Poster: A Hybrid Indoor Audio Localization System

Abstract

User localization with mobile devices remains a challenging research problem if the environment or required accuracy prevent the use of satellite navigation (GPS). In this paper, we present a hybrid system that performs real time localization on a mobile device, using the audio signals emitted by nearby loudspeakers. Such an approach is useful in controlled environments with background music and sound, such as shopping centers, malls, or entertainment theme parks (Fig. 1). Two known audio identification methods are used in combination: (1) watermarking [1] with hidden location markers and (2) fingerprinting [2] with sound matches in a database. Based on the performance of the two methods in terms of accuracy and energy consumption, a particle filter-based hybrid combination [3] of CRC-based watermarking and fingerprinting is proposed. Our localization scheme runs as stand-alone mobile application and enables a user to identify the current location. It also respects the user’s privacy as it runs locally on a mobile device. We present the testbed implementation and experimental evaluation in an indoor environment with respect to the presence of noise, interference, user mobility, and power consumption. This implementation is the first step towards fine-grained localization using sound in the future.