Machine Learning Models for Channel Status Classification in M-Mimo Systems Using Limited CSI Feedback

In next generation networks, knowing if a channel has a Line of Sight (LOS) path between the transmitter and the receiver is becoming increasingly important. For example, researchers have optimized channel estimation and wireless localization algorithms for both LOS and Non-LOS (NLOS) scenarios. Knowing the LOS status of a channel will allow system performance enhancement by employing the best algorithm available. This study explores the use of various machine learning classifiers to identify the LOS status of simulated massive-MIMO channels. The classifiers make their predictions based on limited Channel State Information (CSI) feedback received at the base station. This study identifies and properly manages the class imbalance problem present in LOS/NLOS identification. Promising results are achieved and demonstrated using a synthetic benchmark.