Toward better indoor localization

Abstract

Location-based service (LBS) is believed to be next big thing for mobile applications. Such service depends largely on accurate localization of mobile devices. Though Global Positioning System (GPS) has already achieved good accuracy, it only works in outdoor open environment. Indoor localization still remains challenging. Although many indoor localization techniques have been studied and deployed, each of them has its own limitation with accuracy depending on the environment being applied (e.g., penetration of wireless networks, availability of special devices, etc.).In this talk, I discuss two of our recent advances on indoor localization which augment current indoor estimation techniques to improve their accuracies. I first discuss a cost-effective, distributed and cooperative (i.e., peer-to-peer) scheme for localization. The scheme achieves high accuracies even with coarse measurements and sparse landmarks. It works for dynamic networks and nodes with limited power.Given that many localization techniques may co-exist at the same time, I then share our initial study on estimation fusion, which achieves localization by combining the outputs of multiple estimators. Such fusion is able to mitigate the measurement uncertainty or error of each of the estimators. Encouraging preliminary results show that the approach is promising.