Estimating User Location in a WLAN Using Backpropagation Neural Networks

Context-aware computing refers to an application’s ability to adapt to changing circumstances and respond based on the context of use. The estimation of user location is crucial to many context-aware applications. In this paper we propose a technique to infer user location in a wireless LAN inside buildings based on backpropagation neural networks. The strengths of the radio-frequency (RF) signals arriving from several access points in a wireless LAN are related to the position of the mobile device. Estimating the position of the mobile device from the RF signals is a complex inverse problem since the signals are affected by the heterogeneous nature of the environment. Backpropagation neural networks represent a viable alternative to tackle this problem given their property of generalizing from examples. Experimental results provide an average distance error of 1.87 meters from the real location, which is considered to be adequate for many applications. A simple contextaware application is presented as an example. The estimation errors obtained are similar to those using k-nearest neighbors; however the approach presented here uses less memory, an important concern for handheld devices with limited storage capacity an operating on relatively slow WLANs.