Cost-Effective Localization in Underground Mines Using New SIMO/MIMO-Like Fingerprints and Artificial Neural Networks

Safety measures have always been a main concern in the mining industry that, despite the modern practices, utilizes old-fashioned surveillance and monitoring systems. Our mission in underground mines stems from the profound need of geo-positioning systems that can accurately localize endangered miners and their heavy machinery in one of Earth's most harsh and rough environments. In underground mines, complex channels' responses to wireless transmitted signals challenge traditional localization techniques, yet they fail to defeat our innovative, cost-effective and accurate fingerprint-based positioning techniques that use artificial neural networks (ANNs) and exploit space-time diversity. Being among the pioneers in underground communications research, we bring forward a more sophisticated and accurate fingerprint-based positioning technique that exploits spatial transmission diversity in the presence of more than one transmitter T-x and/or receiver R-x antenna, such as in the case of single/multiple input multiple output (SIMO/MIMO) communication systems. More importantly, an advanced study is conducted to reduce the cost of fingerprint-acquisition trading off pinpoint accuracy for lower complexity and better ANNs' design. By challenging the localization system using less data measurements, we prove that ANNs, when properly designed, succeed to attain high positioning accuracies even when localizing in measurement gaps that were not seen in the training phase.