An Intelligent Solution to In-Home Power Line Communication

Yingxue Li,Huaide Peng,Boxuan Chang,Xiaofeng Zhang,Mengyuan Cheng,Yongpeng Wu
DOI: https://doi.org/10.1109/icicas48597.2019.00171
2019-01-01
Abstract:This paper provides an intelligent solution to power line communication (PLC) by applying deep learning (DL) to the transmission system of PLC. To this end, we choose low voltage (LV) power line as a study case. Based on the channel measurement campaign on residential LV distribution areas in Ganzhou city, Jiangxi province, China, the in-home PLC channel is modeled as a multi-path channel. As for the noise model, we adopt the additive Middleton Class-A noise (AWCN) composed of impulsive and background noise. In order to combating this hostile PLC environment, a proprietary data encoder is recommended in the physical layer of the IEEE standard for broadband over power line networks. Accordingly, we design an intelligent data decoder combining deep neural network (DNN) and domain knowledge in decoding algorithms, able to learn and adapt the environment of PLC channel until achieving the optimal performance. The simulation results confirm the high-performance and low-complexity of the proposed decoder.
What problem does this paper attempt to address?