AROMA: Automatic Generation of Radio Maps for Localization Systems

Abstract

WLAN localization has become an active research field recently. Due to the wide WLAN deployment, WLAN localization provides ubiquitous coverage and adds to the value of the wireless network by providing the location of its users without using any additional hardware. However, WLAN localization systems usually require constructing a radio map, which is a major barrier of WLAN localization systems' deployment. The radio map stores information about the signal strength from different signal strength streams at selected locations in the site of interest. Typical construction of a radio map involves measurements and calibrations making it a tedious and time-consuming operation. In this paper, we present the AROMA system that automatically constructs accurate active and passive radio maps for both device-based and device-free WLAN localization systems. AROMA has three main goals: high accuracy, low computational requirements, and minimum user overhead. To achieve high accuracy, AROMA uses 3D ray tracing enhanced with the uniform theory of diffraction (UTD) to model the electric field behavior and the human shadowing effect. AROMA also automates a number of routine tasks, such as importing building models and automatic sampling of the area of interest, to reduce the user's overhead. Finally, AROMA uses a number of optimization techniques to reduce the computational requirements. We present our system architecture and describe the details of its different components that allow AROMA to achieve its goals. We evaluate AROMA in two different testbeds. Our experiments show that the predicted signal strength differs from the measurements by a maximum average absolute error of 3.18 dBm achieving a maximum localization error of 2.44m for both the device-based and device-free cases.