Measurement-based RSS-multipath neural network indoor positioning technique

Significant developments in indoor positioning techniques based on location fingerprint have been seen recently. RSS (received signal strength) is the most frequently-used indoor fingerprint information. The precision and accuracy of indoor positioning can be improved if we make better use of channel state information and apply more effective matching algorithms. In this study, a method for multipath similarity measurement using multipath time delay and amplitude is proposed. We expand the positioning fingerprint based on the proposed multipath similarity measurement method. Neural network technique is an effective classification and prediction method. An RSS-multipath joint neural network positioning technique is proposed to improve the indoor positioning performance. Distributed MISO (Multiple-Input Single-Output) channel measurement campaign using the THU channel sounder is carried out in indoor environments. Analysis of the experimental results shows that the proposed RSS-multipath joint neural network positioning technique outperforms classical fingerprint algorithms and can improve the positioning accuracy effectively.