A Hybrid Wi-Fi Fingerprint-Based Localization Scheme Achieved by Combining Fisher Score and Stacked Sparse Autoencoder Algorithms

Algorithm 1

Feature extraction algorithm Fisher score–stacked sparse autoencoder (Fisher–SSAE).

Input:
(1)	Training fingerprint data
(2)	Number of features selected by Fisher criteria
(3)	Maximum dimension of hidden layer
(4)	Maximum depth for SSAE
Output:
(1)	The structure SSAE including the dimension of hidden layer and depth
(2)	The fingerprint feature extracted
Calculate the Fisher score for each AP feature
Rank the Fisher score according to its value (large to small), and keep the features that correspond to the first k values
Set the initial depth of SSAE
Set the initial the dimension of hidden layer
t = 1
Repeat
	t = t + 1
	Calculate the accuracy of classification which is utilized to achieve subregional localization
Until or
Determine
t = 1
Repeat
	t = t + 1
	Calculate the accuracy of classification which is utilized to achieve subregional localization
Until or
Determine
Training fingerprint data by SSAE
Return the data in the hidden layer