| |
Research Statement
I was a Masters student in the Viral Communications group at the MIT Media Lab, focussed on building agile, scalable, and collaborative systems. During my 2 years at the Lab, I worked primarily on 2 research projects: Fluidvoice (initially called VoiceMesh) and UniPlug.
Advisors:
Dr Andy Lippman (MIT Media Lab), and Dr David Reed (MIT Media Lab)
Collaborators:
Dr Sean Moore and Mads Emborg, Avaya Inc.
Rick Shrenker and Dr Julian Goldman, Medical Device Plug and Play Program, and Operating Rooms of the Future group, Massachusetts General Hospital.
|
|
Outline: FluidVoice (initially called VoiceMesh)
FluidVoice is a decentralized voice-conferencing system running entirely on ad-hoc mobile nodes. Participants can smooth join or leave the system. Voice mixing is done locally, and is controlled by a web-based UI. FluidVoice allowed multiple conversations, making it a generalized realtime voice system on an adhoc network.
The motivation behind FluidVoice was to build a working application on an ad-hoc wireless network, and in doing so stress the system to expose its limitations and fix them. Realtime voice, being sensitive to delay and jitter, can be an accurate indicator of the underlying network's performance. To have a credible measurement that goes beyond perception, I carried out extensive channel quality tests using an implementation of PESQ - which is an ITU standard for measuring voice quality in a telecom system.
To extend the system further, I created a software adapter in Python that allowed Avaya peer-to-peer Ethernet based IP phones to call into the network.
Outline: UniPlug
For my masters thesis, I developed a framework called UniPlug for software invention sharing on an ad-hoc campus network. The motivation for UniPlug came from the fact that so many students on the MIT campus write software for various devices either for fun or part of class projects. Such software inventions are usually lost over time because not everyone has the flair or motivation to go and create an open source project on a site such as SourceForge to distribute the software. Many of such applications are interesting and useful, and their easy dissemination would have value for others.
I architected and implemented a working prototype of
UniPlug. UniPlug consists
of client software (UniClient) that searches the local environment for software inventions when a
device is plugged in, and server software (UniServer) that allows users to share their
software inventions to others in proximity. It has reputation and safety mechanisms to
ensure that the system is not misused. Where there is high activity, UniPlug allows credit
based "election" of personal machines to serve as UniServers, and opportunistically
caches locally relevant software. The UniClient is context-aware (user, device, or
surroundings based context) to maximize device efficiency.
This project involved
collaboration with Avaya in building the working prototype and with the Operating Room of the Future group at Massachusetts
General Hospital in designing the framework to be extendable to the medical domain.
Misc
Apart from the projects outlined above, I have worked on Sensory for a Media Lab design class, and IndiMa (executive summary, slides) for a Developmental Entrepreneurship class, respectively. Sensory is a system for measuring and displaying environmental information to increase awareness and prevent pollution in a model green city in Italy. IndiMa is a system for tracking women with complicated pregnancies in India and providing them with priority access to medical treatment in case of emergencies.
Here is a peer review I wrote of 3 unpublished papers as part of Bruce Davie's class on Advanced Topics in Service Provider Networking at MIT.
|
|
Papers:
UniPlug: a framework for ad-hoc invention sharing over a campus network
Durga Prasad Pandey
Masters thesis, Massachusetts Institute of Technology, September 2007.
Voice over AODV
Durga Prasad Pandey
6.829 Course Research Paper, Department of Electrical Engineering and Computer
Science, Massachusetts Institute of Technology, Cambridge, MA, December, 2006.
Liquid Communications
Durga Prasad Pandey
Poster, Digital Life meeting, MIT Media Lab, October, 2005.
| |
Talks
Ad-hoc Infrastructure-less Communications
Atheros Communications, Santa Clara, CA, August, 2007.
Experiments with User-Centric Ad-hoc Applications
IBM Watson Research Center, Cambridge, MA, July, 2007.
FluidVoice
Communications Futures Program Bi-Annual meeting, Nortel Networks, Richardson, TX, January, 2007.
AODVoice: Voice over AODV
6.829 Project presentation, CSAIL, Massachusetts Institute of Technology, Cambridge, MA, December 2006.
VoiceMesh: Voice Communcations on a Wireless Mesh Network
Cisco Tech Center, San Jose, CA, November, 2005.
| |
|
| | | | |