Multi-Device Fusion for Enhanced Contextual Awareness of Localization in Indoor Environments

Recently, with various developing sensors, mobile devices have become interesting in the research community for indoor localization. In this paper, we propose Twi-Adaboost, a collaborative indoor localization algorithm with the fusion of internal sensors, such as the accelerometer, gyroscope, and magnetometer from multiple devices. Specifically, the data sets are collected first by one person wearing two devices simultaneously: a smartphone and a smartwatch, each collecting multivariate data represented by their internal parameters in a real environment. Then, we evaluate the data sets from these two devices for their strengths and weaknesses in recognizing the indoor position. Based on that, the Twi-AdaBoost algorithm, an interactive ensemble learning method, is proposed to improve the indoor localization accuracy by fusing the co-occurrence information. The performance of the proposed algorithm is assessed on a real-world dataset. The experiment results demonstrate that Twi-AdaBoost achieves a localization error about 0.39 m on average with a low deployment cost, which outperforms the state-of-the-art indoor localization algorithms.