Fast-converging indoor mapping for wireless indoor localization

Abstract

The construction of indoor floor plans for wireless localization with minimal deployment effort and with low cost is relevant in many ambient intelligence and mobile computing applications. Simultaneous Localization and Mapping (SLAM) is a well-known technique for solving the problem of localization of an object in an unknown environment while simultaneously constructing a map of the surrounding area. A majority of previous indoor mapping methods in SLAM use odometry measurement with exteroceptive sensing to simultaneously estimate user position and construct a map of the environment. Inertial navigation is used in SLAM by constantly monitoring heading changes and distance traveled by means of inertial sensors. This offers slow convergence of floor plan construction and generates large errors due to the relative lack of indoor position landmarks. The goal of the present thesis is to explore the wall penetration loss effect of wireless signals in order to construct a floor map using Received Signal Strength (RSS) readings from moving smartphones. A method by which to construct indoor maps at low cost and with fast convergence without the need to collect the odometry data of the user.