BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Cooperative Localization with Turbo-Networks - Dr Henk Wymeersch\,
  Massachusetts Institute of Technology
DTSTART:20080911T130000Z
DTEND:20080911T140000Z
UID:TALK13134@talks.cam.ac.uk
CONTACT:Ioannis Chatzigeorgiou
DESCRIPTION:The ability to locate devices and people will be an important 
 feature in future wireless networks. Current approaches to achieve locatio
 n-awareness are limited in several ways: systems such as GPS tend to fail 
 in indoor environments\, under tree canopies\, and in urban canyons\, due 
 to the inability of GPS signals to penetrate obstacles\; WiFi or GSM local
 ization systems provide only very coarse location information\, making the
 m unsuitable for many applications\; ultra-wide bandwidth (UWB) systems ca
 n provide centimeter-level localization accuracy\, but require a significa
 nt infrastructure investment. To address these challenges\, we propose to 
 leverage the expected ability of wireless devices to communicate directly\
 , rather than solely through access points or base stations. This allows d
 evices to help one another in determining locations. Such cooperative loca
 lization can improve availability and accuracy dramatically.\n\nIn this ta
 lk\, I will present our recent progress on cooperative localization. We wi
 ll cover fundamental performance bounds\, practical algorithms as well as 
 experimentation with UWB technology. Particular emphasis will be placed on
  algorithm development\, applying techniques from iterative decoding and g
 raphical models\, and extending them to a network setting. Performance res
 ults for large-scale scenarios will be presented\, based on extensive expe
 rimental data from a recent measurement campaign. Our results indicate tha
 t cooperation can offer orders of magnitude performance improvement\, as c
 ompared to conventional approaches. 
LOCATION:Computer Laboratory\, Room FW26
END:VEVENT
END:VCALENDAR
