Comparison of Supervised Learning Methods for Spatial User Clustering in Downlink NOMA

Abstract

The performance of Power Domain Non-Orthogonal Multiple Access (PD-NOMA) is affected by the performance of Successive Interference Cancellation (SIC) in decoding user data. The large number of users will cause error propagation in SIC, which results in decreased SIC performance. This research aims to optimize the performance of SIC in PD-NOMA by applying spatial concepts to classify users. This research applies various supervised machine learning classification algorithms, including Decision Tree, K-Nearest Neighbors (K-NN), Support Vector Machine (SVM), Random Forest, Logistic Regression, and Naive Bayes. The experimental results show that Random Forest achieves the highest accuracy in classifying users, followed by Decision Tree. In addition, in performance measurement using ROC (Receiver Operating characteristic) and AUC (Area under the Curve) curves, the Random Forest method achieved the best results. In terms of experimentation process time, a decision tree has a faster time compared to a random forest. Overall, the Random Forest algorithm is suitable for the task of user clustering in the context of PD-NOMA, which utilizes the spatial concept from user to base station (BS).

Keywords

• supervised learning
• spatial clustering
• noma
• pdnoma
• user clustering