Towards intelligent user clustering techniques for non-orthogonal multiple access: a survey

EURASIP Journal on Wireless Communications and Networking

Table 3 Supervised machine learning-based user clustering schemes in NOMA

Clustering scheme	Related work	Scenario	Assumptions	Objective functions	Application	Findings
Artificial neural network (ANN)	[41]	Downlink	Base station knows the CSI of all the users	Sum rate maximization	5G	Predict the cluster formation automatically to reduce the computational complexity.
Deep neural network (DNN)	[42]	Uplink	Base station knows the CSI of all the users	Sum rate maximization	MIMO	Cluster formation based on feed-forward neural network.
Deep neural network (DNN)	[43]	Downlink	Base station knows the CSI of all the users	Sum rate maximization	MISO	It is suitable for clustering more complex scenario and enhance computational complexity.
Long short term memory (LSTM)	[44]	Downlink	Base station knows the CSI of all the users	Sum rate maximization	5G	Clustering is based on time series data which effectively predicts the number of cluster as compare to exhaustive search method.
Extreme learning machine (ELM)	[45]	Downlink	Base station knows the CSI of all the users	Throughput maximization	5G	Optimized clustering based on fast learning speed of ELM at low complexity.
Genetic Algorithm	[46]	Downlink	Base station knows the CSI of all the users	Throughput maximization	MISO	Reduces the complexity of clustering as compared to the other exhaustive search method.