Artificial intelligence enabled 5G-NR optimisation

Abstract

Radio-frequency spectrum chunks/channels are like invisible traffic lanes that can be used simultaneously by multiple telecoms operators. For realising spectrum efficient intelligent communication in 5G-NR, the latency or time delay of corresponding channel may have to be reduced up to 5 ms. This paper discusses how artificial intelligence techniques can be applied at the physical layer of base station to train machines/devices for smooth user experience at application layer. An estimated channel can be trained by generating meaningful data/information at physical layer itself and then neural networks principal techniques like convolution neural networks (CNNs) can be applied to train the channel/spectrum for best schedules. Estimated best schedules can be predicted with negligible time delay by applying artificial intelligence techniques.