Potentials and limitations of WIFI-positioning using Time-of-Flight

Many existing approaches for WIFI-based indoor positioning systems use the received signal strength indicator (RSSI) of all the access points in reach to estimate the current location of a mobile device. Those systems´ accuracy, however, is strongly influenced by external interferences and suffers both from short-term and long-term changes in the respective environments. Time-of-Flight-based (TOF) approaches bypass these problems by relying on the relation between distance and the time it takes a radio signal to travel that distance. Furthermore and in contrast to RSSI-based fingerprinting methods, they require neither a time-consuming calibration phase nor an extensive database. In this paper, we assess the advantages and limitations of TOF-based positioning techniques and compare different approaches in literature in terms of communication flow, hardware components, method of time measurement and positioning accuracy. Additionally, we present a novel approach using NULL-ACK-sequences, off-the-shelf hardware components and the CPU´s time stamp counter offering a nanosecond resolution. An association with access points is hence not required and there is no need for modifications of client or infrastructure WIFI-components. We evaluate our system in various settings. Our results indicate that the accuracy of such TOF-based approaches depends on both the used hardware and the characteristics of the given environment. We find that time fluctuations caused by varying delays of the interrupt service routine as well as multipath effects render precise distance estimations based on a single measurement infeasible. In order to obtain stable ranging results we try to minimize these effects by utilizing a high amount of NULL-ACK-sequences. We investigate several filters and statistical estimations and compare them within our system settings. Using a band-pass filter and taking the average of a series of measurements, we are able to achieve a ranging accuracy with a mean ab- olute error of less than 1.33 meters in an ideal environment.