Chameleon: Survey-Free Updating of Fingerprint Database for Indoor Localization

MPhil Thesis Defence


Title: "Chameleon: Survey-Free Updating of Fingerprint Database for
Indoor Localization"

By

Mr. Bo JI


Abstract

Many fingerprint-based techniques have been proposed and studied in recent 
years for indoor localization. As wireless signals may change overtime or 
suddenly (due to addition/removal of APs, construction/tear-down of partitions, 
adjustment of AP powers, etc.), the fingerprint database needs to be updated 
regularly and frequently in order to achieve high localization accuracy. 
Traditionally the site is surveyed to collect the new radio signatures by means 
of trained surveyors or crowdsourcing. However, these approaches are not 
satisfactory due to their laborious or intrusive nature.

We propose Chameleon, a novel survey-free technique to maintain an updated 
fingerprint database which is adaptive to signal changes. Chameleon eliminates 
the need of any surveyors or explicit user participation by making use of the 
signals measured by the clients, thereby making clients as surveyors as well. 
Independent of any fingerprint-based localization algorithm, it employs an 
efficient algorithm to filter out the altered AP (access point) signals to 
achieve highly accurate indoor localization. Using the calculated location and 
the measured signals, the radio map in the database can be continuously updated 
without manual intervention. Both extensive simulation and experimental trials 
conducted on our campus and Hong Kong International Airport confirm that 
Chameleon is able to adapt the radio map to signal environment, and achieves 
low localization error despite signal changes.


Date:			Thursday, 24 July 2014

Time:			10:00am - 12:00noon

Venue:			Room 3501
 			Lifts 25/26

Committee Members:	Prof. Gary Chan (Supervisor)
 			Dr. Raymond Wong (Chairperson)
 			Dr. Lin Gu


**** ALL are Welcome ****