Lahore University of Management and Sciences

1
Lahore University of Management and Sciences
2
Founding Vision

• To be a pre-eminent academic
• institution, serving as a catalyst
• for economic prosperity and
• social development
• Focus on development of world-class
• management talent

3
LUMS Overview

• Private university founded in 1985
• Founded by corporate sector (including Siemens)
• Modern 100 acre campus
• Located in Lahore, the cultural capital of
  Pakistan
• 2200 student body
• 1700 undergrad 500 graduate students
• 35 women
• 120 faculty members
• PhDs from Harvard, Stanford, MIT, Oxford
• Over 60 international publications in 2006

4
Undergraduate Programs

• Broad based education
• BSc (Honors) - 4 Years Degree
• Computer Science
• Computer Engineering
• Mathematics
• Economics
• Social Science
• Accounting and Finance
• Law
• Entry Requirements 12 years of education,
• and SAT I SAT II or LUMS Admission Test
• Average SAT I score 1300 / 1600

5
MS/PhD Programs

• Graduate programs in
• Computer Science (MS/PhD)
• Computer Engineering (MS/PhD)
• Mathematics (PhD)
• Business Studies (PhD)
• Economics (MS)
• Develop professionals with sound
• technical and management skills
• Produce highly qualified teachers
• Inculcate research culture

6
Quality Focus

• Best students inducted because of need blind
  admission policy
• Over Rs.40 million disbursed as financial aid in
  2005
• Excellent Graduate Placement Record
• 8 Rhodes Scholars in last 12 years
• 10 Fulbright Scholars in 2005
• Over 2500 alumni studying at top universities or
  employed in the top multinationals, local
  corporates, NGOs, Multilateral agencies,
  Government

7
Graduate Placement
8
Selected Employers
9
International Linkages Collaborations

• Student Exchange Programs
• Copenhagen Business School
• FH Joanneum Institute of Management
• Asian Institute of Technology
• University Utara Malaysia
• Indian School of Business
• Stockholm School of Business
• Linkages
• McGill University (CIDA)
• University of Essex
• Harvard
• MIT
• University of Chicago
• University of Sussex

10
Going Forward

• Refocusing of existing Schools
• School of Arts Sciences
• Suleman Dawood School of Business
• Two new schools and expansion of teaching
  programs
• School of Law and Policy
• School of Science and Engineering
• Doubling of student body to 4000
• 2400 undergraduate
• 1600 graduate
• Tripling of faculty strength to 360

11
Science Engineering Plans
Mathematics Computer Science Computer Engineering

• Modeled after US Research Universities
• 30 million already raised
• First batch of students enrolled by 2008
• By 2020 have 1200 undergrads, 800 grad
  students and 200 faculty

Physics Chemistry Biology Electrical
Engineering Industrial Engineering
Mechanical Engineering Material
Science Biological Engineering Chemical
Engineering

• Produce graduates who are
• Technically competent,
• creative hands-on
• Able to sell their ideas
• Socially responsible

12
(No Transcript)
13
Science/Eng Advisory Committee

• Robert Jaffe (MIT)
• John Kassakian (MIT)
• Richard Larson (MIT)
• Khalid Aziz (Stanford)
• Mir Imran (Incube)
• Hubertus Von Dewitz (Siemens CT)
• Obaid Siddiqi (TATA Inst. of Fundamental
  Research)
• Xiang-Sun Zhang (Chinese Academy of Sciences)
• Ashok Mittal (IIT Kanpur)
• Amos Nur (Stanford)
• Stephen Berry (Univ of Chicago)
• Hassan Ahmed (Sonus Networks)
• Zia Chishti (The Resource Group)

14
Vision

• To become an internationally
• acclaimed research university that
• serves society through excellence in
• education and research

15
Research Work in Department of Computer Science
and Engineering

• Lahore University of Management Sciences

16
Computer Science and Engineering Faculty

• Telecommunication and Networks
• Shahab Baqai, PhD Purdue Amjad Luna, PhD
  Georgia Tech
• Tariq Jadoon, PhD Strathclyde Zartash Uzmi, PhD
  Stanford
• Umar Saif, PhD Cambridge
• Video and Image Processing
• Nadeem A. Khan, PhD Eindhoven Shahid Masud,
  PhD Queens Belfast
• Sohaib A. Khan, PhD U of Central Florida
• Artificial Intelligence and Pattern Recognition
• Mian M Awais, PhD Imperial College Asim Karim,
  PhD Ohio State
• Electronics, Systems and Control
• Khurram Afridi, PhD MIT Anwar Khurshid, PhD
  Michigan
• M Jahangir Ikram, PhD Manchester
• Theoretical Computer Science and Software
  Engineering
• M Ashraf Iqbal, PhD UET Shafay Shamail, PhD
  Bath
• Arif Zaman, PhD Stanford Nabil Mustafa, PhD
  Duke

17
Research Areas and Funded Projects

• Artificial Intelligence, Pattern Recognition and
  Computer Vision
• Multi-Camera Multiple-Person Tracking and
  Surveillance
• 3D Reconstruction from Video (Non Rigid Structure
  from Motion)
• Freeway Incident Detection
• Automatic Multiple Party Lecture and Speech
  Summarization
• Analysis of Satellite Imagery for Rural
  Applications
• Telecommunication and Networks
• Restoration Routing in MPLS Networks (Funded by
  Cisco)
• Poor Mans Broad-band (Funded by Microsoft
  Research)
• Resource Management for Multimedia Services
• Fault Tolerant TCP Stack for Reliable Routing
• GMPLS Control for Ethernet (In collaboration with
  Siemens)
• Benchmarking of Pakistan Internet Backbone for
  VOIP
• Optical Network Management using GIS-based
  Information

18
Research Areas and Funded Projects

• Video and Image Processing
• Error Resilience of Multimedia Information
• Efficient H.264/MPEG4 Video Encoding
• Fast Correlation Methods for Image/Video
  Registration
• Automatic Target Recognition
• System Simulation and Design
• Predictive Model Development for Industrial
  Distillation Column
• Internet-based Environmental Watch System
• Simulation of GELS-based Ethernet Networks
• Automotive Systems
• Vision-based Smart Airbag
• Advanced Electrical System Design (MAESTrO)

Awarded BMW Scientific Award 1999
19
Evaluation of GELS Architecture
Evaluation of GELS Architecture
Dr. Zartash A Uzmi

• Problem
• Is it feasible to use GMPLS to control Ethernet
  traffic instead of traditional (STP) control?
• Approach
• Consider different networks and place random
  traffic on each of those networks
• Consider the normal as well as failed network
  conditions
• Compare amount of traffic placed in normal state
  and recovery time after a failure
• Solution
• Software tools for simulation of traditional
  Ethernet control and newly proposed control

• Project span one year (01/2006-12/2006)
• People involved
• Siemens Herbertus Dewitz (Liaison and advisory
  role) Johannes Riedl, Martin Nathensen, Kulkarni
• LUMS Zartash Afzal Uzmi (Technical and
  Supervisory role), three other researchers

• Future Directions
• Evaluate different mechanisms for replacing the
  traditional STP-based control with the new
  GMPLS-based control (GELS)

Collaboration with Siemens Possible Publication
in Globecom 07
20
Poor Mans Broadband Peer-to-Peer Dialup
Networking
Poor Mans Dialup Peer to Peer Dialup Networking

• Problem
• Provide better dialup internet access to rural
  areas than what is currently available
• Approach
• The bottleneck in internet access is not the
  last mile in developing countries rather,
  internet access is often much slower than the
  bandwidth of a modem
• Allow multiple users download internet content
  collaboratively, and use remaining bandwidth of
  their modem for sharing
• Solution
• Internet dialup and dialup underlay connections
  are interleaved to optimize available bandwidth
• Implementation is named Dittorrent, motivated
  by Bit-torrent

Dr. Umar Saif

• Future Directions
• Developing a comprehensive security architecture
  for this application
• Extend implementation to high-bandwidth WLL and
  WiMAX

Funded by Microsoft Research Published in ACM
SIGCOMM CCR 2007
21
Restoration Routing in MPLS Networks
Restoration Routing in MPLS Networks
Dr. Zartash A Uzmi

• Problem
• Develop framework to accommodate maximum
  restorable traffic in an MPLS ISP network
• Approach
• Allow maximum sharing of backup paths
• Identify optimal information to be propagated
• Identify optimal set of nodes to compute backup
  paths
• Solution
• A restoration routing scheme able to accommodate
  10 more traffic over an MPLS network compared to
  best known methodologies

Primary Path
Single Failure Assumption allows these backup
paths to share bandwidth
Backup Path

• Future Directions
• Create MPLS Traffic Engineering test bed with 26
  nodes for experimentation

Funded by CISCO Published in ICC 05 and ICC 07
22
Resource Management in Multimedia Services
Dr Shahab Baqai

• Problem
• Transport of multimedia over a network for
  play-out within user specified quality
  constraints
• Approach
• Formulate resource allocation as optimization
  problem with constraints and fairness
• Exploit tradeoff between available resources and
  minimum required play-out quality. Use fairness
  condition to spread any degradation evenly across
  active objects
• Solution
• Heuristic based dynamic resource allocation
  solution able to transmit better quality video
  over limited resources versus static assignment

Time process for I/O or Network channel assignment
Dynamic channel allocation for above process

• Future Directions
• Extension to resource reservation by intermediate
  nodes for end-to-end quality

Published in ACM Multimedia Systems Journal
23
Error Resilience of Multimedia Information
Error Resilience of Multimedia Information
Dr Shahab Baqai

• The Problem
• Incorporate error resilience in coding of
  multimedia information against correlated
  network losses.
• Approach
• Joint source and channel coding
• Interleaving in compressed bit-stream to preserve
  compression
• Compliance with standards
• Minimum change in encoding
• Decoding using standard decoder
• Solution
• Split and interleave related information in the
  frequency domain so that partial information is
  retrievable even under correlated network losses
• Different schemes for standard JPEG, JPEG 2000
  and MPEG standards optimized for application
  scenarios.

Triangular Interleaving to preserve partial high
frequency information in case of brusty network
loss for better edge reconstruction by error
concealment algorithms

• Applications
• Wireless multimedia applications
• Layered coding for progressive decoding display
  of multimedia data
• Graceful degradation of quality with increasing
  resource constraints
• Future Directions
• Error resilience and subsequent concealment of
  different video and audio coding standards
• Modeling of loss in different networks

Published in J. of Image Representation Video
Coding
24
Error Resilience of Multimedia Information
Wireless Network Research Initiative
Various faculty members

• Observation
• Developing countries are offering the fastest
  growth in mobile technology
• Largest WiMax deployment is being done in
  Pakistan (Motorola/Wateen)
• Research Strategy
• Address the networking issues for the 4th
  generation multi-hop broadband wireless networks
• Emphasize the components in the 802.16J
  architecture
• Work with the industry to translate the research
  into technology
• Icebreaking with 802.16J
• Relay functionality
• Access network frequency management
• Admission control
• Local and network mobility management
• QoS enhancements via relays
• Broadband multimedia services

• Future Plan
• Address ALL relevant research areas
• Physical Layer (Collaborative MIMO and OFDMA, ST
  diversity, Cognitive radio)
• RF Hardware (High efficiency linear power amps,
  tunable receivers, antenna arrays)
• MAC (opportunistic scheduling with fairness)
• Networking (Mobility management, handoff and
  roaming, Ambient networking)
• Services (VoIP, Mobile TV, High Res Games)

25
Thank You