Machine Learning in Visible Light Communication System: A Survey

Table 6

Comparison of various ML schemes.

ML algorithms	Applications	Ref.	Action position	Characteristic parameters
-NN	Positioning	[14–17]	Receiver	Supervised, moderate complexity
SVM	Phase estimation	[50]	Receiver	Supervised, moderate complexity
ANN and DNN	Nonlinearity mitigation, positioning, and channel estimation	[10, 11, 23, 33, 41, 46, 47]	Receiver	Supervised, high complexity
Multiple classifier	Positioning	[44, 45]	Receiver	Supervised, high complexity
-Means	Nonlinearity mitigation, positioning, phase estimation, and modulation identification	[18, 19, 25, 26, 42, 43, 53]	Transmitter and receiver both	Unsupervised, low complexity
GMM	Modulation identification and phase estimation	[20, 26]	Receiver	Unsupervised, moderate complexity
DBSCAN	Jitter compensation	[30, 48, 49]	Receiver	Unsupervised, low complexity