WLAN location determination model based on the artificial neural networks

Along with increasing popularity of wireless LAN, problem of location determination for mobile users becomes more important. The strengths of RF signals arriving from several access points can be used for location determination of the mobile terminal. In indoor environments the received signal level is very complex function of the distance. The solution can be found in the area of artificial neural networks. The neural networks can be learned to classify data. Labeled data examples of signal strengths at known locations must be collected by the measurement. This data will serve for the training of the network with appropriate training algorithm. The trained network is capable to determine location on the base of new signal strengths as a process of generalization. The advantage of the method is that it doesn't need any extra hardware, while with flexible neural network model achieves lower distance errors in determining position comparable with other methods. For successful position determination only what is needed are a map of indoor space and several identified locations to train the network.