Add to Google Calendar
Date:  Thu, May 05, 2016
Time:  9:00 am - 11:00 am
Location:  Holmes Hall 389
Speaker:  Darcy Bibb, MS Student, UH Manoa

Abstract

The focus of this thesis is the examination and development of some source localization methods which are effective in urban environments. Multipath, non-line-ofsight, and other effects present in complex environments can be challenging for traditional positioning techniques. Localization methods are investigated with the use of ray tracing simulation along with 3-D models of real-world areas constructed from geospatial resources such as Google Earth to accurately model radio propagation in realistic environments. Two source localization methods are presented, including a time reversal analysis approach as well as the use of machine learning. The time reversal method proves to benefit from multipath conditions, and machine learning shows to be accurate and efficient even for large areas. A combined localization scheme using both methods is also demonstrated, achieving localization accuracies of up to 0.05 m in urban environments.