Navigation using VLF environmental features

This paper examines the use of Very Low Frequency (VLF) electromagnetic signals to perform location estimation with-out any knowledge of the source of the VLF signal or their interaction with the environment. Signals were collected from different locations in a variety of indoor environments. The signals are divided into equal time length segments and processed into feature sets. A logistic regression and an Artificial Neural Network are then both trained on the sample statistics to give a set of parameters to return likelihood of each locations with the correct location being determined as the position with the highest likelihood. This approach using either method achieved an error of less than 1% for 17 locations recorded for 10 minutes each.