Centaur : Locating Devices in an Office Environment

We consider the problem of locating devices such as laptops, desktops, smartphones etc. within an office environment, without requiring any special hardware or infrastructure. We consider two widely-studied approaches to indoor localization: (1) those based on Radio Frequency (RF) measurements made by devices with WiFi or cellular interfaces, and (b) those based on Acoustic Ranging (AR) measurements made by devices equipped with speaker/microphone. A typical office environment today comprises devices that are amenable either one or both these approaches to localization. In this paper we ask the question, “How can we combine RF and AR based approaches in synergy to locate a wide range of devices, leveraging the benefits of both approaches?” The key contribution of this paper is Centaur, a first of its kind system that fuses RF and AR based localization techniques into a single framework. Centaur uses Bayesian inference techniques and is agnostic to the specific RF or AR technique used, giving users the flexbility of choosing their preferred RF or AR schemes. We evaluate the performance of Centaur through a deployment in an office environment across various RF and AR techniques.