Top Banner
CSCI 599: Science of CSCI 599: Science of Design for Continuous Design for Continuous Media Applications Media Applications Shahram Ghandeharizadeh Shahram Ghandeharizadeh Associate Professor Associate Professor Computer Science Department Computer Science Department University of Southern California University of Southern California
34

CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

Dec 24, 2015

Download

Documents

Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

CSCI 599: Science of Design CSCI 599: Science of Design for Continuous Media for Continuous Media

ApplicationsApplications

Shahram GhandeharizadehShahram GhandeharizadehAssociate ProfessorAssociate ProfessorComputer Science DepartmentComputer Science DepartmentUniversity of Southern CaliforniaUniversity of Southern California

Page 2: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

OutlineOutline H2O & C2P2 frameworksH2O & C2P2 frameworks Uses and challengesUses and challenges What makes these devices feasible?What makes these devices feasible? This seminar’s objectiveThis seminar’s objective How am I graded?How am I graded?

Page 3: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

Home-to-Home Online (H2O) devices collaborate to Home-to-Home Online (H2O) devices collaborate to deliver continuous media:deliver continuous media:

A H2O device is a wireless device with a powerful A H2O device is a wireless device with a powerful processor and abundant amount of storage.processor and abundant amount of storage.

H2O FrameworkH2O Framework

S. Ghandeharizadeh, H2O Clouds: Issues, Challenges and Solutions, in IEEE Pacific-Rim S. Ghandeharizadeh, H2O Clouds: Issues, Challenges and Solutions, in IEEE Pacific-Rim Conference on Multimedia, 2003.Conference on Multimedia, 2003.

Page 4: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

Home-to-Home Online (H2O) devices collaborate to Home-to-Home Online (H2O) devices collaborate to deliver continuous media:deliver continuous media:

A H2O device is a wireless device with a powerful A H2O device is a wireless device with a powerful processor and abundant amount of storage.processor and abundant amount of storage.

H2O FrameworkH2O Framework

S. Ghandeharizadeh, H2O Clouds: Issues, Challenges and Solutions, in IEEE Pacific-Rim S. Ghandeharizadeh, H2O Clouds: Issues, Challenges and Solutions, in IEEE Pacific-Rim Conference on Multimedia, 2003.Conference on Multimedia, 2003.

Page 5: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

H2O Framework (Cont…)H2O Framework (Cont…) H2O devices complement existing wired H2O devices complement existing wired

infrastructureinfrastructure

A H2O device may:A H2O device may: Display a clip: video-on-demandDisplay a clip: video-on-demand Store and produce a clip for a displayStore and produce a clip for a display Act as a router of data from a producer to a Act as a router of data from a producer to a

displaydisplay

Page 6: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

H2O Uses & Challenges [2]H2O Uses & Challenges [2] A household may:A household may:

Access clips for entertainment, education, etc.Access clips for entertainment, education, etc. Publish video library (never-erase).Publish video library (never-erase). Establish time-shifted recording of live events: Establish time-shifted recording of live events:

monitor the house while on travel, Memex, monitor the house while on travel, Memex, MyLifeBits, etc.MyLifeBits, etc.

Research challenges:Research challenges: Privacy of user profiles and content.Privacy of user profiles and content. Effective user-interfaces.Effective user-interfaces. How to minimize loss-of-data in the presence of How to minimize loss-of-data in the presence of

node removals.node removals. Hiccup-free display:Hiccup-free display:

Placement of dataPlacement of data

Page 7: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

A C2P2 device is equipped with a wireless device that may A C2P2 device is equipped with a wireless device that may operate in a hybrid manner using either ad-hoc, cellular/802.16, or operate in a hybrid manner using either ad-hoc, cellular/802.16, or both modes of communication.both modes of communication.

Data services: Video/audio-on-demand, road congestion reports.Data services: Video/audio-on-demand, road congestion reports.S. Ghandeharizadeh and B. Krishnamachari, C2P2: A Peer-to-Peer Network for On-S. Ghandeharizadeh and B. Krishnamachari, C2P2: A Peer-to-Peer Network for On-

Demand Automobile Information Services, in Globe 2004.Demand Automobile Information Services, in Globe 2004.

C2P2: Car-to-Car Peer-to-PeerC2P2: Car-to-Car Peer-to-Peer

Ad-hoc Cellular base station

Page 8: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

CHaMeLeoN Software ArchitectureCHaMeLeoN Software Architecture

Static/Dynamic Content

Application Requirements

QoS, Efficiency, Availability

Data and Resource Discovery

Design for Mobility

Routing Protocols

Data Placement

Admission

Control

Data Scheduling and Stream Merging

CHaMeLeoN

Environmental Conditions

Mobility

Traffic Patterns andInformation Association

Wireless ChannelsConstraints

Page 9: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

Smaller & Inexpensive DisksSmaller & Inexpensive Disks 25% annual reduction in size; 40% reduction in cost25% annual reduction in size; 40% reduction in cost

1 inch in height, weighs 1 ounce (16 grams)

1 GB, Year 2004

IBM Microdrive @ $185

Size of a refrigerator, 550 pounds (250 Kg)

1 GB, Year 1980

IBM 3380 @ $40,000

Page 10: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

Inexpensive wireless devicesInexpensive wireless devices

Technology Frequency band Spec B/W Typical B/W Radio-range(indoor)Bluetooth 2.4Ghz 1Mbps 700Kbps 30 feet802.11b 2.4-2.48Ghz 11Mbps 4-5Mbps 300 feet802.11a 5.725-5.85Ghz 54Mbps 20-25Mbps 40 feet

Year 2002

Intel 802.11a @ $185

Year 2004

SMC 802.11g @ $60

S. Bararia, S. Ghandeharizadeh and S. Kapadia, Evaluation of 802.11a for Streaming Data in S. Bararia, S. Ghandeharizadeh and S. Kapadia, Evaluation of 802.11a for Streaming Data in Ad-hoc Networks, in ASWN 2004.Ad-hoc Networks, in ASWN 2004.

Page 11: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

For e.g.For e.g. (a) 3:1 hop (a) 3:1 hop

transmissiontransmission

(b) 1:3 hop (b) 1:3 hop transmissiontransmission

TerminologyTerminology In general, any scenario is m transmissions k hops eachIn general, any scenario is m transmissions k hops each

Denoted as m:k, m,k>=1Denoted as m:k, m,k>=1

Movie Movie

DD

Node 1Node 1 Node 2Node 2 Node 3Node 3 Node 4Node 4

Movie Movie

DD

Movie Movie

DD

Node 1Node 1 Node 2Node 2 Node 3Node 3 Node 4Node 4

Movie Movie

AA

Movie Movie

BB

Movie Movie

CC

1 foot1 foot 1 foot1 foot 1 foot1 foot

Page 12: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 13: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

ADU size defines the size of ADU size defines the size of chunk handed from the chunk handed from the application to the network layer.application to the network layer.

Page 14: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 15: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 16: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 17: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 18: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 19: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 20: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 21: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 22: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 23: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 24: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 25: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ADU SizeADU Size Application Data Unit sizes are dictated by Application Data Unit sizes are dictated by

an application at the data source:an application at the data source:

Node 1Node 1

ApplicationApplication NetworkNetwork

Page 26: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

TCP and UDP performance for a 1:3 hop connectionTCP and UDP performance for a 1:3 hop connection

Bandwidth (Good put) and loss rate for a 1:3 hop Bandwidth (Good put) and loss rate for a 1:3 hop connection.connection.

Movie Movie

DD

Node 1Node 1 Node 2Node 2 Node 3Node 3 Node 4Node 4

Movie Movie

DD

Movie Movie

DD

Page 27: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

TCP and UDP performance for 3:1 TCP and UDP performance for 3:1 hop connectionhop connection

Node 1Node 1 Node 2Node 2 Node 3Node 3 Node 4Node 4

Movie Movie

AA

Movie Movie

BB

Movie Movie

CC

1 foot1 foot 1 foot1 foot 1 foot1 foot

Page 28: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

ObservationsObservations UDP Loss rate between 15-30% with a large UDP Loss rate between 15-30% with a large

variancevariance Losses occur due to transient bottlenecks at Losses occur due to transient bottlenecks at

intermediate routersintermediate routers k participants competing for the channel k participants competing for the channel Due to randomness intermediate router is Due to randomness intermediate router is

flooded occasionally and drops dataflooded occasionally and drops data TCP performs well even though there is the ACK TCP performs well even though there is the ACK

overheadoverhead A protocol with flow control and congestion A protocol with flow control and congestion

control does well in case multiple senders in control does well in case multiple senders in the same radio rangethe same radio range

System may produce data at a slower rate than System may produce data at a slower rate than available network bandwidth available network bandwidth Introduce a delay between successive ADUsIntroduce a delay between successive ADUs

Page 29: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

Seminar’s ObjectivesSeminar’s Objectives1.1. Study fundamental principles that make Study fundamental principles that make

delivery of continuous media across devices delivery of continuous media across devices such as H2O and C2P2 feasible:such as H2O and C2P2 feasible: Space-timeSpace-time Law of large numbersLaw of large numbers Divide-and-conquerDivide-and-conquer

2.2. Implement a prototype of CHaMeLeoNImplement a prototype of CHaMeLeoN

3.3. Investigate complementary frameworks: Investigate complementary frameworks: P2P and Web ServicesP2P and Web Services

Page 30: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

WHAT WILL I LEARN?WHAT WILL I LEARN?

Design and implementation of techniques to Design and implementation of techniques to deliver continuous media using wireless deliver continuous media using wireless networksnetworks Design principlesDesign principles Hands on experience with wireless devicesHands on experience with wireless devices Team-work to implement softwareTeam-work to implement software

On-going research at USC and other places.On-going research at USC and other places.

This course assumes a knowledge of relational databases and a This course assumes a knowledge of relational databases and a programming language such as Java, C, Cprogramming language such as Java, C, C##

Page 31: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

WHAT WILL WE DO?WHAT WILL WE DO? Weekly lectures to cover the fundamentals.Weekly lectures to cover the fundamentals.

Break into groups that is working on a certain Break into groups that is working on a certain project. I will try to meet with all groups each week project. I will try to meet with all groups each week to discuss:to discuss: Current project status.Current project status. What each member is doing to bring the project to an end.What each member is doing to bring the project to an end.

An estimate of when the project will be complete.An estimate of when the project will be complete.

Design and implement novel research ideas, and Design and implement novel research ideas, and write technical manuscripts.write technical manuscripts.

Page 32: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

HOW AM I GRADED?HOW AM I GRADED? This is a seminar course, you should This is a seminar course, you should

enroll because:enroll because: You are interested in the topicYou are interested in the topic You have certain ideas of your own to pursueYou have certain ideas of your own to pursue You want to learn what research is all aboutYou want to learn what research is all about

Your grade will depend on how fully you Your grade will depend on how fully you complete the project that you assume to complete the project that you assume to work on. Generally speaking, projects work on. Generally speaking, projects are team oriented. So, your grade is are team oriented. So, your grade is somewhat dependent on your team somewhat dependent on your team members.members.

There are no exams.There are no exams. How do I choose a project? Either pick from How do I choose a project? Either pick from

one of the offered projects or come up with your one of the offered projects or come up with your own project.own project.

Page 33: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

WHERE WILL I CONDUCT MY PROJECTS?WHERE WILL I CONDUCT MY PROJECTS?

SAL 102 and 200C:SAL 102 and 200C: Linux 7.0, SAL 102 Linux 7.0, SAL 102 Microsoft Windows NT, SAL 200C (Microsoft’s tool set)Microsoft Windows NT, SAL 200C (Microsoft’s tool set) Is there anyone to help me?Is there anyone to help me?

YES!YES! Shahram Ghandeharizadeh, Esam Alwagait, Tooraj Helmi, Shahram Ghandeharizadeh, Esam Alwagait, Tooraj Helmi,

Shyam Kapadia, Sung-ah Kim, Mayank Saxena Shyam Kapadia, Sung-ah Kim, Mayank Saxena

Page 34: CSCI 599: Science of Design for Continuous Media Applications Shahram Ghandeharizadeh Associate Professor Computer Science Department University of Southern.

http://dblab.usc.eduhttp://dblab.usc.edu

SHORT-TERM PLANSHORT-TERM PLAN Thursday, August 26:Thursday, August 26:

S. Ghandeharizadeh, T. Helmi, S. Kapadia, and B. S. Ghandeharizadeh, T. Helmi, S. Kapadia, and B. Krishnamachari. Krishnamachari. A Case for a Mobility Based Admission Control PolicyA Case for a Mobility Based Admission Control Policy. In . In Proceedings of the International Conference on Distributed Proceedings of the International Conference on Distributed Multimedia Systems, San Francisco, September 2004. (Presenter: Multimedia Systems, San Francisco, September 2004. (Presenter: Shyam Kapadia) Shyam Kapadia)

V. Bush.V. Bush. As We May ThinkAs We May Think. . The Atlantic Monthly. July 1945.The Atlantic Monthly. July 1945. (Discussion leader: Shahram Ghandeharizadeh)(Discussion leader: Shahram Ghandeharizadeh)

Overview of a 802.11a mini-projectOverview of a 802.11a mini-project Tuesday, Aug 31, 2004:Tuesday, Aug 31, 2004:

S. Bararia, S. Ghandeharizadeh, and S. Kapadia.S. Bararia, S. Ghandeharizadeh, and S. Kapadia. Evaluation of 802.11a for Streaming Data in Ad-hoc NetworksEvaluation of 802.11a for Streaming Data in Ad-hoc Networks. . In In 44thth Workshop on Applications and Services in Wireless Networks, Workshop on Applications and Services in Wireless Networks, Boston, Massachusetts, August 2004.Boston, Massachusetts, August 2004. (Presenter: Shyam (Presenter: Shyam Kapadia)Kapadia)

Analyze the lap-tops and software to run codeAnalyze the lap-tops and software to run code Thursday, Sept 2, 2004: Student presentations on their Thursday, Sept 2, 2004: Student presentations on their

results.results.