LASR Colloquia - Venkat Padmanabhan, Microsoft Research, "Indoor Localization: The Quest for Zero-Effort and for a Killer App"

Contact Name: 
Lili Qiu
Location: 
GDC 6.516
Date: 
Dec 4, 2013 11:00am - 12:00pm
Host: 
Lili Qiu
Speaker Affiliation: 
Microsoft Research

Signup Schedule: http://apps.cs.utexas.edu/talkschedules/cgi/list_events.cgi

Talk Audience: UTCS Faculty, Grads, Undergrads, Other Interested Parties

Host: Lili Qiu

Talk Abstract: Devising a method to locate people and devices in indoor spaces, where GPS is not an option, has been a long-standing research problem, receiving much attention for over two decades. WiFi-based indoor localization has emerged as perhaps the most promising approach because it holds the promise of providing good accuracy at a low cost, by leveraging the near-ubiquitous coverage of WiFi in indoor spaces of interest such as office buildings and malls. However, widespread deployment has been stymied by the need for significant calibration effort.

In this talk, we will trace through our research at Microsoft Research, moving towards the goal of zero-effort indoor localization. We will start with the Radar system, which pioneered the idea of piggybacking on an existing RF-based wireless LAN to perform indoor localization. Radar introduced two approaches to indoor localization: (a) RF fingerprinting through empirical measurements made at known locations, and (b) RF computation using mathematical modeling. While these techniques offer good localization accuracy and there has been much follow-on work on further improving accuracy, making empirical measurements in a new environment is expensive and customizing the mathematical model to such an environment is challenging. To overcome these difficulties, we advocate a crowdsourcing-based approach, wherein smartphones carried by users in normal course are used to make measurements and construct models, without any explicit effort on the part of users. The key is to use the radios and sensors on these smartphones, such as WiFi, GPS, accelerometer, and compass, in unison to (i) track users and thereby enable empirical measurement of training data, using a system called Zee, and (ii) construct an RF model with patchy information, using a system called EZ.

We will conclude by touching on a new effort on Physical Analytics, which builds on indoor localization and could well turn out as its "killer app".

[EZ is joint work with Krishna Chintalapudi and Anand Iyer, and Zee with Krishna, Anshul Rai, and Rijurekha Sen.]

Speaker Bio: Venkat Padmanabhan is a Principal Researcher and Research Manager at Microsoft Research India in Bangalore, where he founded and has led the Mobility, Networks, and Systems group since 2007. Venkat was previously with Microsoft Research Redmond for nearly 9 years. His research interests are broadly in networked and mobile systems, and his current work is on indoor localization and efficient mobile communication. He was General Co-Chair for ACM Sigcomm 2010 in New Delhi, program co-chair for ACM Sigcomm 2012, and presently serves as Chair of the Sigcomm Conference Technical Steering Committee. He has also served as an affiliate faculty member at the University of Washington, where he has taught and served on student thesis committees. Venkat holds a B.Tech. from IIT Delhi and an M.S. and a Ph.D. from UC Berkeley, all in Computer Science. He has been elected a Fellow of the Indian National Academy of Engineering (INAE) and the IEEE, and is a Distinguished Scientist of the ACM.

Tags: