An indoor radio propagation model considering angles for WLAN infrastructures
Shuai Han,Zijun Gong,Weixiao Meng,Cheng Li
DOI: https://doi.org/10.1002/wcm.2596
2015-01-01
Wireless Communications and Mobile Computing
Abstract:Wireless local area network fingerprint-based indoor location system is a hot topic these years because it needs no extra hardware and is very easy to deploy. However, it demands a database containing the distribution of received signal strength RSS of the area of interest,called radio map. Conventionally, we need to grid the area densely and manually measure RSS values on intersections, which will consume a lot of time and human resources. What is worse, change of the environment may render this database totally useless. Our consideration is to measure RSS on a small amount of these intersections and use them to build a radio propagation model. Then, this model can be deployed to predict RSS values of other intersections and reconstruct the radio map. In other words, we only need to collect a very small part the radio map and utilize the radio propagation model to recover the whole one. So far, many models have been proposed, among which the one suggested by Seidel, named floor attenuation factor propagation model, achieves great balance between computational request and accuracy. But it is not compatible with environments in some scenarios. So as to compensate for this deficiency, we take into account the angles formed by signal and surfaces of obstacles, and the results show better compatibility. The proposed model has four parameters that are related to the environments, and our second contribution in this paper is to propose a method to determine them. In fact, after collecting a small part of the radio map, we can estimate these parameters with least square method. Then, these parameters can be used to predict the signal strength at any other points in the same environment, and the whole radio map is rebuilt. According to practical experiments, performance of the radio map built by the proposed model is not as good as the manually collected one, but 80% of collecting labor is saved. Copyright © 2015 John Wiley & Sons, Ltd.