On the empirical performance of self-calibrating WiFi location systems

Abstract

Abstract-The pervasive deployment of 802.11 in modern enterprise buildings has long made it an attractive technology for constructing indoor location services. To this end, a broad range of algorithms have been proposed to accurately estimate location from 802.11 signal strength measurements, some without requiring manual calibration for each physical location. Prior work suggests that many of these protocols can be highly effectivereporting median errors of under 2 meters in some instances. However, there are few studies validating these claims at scale, nor comparing the algorithms in a uniform, realistic environment. Our work provides precisely this kind of empirical evaluation in a realistic office building environment. Surprisingly, we find that median errors in our environment are consistently greater than 5 meters and, counter-intuitively, that simpler algorithms frequently outperform their more sophisticated counterparts. In analyzing our results, we argue that unrealistic assumptions about access point densities and underlying variability in the indoor environment may preclude highly accurate location estimates based on 802.11 signal strength.

LiPro: light-based indoor positioning with rotating handheld devices

“…Given the ubiquitous radio frequency (RF) signals, measurement of RSS of WiFi [1,5,28,32], cellular radios [24], or FM radios [4], can be used to build an RF signature database, or to establish a signal propagation model for positioning. However, RF signal is susceptible to noise and interference, and thus is highly unstable, leading to rather coarse-grained positioning results, with distance errors typically to a few meters [23]. Hence, many efforts have been devoted to mitigate the instability in a multi-path fading environment [8,12,21,27].…”

Last-Mile Navigation Using Smartphones

“…The rationale behind the step-constraint is simple: the difference between the number of steps taken by the leader and the follower within a small given physical distance should be bounded. == true) in the previous iteration, the algorithm calculates and outputs the matching result uv, and then reads a new sample (lines [8][9][10][11][12]. Between line 14 and 27, the algorithm calculates a partial row or column of the warping cost matrix D (i.e., compare magnetic field values and compute similarities).…”

“…Although multiple model-based [5,8] and crowdsourcing [6,9,10] techniques have been proposed, the need for precise building structure information and a sustainable incentive crowdsourcing mechanism limits their applicability [11]. Even if deployed, indoor localization systems may face an onerous calibration process (for those radio-based fingerprint systems) and need to deploy path-planning algorithms to enable online navigation services.…”

Platform and Algorithm Development for a RFID-Based Indoor Positioning System

“…For instance, the log-distance PL model is used to establish the relationship between the measured RSS and the Radio Frequency (RF) propagation distance [21,22]. Several model-based approaches employing radio propagation models have been investigated in [23]. The average localization accuracy of these IPSs is around 5 m.…”