Object tracking through RSSI measurements in wireless sensor networks

Abstract

International audienc

Noise reduction for variance-based device-free localization and tracking

“…RELATED WORK DFL using RF sensor networks has potential applications in surveillance for police and firefighters. Different measurements and algorithms have been proposed [4], [6], [7], [5]. For RSSbased DFL, there are essentially two types of algorithms: fingerprint-based algorithms and model-based algorithms.…”

“…For RSSbased DFL, there are essentially two types of algorithms: fingerprint-based algorithms and model-based algorithms. Like fingerprint-based real-time location service (RTLS) systems, fingerprint-based DFL methods use a database of training measurements, in which a person stands at all possible locations, and estimate people's locations by comparing the measurements during the online phase with the training measurements [6], [7], [17]. Since a separate training measurement dataset is necessary, fingerprint-based DFL needs substantial calibration effort.…”

“…For example, [5] deploys an RF sensor network around a residential house and uses sample variance during a short window to track people walking inside the house; [6] places RF sensors on the ceiling of a room, and track people using the RSSI dynamic, which is essentially the variance of RSS measurements with and without people moving inside the room. In this paper we use windowed variance to describe the various functions of RSS measurements recently used in different DFL studies [4], [6], [5], [7], and we call these methods RSS variance-based DFL methods.…”

RF Sensor Networks for Device-Free Localization: Measurements, Models, and Algorithms

“…For example, the parameters may only be a function of the distances x o −x t and x o −x r . These assumptions are used in the algorithms presented in Sections IV-B through IV-H. 2) Absolute position dependence: The channel parameter dependence cannot be simplified using the relative positions of x o , x t , and x r [1], [18], [17], [10]. In the latter case, the dependence of the measurement on x o must be determined for every link (and thus x t and x r ), and for the entire range of x o , for each environment.…”

“…Results have been presented which count the number of people moving [9], estimate a person's location [18], [13], [17], [14], and image the movements in an area of interest [11], [12], [13], [15], all in real-world multipath environments. Both location estimation (estimating a person's coordinate at one time) and tracking (estimating a person's velocity and sequence of positions over a duration of time) have been reported, with accuracy of less than one meter of average error [8], [14], [13] or less than two meters median error over a 1500 m 2 area [17]; these results are at least as good as reported location error when locating radio tagged objects [4], [19], so the accuracy is surprising.…”