A Hybrid Model Based on Constraint OSELM, Adaptive Weighted SRC and KNN for Large-Scale Indoor Localization

150 150 MIMOS Berhad


Hengyi Gan, Mohd Haris Md Khir, Gunawan Witjaksono Djaswadi and Nordin Ramli;



In this paper, a novel hybrid model based on the constraint online sequential extreme learningmachine (COSELM) classier with adaptive weighted sparse representation classication (WSRC) and knearest neighbor (KNN) is proposed for the WiFi-based indoor positioning system. It is referred to as A Fast-Accurate-Reliable Localization System (AFARLS). AFARLS exploits the speed advantage of COSELMto reduce the computational cost, and the accuracy advantage of WSRC to enhance the classicationperformance, by utilizing KNN as the adaptive sub-dictionary selection strategy. The understanding isthat the original extreme learning machine (ELM) is less robust against noise, while sparse representationclassication (SRC) and KNN suffer a high computational burden when using the over-complete dictionary.AFARLS unies their complementary strengths to resolve each other’s limitation. In large-scale multi-building and multi-oor environments, AFARLS estimates a location that considers the building, oor, andposition (longitude and latitude) in a hierarchical and sequential approach according to a discriminativecriterion to the COSELM output. If the classier result is unreliable, AFARLS uses KNN to achieve the bestrelevant sub-dictionary. The sub-dictionary is fed to WSRC to re-estimate the building and the oor, whilethe position is predicted by the ELM regressor. AFARLS has been veried on two publicly available datasets,theEU Zenodoand theUJIIndoorLoc. The experimental results demonstrate that AFARLS outperforms thestate-of-the-art algorithms on the former dataset, and it provides near state-of-the-art performance on thelatter dataset. When the size of the dataset increases remarkably, AFARLS shows that it can maintain itsreal-time high-accuracy performance



IEEE Access V7 2019