Enhancing Indoor Visible Light Localization Accuracy through Hybrid Fingerprint-KNN Scheme in IoT-enabled Smart Cities

Location-based services have become increasingly popular in the construction of smart cities as the Internet of Things (IoT) technology has advanced quickly. While satellite-based location services are useful in outdoor settings, they are ineffective indoors due to the lack of a GPS signal. As a result, many indoor localization methods and platforms have emerged that use alternative signals. Among these, fingerprinting localization has recently received interest for its encouraging outcomes. To further enhance positioning accuracy, a hybrid scheme combining Fingerprint and K-Nearest Neighbour (KNN) models has been proposed. The performance of this proposed hybrid scheme has been compared with another hybrid scheme combining Voronoi and KNN. The positioning error was simulated at all receiver positions using Trilateration and these hybrid schemes. The results show that the average positioning errors for Trilateration, the hybrid Fingerprint-KNN, and the hybrid Voronoi-KNN are 0.0548m, 0.0373m, and 3.5357m respectively. Hybrid Fingerprint-KNN has a slightly higher average error but its error range is much narrower, with a maximum error of only 0.1803m as compared with maximum error achieved using Voronoi-KNN. This indicates more consistent performance with less variability and no extreme outliers using Fingerprint-KNN.