This work presents a Single Access Point indoor localization system that can be easily deployed with commodity WiFi infrastructure and that estimates both Angle of Arrival (AOA) and ranges to commodity smartphones. Our system consists of an enrichment of Spring localization method including appropriate improvements to the SpotFi algorithm for AOA estimation. First, a new Channel State Information (CSI) calibration approach is proposed in order to eliminate the existing bias caused by hardware problems. Second, we improve the AOA resolution by showing the benefits of changing the dimensions of the smoothed CSI matrix and applying MUSIC as initial step before the smoothing algorithm. Third, we propose an alternative approach to the Sanitization algorithm of SpotFi in order to improve the time resolution too.
Finally, an analysis of different clustering techniques takes place, showing that the DBSCAN is the most appropriate one, thus giving a more accurate calculation of AOA. As for the estimation of distance, Fine Time Measurements (FTM) procedure is used, and we apply an appropriate method in order to filter these measurements. Our system works with only one AP, and it achieves an 80-th percentile error of 1.67 and 4.1 meters in two different testbeds.
Stavros Eleftherakis has completed his BSc Degree in Mathematics (2017) and MSc Degree in Applied and Computational Mathematics (2020), both from University of Crete. He has also completed a MSc Degree in Telematics Engineering from UC3M (2021). He is a PhD student within the Pervasive Wireless Systems Group at IMDEA Networks Institute and his main research interests are 5G Localization and Machine Learning.
This event will be conducted in English