Home // International Journal On Advances in Telecommunications, volume 14, numbers 1 and 2, 2021 // View article

Supervised Machine Learning in Inter-Level, Ultra-Low Frequency Power Line Communications

Authors:
Kushal Thapa
Stan McClellan
Damian Valles

Keywords: ultra-low frequency power line communications; machine learning; smart grid; feature based learning; featureless learning

Abstract:
Power Line Communications (PLC) is a technology that uses power lines to carry communication data alongside electrical signals. This technology has a huge potential and low infrastructure cost due to the pre-existing and ubiquitous power grid. However, the electrical components in the grid and the highly dynamic properties of the grid make power lines a hostile medium for PLC, especially when the PLC signal path extends over multiple levels (current or voltage) of the grid. Subsequently, efficient transmission of the PLC signals from the transmitter end and effective demodulation at the receiver end are both challenging. In our research, we limit PLC transmission frequencies to the ultra-low spectrum and investigate supervised machine learning as a potential signal demodulation technique. Employing inter-level PLC architecture, we transmit and collect baseband-modulated data, then use various machine learning algorithms to recover the data. We also investigate various feature-based and featureless machine learning methods for PLC and conclude that feature-based methods provide better generalization for our dataset.

Pages: 51 to 69

Copyright: Copyright (c) to authors, 2021. Used with permission.

Publication date: December 31, 2021

Published in: journal

ISSN: 1942-2601