. 2021 Dec 3;21(23):8086. doi: 10.3390/s21238086

Table 4.

Computational comparison of existing localization techniques.

Reference	Technique	Complexity	Symbol and Notation
[18]	ANN-based deep learning techniques	$O (N_{n e u r o n s}^{2})$	$N_{n e u r o n s}$ is the number of neurons of the trained ANN
[28]	CNN-based completed distance refinement and DNN-based recovery scheme	$O (6 U^{3} + 9 U^{2})$	U is the total number of sensing nodes, including M known reference points (RPs) and N unknown points (UPs) to be localized
[24]	KNN-based and Naive Bayes-based methods	$O (m n)$	m is the number of possible transmitters to verify RSSI measurement; n is the number of comparisons performed between RPs and UPs on RSSI measurement
[130]	Local Gaussian Process method for fingerprint indoor localization based on WLAN radio map	$O (n L + L^{3})$	n is the number of RPs; and L is the number of RPs in a training set
[37]	weight estimation of Unscented Kalman Filter (UKF)	$O (L^{2})$	L is the number of weights
[57]	high dimensional state estimation by Cubature Kalman Filters (CKF)	$O (n^{3})$	n is the number of state-vector dimensions