operation Availability 446
Kinga Kiss Iakab, Christian Storm, and Oliver Theel
Hamiltonicity of a General OTIS Network (Short Paper) 459
Nagendra Kumar, Rajeev Kumar, Dheeresh K. Mallick, and
Prasanta K. Jana
Specifying Fault-Tolerance Using Split Precondition Logic
(Short Paper) 466
Awadhesh Kumar Singh and Anup Kumar Bandyopadhyay
Table of Contents XIX
Network Protocols
Fast BGP Convergence Following Link/Router Failure 473
Swapan Kumar Ray and Susmit Shannigrahi
On Using Network Tomography for Overlay Availability 485
Umesh Bellur and Mahak Patidar

QoSBR: A Quality Based Routing Protocol for Wireless Mesh
Networks 497
Amitangshu Pal, Sandeep Adimadhyam, and Asis Nasipuri
An ACO Based Approach for Detection of an Optimal Attack Path in
a Dynamic Environment 509
Nirnay Ghosh, Saurav Nanda, and S.K. Ghosh
Author Index 521
An Intelligent IT Infrastructure for the Future
Prith Banerjee
HP Labs, Hewlett Packard Corporation

Abstract. The proliferation of new modes of communication and collaboration
has resulted in an explosion of digital information. To turn this challenge into
an opportunity, the IT industry will have to develop novel ways to acquire, store,
process, and deliver information to customers - wherever, however, and whenever
they need it. An ”Intelligent IT Infrastructure,” which can deliver extremely high
performance, adaptability and security - will be the backbone of these develop-
ments. At HP Labs, the central research arm for Hewlett Packard, we are taking
a multidisciplinary approach to this problem by spanning four areas: computing,
storage, networking and nanotechnology. We are working on the design of an
exascale data center that will provide 1000X performance while enhancing avail-
ability, manageability and reliability and reducing the power and cooling costs.
We are working on helping the transition to effective parallel and distributed
computing by developing the software tools to allow application developers to
harness parallelism at various levels. We are building a cloud-scale, intelligent
storage system that is massively scalable, resilient to failures, self-managed and
enterprise-grade. We are designing an open, programmable wired and wireless
network platform that will make the introduction of new features quick, easy and
cost-effective. Finally, we are making fundamental breakthroughs in nanotech-
nology - memristors, photonic interconnects, and sensors - that will revolutionize

the way data is collected, stored and transmitted. To support the design of such an
intelligent IT infrastructure, we will have to develop sophisticated system-level
design automation tools that will tradeoff system-level performance, power, cost
and efficiency.
K. Kant et al. (Eds.): ICDCN 2010, LNCS 5935, p. 1, 2010.
c
 Springer-Verlag Berlin Heidelberg 2010
Heavy Tails and Models for the Web and Social
Networks
Prabhakar Raghavan
Yahoo! Labs

Abstract. The literature is rich with (re)discoveries of power law phenomena;
this is especially true of observations of link and traffic behavior on the Web. We
survey the origins of these phenomena and several (yet incomplete) attempts to
model them, including our recent work on the compressibility of the Web graph
and social networks. We then present a number of open problems in Web research
arising from these observations.
K. Kant et al. (Eds.): ICDCN 2010, LNCS 5935, p. 2, 2010.
c
 Springer-Verlag Berlin Heidelberg 2010
Data Structures and Algorithms for Packet Forwarding
and Classification:
Prof. A.K. Choudhury Memorial Lecture
Sartaj Sahni
Computer and Information Science and Engineering Department
University of Florida

Abstract. Packet forwarding and classification at Internet speed is a challenging
task. We review the data structures that have been proposed for the forwarding

and classification of Internet packets. Data structures for both one-dimensional
and multidimensional classification as well as for static and dynamic rule tables
are reviewed. Sample structures include multibit one- and two-dimensional tries
and hybrid shape shifting tries. Hardware assisted solutions such as Ternary Con-
tent Addressable Memories also are reviewed.
K. Kant et al. (Eds.): ICDCN 2010, LNCS 5935, p. 3, 2010.
c
 Springer-Verlag Berlin Heidelberg 2010
Spoken Web: A Parallel Web for the Masses:
Industry Keynote
Manish Gupta
IBM Research, India

Abstract. In India and several other countries, the number of mobile phone sub-
scribers far exceeds the number of personal computer users, and continues to
grow at a much faster pace (it has already crossed the 450 million mark in India).
We will present Spoken Web, an attempt to create a new world wide web, acces-
sible over the telephone network, for the masses in these countries. The Spoken
Web is based on the concepts of Hyperspeech and Hyperspeech Transfer Protocol
that allow creation of ”VoiceSites” and traversal of ”VoiceLinks”. We describe a
simple voice-driven application, which allows people, without any information
technology background, to create, host, and access such VoiceSites, and traverse
VoiceLinks, using a voice interface over the telephone. We present our experi-
ence from pilots conducted in villages in Andhra Pradesh and Gujarat. These
pilots demonstrate the ease with which a semi-literate and non-IT savvy popu-
lation can create VoiceSites with locally relevant content, including schedule of
education/training classes, agicultural information, and professional services, and
their strong interest in accessing this information over the telephone network. We
describe several outstanding challenges and opportunities in creating and using
a Spoken Web for facilitating exchange of information and conducting business

transactions.
K. Kant et al. (Eds.): ICDCN 2010, LNCS 5935, p. 4, 2010.
c
 Springer-Verlag Berlin Heidelberg 2010
India’s Mobile Revolution and the Unfinished Tasks:
Invited Lecture
Ashok Jhunjhunwala
IIT Madras, Chennai, India

Abstract. India has made great strides in use of Mobile telephones in recent
years. Adding over 10 million phones a month, it is the fastest growing market
today. The cell-phones are quickly reaching the deepest parts of the nation and
serving the poorest people. The talk will examine what made this possible. It will
also focus on what the unfinished telecom tasks for India are. It will examine
what India is doing in terms of providing Broadband wireless connectivity to its
people; what it is doing towards R&D and technology development in the county;
and how it aims at building global telecom manufacturing and telecom operation
companies in India.
K. Kant et al. (Eds.): ICDCN 2010, LNCS 5935, p. 5, 2010.
c
 Springer-Verlag Berlin Heidelberg 2010
Scheduling in Multi-Channel Wireless Networks

Vartika Bhandari

and Nitin H. Vaidya
University of Illinois at Urbana-Champaign, USA
,
Abstract. The availability of multiple orthogonal channels in a wireless net-
work can lead to substantial performance improvement by alleviating contention

and interference. However, this also gives rise to non-trivial channel coordina-
tion issues. The situation is exacerbated by variability in the achievable data-
rates across channels and links. Thus, scheduling in such networks may require
substantial information-exchange and lead to non-negligible overhead. This pro-
vides a strong motivation for the study of scheduling algorithms that can operate
with limited information while still providing acceptable worst-case performance
guarantees. In this paper, we make an effort in this direction by examining the
scheduling implications of multiple channels and heterogeneity in channel-rates.
We establish lower bounds on the performance of a class of maximal sched-
ulers. We first demonstrate that when the underlying scheduling mechanism is
“imperfect”, the presence of multiple orthogonal channels can help alleviate the
detrimental impact of the imperfect scheduler, and yield a significantly better
efficiency-ratio in a wide range of network topologies. We then establish perfor-
mance bounds for a scheduler that can achieve a good efficiency-ratio in the pres-
ence of channels with heterogeneous rates without requiring explicit exchange
of queue-information. Our results indicate that it may be possible to achieve a
desirable trade-off between performance and information.
1 Introduction
Appropriate scheduling policies are of utmost importance in achieving good throughput
characteristics in a wireless network. The seminal work of Tassiulas and Ephremides
yielded a throughput-optimal scheduler, which can schedule all “feasible” traffic flows
without resulting in unbounded queues [8]. However, such an optimal scheduler is diffi-
cult to implement in practice. Hence, various imperfect scheduling strategies that trade-
off throughput for simplicity have been proposed in [5,9,10,7], amongst others.
The availability of multiple orthogonalchannels in a wireless network can potentially
lead to substantial performance improvement by alleviating contention and interference.
However, this also gives rise to non-trivial channel coordination issues. The situation is
exacerbated by variability in the achievabledata-ratesacrosschannels and links. Comput-
ing an optimal schedule, even in a single-channel network, is almost always intractable,
due to the need for global information, as well as the computational complexity. How-

ever, imperfect schedulers requiring limited local information can typically be designed,

This research was supported in part by NSF grant CNS 06-27074, US Army Research Office
grant W911NF-05-1-0246, and a Vodafone Graduate Fellowship.

Vartika Bhandari is now with Google Inc.
K. Kant et al. (Eds.): ICDCN 2010, LNCS 5935, pp. 6–17, 2010.
c
 Springer-Verlag Berlin Heidelberg 2010
Scheduling in Multi-Channel Wireless Networks 7
which provide acceptable worst-case (and typically much better average case) perfor-
mance degradation compared to the optimal. In a multi-channel network, the local in-
formation exchange required by even an imperfect scheduler can be quite prohibitive as
information may be needed on a per-channelbasis. For instance, Lin and Rasool [4] have
described a scheduling algorithm for multi-channel multi-radio wireless networks that
requires information about per-channel queues at all interfering links.
This provides a strong motivation for the study of scheduling algorithms that can
operate with limited information, while still providing acceptable worst-case perfor-
mance guarantees. In this paper, we make an effort in this direction, by examining the
scheduling implications of multiple channels, and heterogeneity in channel-rates. We
establish lower bounds on performance of a class of maximal schedulers, and describe
some schedulers that require limited information-exchangebetween nodes. Some of the
bounds presented here improve on bounds developed in past work [4].
We begin by analyzing the performance of a centralized greedy maximal scheduler.
A lower bound for this scheduler was established in [4]. However, in a large variety of
network topologies, the lower bound can be quite loose. Thus is particularly true for
multi-channel networks with single interface nodes. We establish an alternative bound
that is tighter in a range of topologies. Our results indicate that when the underlying
scheduling mechanism is imperfect, the presence of multiple orthogonal channels can
help alleviate the impact of the imperfect scheduler, and yield a significantly better

efficiency-ratio in a wide range of scenarios
We then consider the possibility of achieving efficiency-ratio comparable to the cen-
tralized greedy maximal scheduler using a simpler scheduler that works with limited
information. We establish results for a class of maximal schedulers coupled with local
queue-loading rules that do not require queue-information from interfering nodes.
2 Preliminaries
We consider a multi-hop wireless network. For simplicity, we largely limit our discus-
sion to nodes equipped with a single half-duplex radio-interface capable of tuning to
any one available channel at any given time. All interfaces in the network have iden-
tical capabilities, and may switch between the available channels if desired. Many of
the presented results can also be used to obtain results for the case when each node is
equipped with multiple interfaces; we briefly discuss this issue.
The wireless network is viewed as a directed graph, with each directed link in the
graph representing an available communication link. We model interference using a
conflict relation between links. Two links are said to conflict with each other if it is only
feasible to schedule one of the links on a certain channel at any given time. The conflict
relation is assumed to be symmetric. The conflict-based interference model provides
a tractable approximation of reality – while it does not capture the wireless channel
precisely, it is more amenable to analysis. Such conflict-based interference models have
been used frequently in the past work (e.g., [11,4]).
Time is assumed to be slotted with a slot duration of 1 unit time (i.e., we use slot
duration as the time unit). In each time slot, the scheduler determines which links should
transmit in that time slots, as well as the channel to be used for each such transmission.
8 V. Bhandari and N.H. Vaidya
We now introduce some notation and terminology.
The network is viewed as a collection of directed links, where each link is a pair of
nodes that are capable of direct communication with non-zero rate.
– L denotes the set of directed links in the network.
– C is the set of all available orthogonalchannels. Thus, |C| is the number of available
channels.

– We say that a scheduler schedules link-channel pair (l, c) if it schedules link l for
transmission on channel c.
– r
c
l
denotes the rate achievable on link l by operating link l on channel c, provided
that no conflicting link is also scheduled on channel c. For simplicity, we assume
that r
c
l
> 0foralll ∈ L and c ∈ C.
1
The rates r
c
l
do not vary with time. We also
define the terms: r
max
= max
l∈L,c∈C
r
c
l
,andr
min
= min
l∈L,c∈C
r
c
l

. When two conflicting links
are scheduled simultaneously on the same channel, both achieve rate 0.
– β
s
denotes the self-skew-ratio, defined as the minimum ratio between rates support-
able over different channels on a single link. Therefore, for any two channels c and
d, and any link l,wehave
r
d
l
r
c
l
≥ β
s
. Note that 0 < β
s
≤ 1.
– β
c
denotes the cross-skew-ratio, defined as the minimum ratio between rates sup-
portable over the same channel on different links. Therefore, for any channel c,and
any two links l and l

:
r
c
l

r

c
l
≥ β
c
. Note that 0 < β
c
≤ 1.
Let r
l
= max
c∈C
r
c
l
.Letσ
s
= min
l∈L
∑
c∈C
r
c
l
r
l
. Note that σ
s
≥ 1 +β
s
(|C|−1). Moreover, in

typical scenarios, σ
s
will be expected to be much larger than this worst-case bound.
σ
s
is largest when β
s
= 1, in which case σ
s
= |C|.
– b(l) and e(l), respectively, denotes the nodes at the two endpoints of a link. In
particular, link l is directed from node b(l) to node e(l).
– E(b(l))and E(e(l))denote the set of links incident on nodes b(l) and e(l), respec-
tively. Thus, the links in E(b(l)) and E(e(l)) share an endpoint with link l.Since
we focus on single-interface nodes, this implies that if link l is scheduled in a cer-
tain time slot, no other link in E(b(l)) or E(e(l)) can be scheduled at the same
time. We refer to this as an interface conflict.LetA(l)=E(b(l)) ∪ E(e(l)).Note
that l ∈ A(l). Links in A(l) are said to be adjacent to link l. Links that have an
interface conflict with link l are those that belong to E(b(l)) ∪ E(e(l)) \{l}.Let
A
max
= max
l
|A(l)|.
–I(l) denotes the set of links that conflict with link l when scheduled on the same
channel. I(l) may include links that also have an interface-conflict with link l.By
convention, l is considered included in I(l). The subset of I(l) comprising interfer-
ing links that are not adjacent to l is denoted by I

(l), i.e., I


(l)=I(l) \ A(l).Let
I
max
= max
l
|I

(l)|.
– K
l
denotes the maximum number of non-adjacent links in I

(l) that can be sched-
uled on a given channel simultaneously if l is notscheduled on that channel. K
l
(|C|)
1
Though we assume that r
c
l
> 0foralll, c, the results can be generalized very easily to handle
the case where r
c
l
= 0 for some link-channel pairs.
Scheduling in Multi-Channel Wireless Networks 9
denotes the maximum number of non-adjacent links in I

(l) that can be scheduled

simultaneously using any of the |C| channels (without conflicts) if l is not sched-
uled for transmission. Note that here we exclude links that have an interface conflict
with l.
– K is the largest value of K
l
over all links l, i.e., K = max
l
K
l
. K
|C|
is the largest value
of K
l
(|C|) over all links l, i.e., K
|C|
= max
l
K
l
(|C|).LetI
max
= max
l
|I

(l)|. It is not
hard to see that for single-interface nodes:
K ≤ K
|C|

≤ min{K|C |, I
max
} (1)
We remark that the term K as used by us is similar, but not exactly the same as
the term K used in [4]. In [4], K denotes the largest number of links that may be
scheduled simultaneously if some link l is not scheduled, including links adjacent
to l. We exclude the adjacent links in our definition of K. Throughout this text, we
will refer to the quantity defined in [4] as κ instead of K.
– Let γ
l
be 0 if there are no other links adjacent to l at either endpoint of l, 1 if there
are other adjacent links at only one endpoint, and 2 if there are other adjacent links
at both endpoints.
– γ is the largest value of γ
l
over all links l, i.e., γ = max
l
γ
l
.
– Load vector: We consider single-hop traffic, i.e., any traffic that originates at a node
is destined for a next-hop node, and is transmitted over the link between the two
nodes. Under this assumption, all the traffic that must traverse a given link can be
treated as a single flow.
The traffic arrival process for link l is denoted by {λ(t)} . The arrivals in each
slot t are assumed i.i.d. with average λ
l
. The average load on the network is denoted
by load vector
−→

λ =[λ
1
, λ
2
, , λ
|L|
],whereλ
l
denotes the arrival rate for the flow
on link l. λ
l
may possibly be 0 for some links l.
– Queues: The packets generated by each flow are first added to a queue maintained
at the source node. Depending on the algorithm, there could be a single queue for
each link, or a queue for each (link, channel) pair.
– Stability: The system of queues in the network is said to be stable if, for all queues
Q in the network, the following is true [2]:
lim
t→∞
sup
1
t
t
∑
τ=1
E[q(τ)] < ∞
(2)
where q(τ) denotes the backlog in queue Q at time τ.
– Feasible load vector: In each time slot, the scheduler used in the network deter-
mines which links should transmit and on which channel (recall that each link is a

directed link, with a transmitter and a receiver). In different time slots, the sched-
uler may schedule a different set of links for transmission. A load vector is said to
be feasible, if there exists a scheduler that can schedule transmissions to achieve
stability (as defined above), when using that load vector.
– Link rate vector: Depending on the schedule chosen in a given slot by the sched-
uler, each link l will have a certain transmission rate. For instance, using our nota-
tion above, if link l is scheduled to transmit on channel c, it will have rate r
c
l
(we