RF Signal Strength Modeling in Indoor Environment for Cost-Effective Deployment of Fingerprinting Technology

Abstract

This paper proposes an RF signal strength modeling algorithm to cost-effectively generate an database for indoor localization. Currently, while outdoors, it is possible to provide reliable location information to users using the Global Navigation Satellite System (GNSS), there is no clear solution indoors. Fingerprinting is a localization technology that can provide reliable location information in an indoor environments where multipath for RF signals is severe. Fingerprinting technology calculates the location by directly comparing the measured Received Signal Strength (RSS) from the surrounding signal sources with the database where RSS is stored for each indoor space. So, the fingerprinting technology is robust to signal multipath, but has the disadvantage that it takes a lot of time and cost to construct database. Indoor localization technology is highly dependent on the displacement of RF signal sources. Therefore, for the high accuracy of the localization technology, it is necessary to consider the displacement of the signal source for each indoor environment. However, the fingerprinting technology should re-construct a new database whenever the displacement of the signal source is changed. Manually measuring RSS for each location indoors requires a lot of time and cost. To solve this problem, we propose a simulation algorithm that models the RF signal strength indoors according to the displacement of the signal sources. The proposed algorithm places RF signal sources virtually in a given indoor space, and calculates an RF signal strength pattern based on the dual-slope path loss model. In particular, the proposed algorithm utilizes both the distance-based propagation model that considers only the distance from a signal source and the propagation model on the path where a user can walk, that is, the link-based propagation model. By using the two models together, the proposed algorithm can accurately reflect changes in signal strength, such as penetrating a wall at a close distance. The calculated indoor RF signal strength pattern can be utilized as a fingerprinting database. That is, the proposed algorithm enables performance analysis of localization technology through virtual simulation indoors. To verify the proposed algorithm, we installed six Bluetooth beacons in a department store. We calculated signal pattern through the proposed algorithm and compared with the measured signal pattern. Also, we constructed fingerprinting database for each pattern. Accordingly, the localization performance was analyzed by comparing measurements collected through a user test with each database, and as a result, a localization error of about 3 m was confirmed for each database. Through these results, it was confirmed that unnecessary time and cost for database construction can be minimized by indirectly checking the RF signal strength pattern through the virtual displacement of the RF signal source.