An Evaluation of the Temporal Stability of Leading RF Technologies for Fingerprinting-based Indoor Localization Systems

Erich P. Stuntebeck, Rahul Rajan, Tom Robertson,

Ramakrishnan Chandrasekharapuram, Gregory Abowd

Ubiquitous Computing Group, Fall 2009

Abstract: RF fingerprinting indoor localization techniques work be- cause site surveys of various features, such as signal strength, reveal significant spatial variability. Each part of an indoor space is uniquely identified by its fingerprint of a set of features of the RF signals. This technique has been exploited for a variety of RF sources, including WiFi, GSM, CDMA, FM and even power line transmitted signals in the home. Closer scrutiny of the research in this area reveals that an- other critical characteristic of a successful fingerprinting technique, the temporal stability of the features, has not been as carefully evaluated. Temporal instability of the features means that a single site survey will not be adequate for good performance over a long period of time. In this paper, we explore the temporal stability of the raw signals underlying all previously reported fingerprinting solutions. Our results reveal the unfortunate conclusion that most of the measured features do not exhibit very good temporal stability, with FM and powerline solutions showing the most promise, at least in domestic settings.