Multi-sensor location estimation for illegal cell-phone use in real-life indoor environment

This paper proposes a novel method for location estimation of a cell-phone in an indoor environment with experimental validation. The proposed method is a location fingerprint scheme which employs the statistical characteristics of the signal cross-correlation among multiple sensors. As the cross-correlation between a pair of antennas fully stores the information of their complex channel responses, the proposed method can be considered as a generalized scheme of all conventional location fingerprint ones i.e. Received Signal Strength Indicator (RSSI), Time Difference of Arrival (TDOA), and Direction of Arrival (DOA). Besides, different from the conventional methods, in which the locality of spatial correlation is utilized, thus fine grid location measurements is required and environment must be static, the proposed one invokes statistical learning technique and estimates location based on the correlation of received samples with the statistical learning database. Therefore, the proposed method is superior to conventional ones in terms of location estimation accuracy and installation simplicity. An experiment conducted in class room validates the superiority of the proposed method.