OBLoc: Online Batch Localization for Large-Scale Indoor Environments

The demand for indoor localization is increasing with ubiquitous computing and the Internet of Things (IoT) emergence. Location awareness is crucial to intelligent systems in large-scale indoor environments, such as smart airports, train stations, hospitals, malls, and smart cities. Hence, indoor localization has been researched for the last decades. However, most site survey based WiFi fingerprint based indoor localization methods involve independent or one by one online localization methods, that are costly, and inefficient regarding computation time, memory, and energy consumption to implement in large-scale indoor environments. We propose an Online Batch Localization (OBLoc) for large-scale indoor environments scheme to address these challenges. The basic idea of OBLoc is that the indoor localization users close to each other have similar WiFi fingerprinting and can be clustered to form batches. Every batch's search space could be reduced using a batch representative. We propose a search space-slicing algorithm that slices part of the large-scale database using the appointed representative to minimize searching overhead. Furthermore, we designed a neighborhood consistency algorithm to identify erroneous location annotations in the sliced search space. We conduct experiments on three field-measured datasets to evaluate the performance of the proposed method. The experiment results show the improvement over the peer schemes of 21.34%, 17.38%, and 27.24% localization accuracy in the three datasets.