Mobile Station Positioning Using GSM Cellular Phone and Artificial Neural Networks

In this paper, we describe a novelapproach to mobile station positioning using a GSMmobile phone. The approach is based on the use of aninherent feature of the GSM cellular system (themobile phone continuously measures radio signalstrengths from a number of the nearest base stations(antennas)) and on the use of this information to estimatethe phone's location. The current values of the signalstrengths are processed by a trained artificial neuralnetwork executed at the computer attached to themobile phone to estimate the position of the mobilestation in real time. The neural network configurationis obtained by using a genetic algorithm that searchesthe space of specific neural network types anddetermines which one provides the best locationestimation results. Two general methods are explored:the first is based on using a neural network forclassification and the second uses functionapproximation. The experimental results are reportedand discussed.