Search

Due to rapid urban development, many indoor and semi-indoor structures, such as underground malls, metro stations, and bus-stops, have emerged. Traditional GNSS outdoor navigation fails in these locations. To meet urban citizens' needs, we require sustainable ubiquitous localization solutions, leveraging existing WiFi infrastructure. In such applications, processing data at the edge within constrained environments is imperative. Many prior studies neglect the practical challenge of implementing machine learning-based edge device localization, as they assume reliance on cloud servers. Selecting well-distributed data instances that preserve location class boundaries is essential for effective training in such conditions. In this work, we propose an instance selection approach that is modeled by applying a modified Binary Particle Swarm Optimization (BPSO) technique. The k-differentiating neighbor-based measure is incorporated to determine instance hardness while exploring the solution space through BPSO. The work is implemented on three publicly available benchmark datasets that are collected from two university campuses and one shopping mall. The proposed work is found to have achieved an instance reduction of 35% with only 1%–2% decrease in the accuracy measurement and an appreciable error deviation metric of 2.78 m. • Exploratory analysis on fingerprint instances to detect superfluous data. • BPSO combined with nearest neighbor analysis for effective instance selection. • 35%–40% instance reduction with decreased localization error deviation. • The approach is compared with state-of-the art undersampling approaches. [ABSTRACT FROM AUTHOR]

For WiFi-based indoor localization, optimal selection of features leads to the increased perceptibility of the localization procedure. It is essential to capture the important sets of Access Points (APs) that best defines the floor map for the positioning process. To maintain sustainable localization, the selection of APs enables scaling the solution and reducing the maintenance cost. In the present work, our contribution is twofold- the power of Particle Swarm Optimization is utilized for the selection of important APs. Then, a feature-based ensemble model is designed for the selected subsets of APs to retain the generality of localization performance. The base learners capture the different ambiance in the training and testing process. Extensive experimentation was carried out using the collected dataset from multiple smartphone devices. The proposed feature selection and training pipeline has also been tested with two popular benchmark datasets- UJIIndoorLoc and JUIndoorLoc. Results indicate that the proposed feature-based ensemble model could achieve 86%–96% accuracy with around 50%–65% reduction in APs for the datasets. The mean absolute error (MAE) indicates the distance between the predicted and actual location points. It is found to be 2.68 m, that is, neighboring location points, which is quite acceptable for user localization in indoor spaces. [ABSTRACT FROM AUTHOR]