A Secure Received Signal Strength based Indoor Localization Algorithm Using OS-ELM and Clustering Analysis Technique

In this paper, a secure machine learning based indoor localization algorithm is proposed, when the received signal strength (RSS) measurement fingerprint based training data set is given by chunk-by-chunk and contains the attacked training data samples. In the off-line phase, the hierarchical clustering approach is proposed to distinguish the attacked training data and the attacked-free training data, firstly. By the above data pre-processing, the attacked RSS measurements is found and can be deleted. Then, the online sequential extreme learning machine (OS-ELM) algorithm is used to training the attacked-free data in turn. In on-line phase, according to the obtained RSS measurements, the obtained regression models are straightly used for final position estimation. Field tests are carried out to show the advantage of the proposed secure localization algorithm over traditional OS-ELM based approach.