Coverage gaps in fingerprinting based indoor positioning: The use of hybrid Gaussian Processes

Indoor positioning based on the received signal strength (RSS) in wireless local area networks (WLAN) is one of the most promising approaches to provide Location-based services. Gaps in the coverage of the fingerprint can lead to significant errors. We propose a localization scheme that minimizes these faults. By using Gaussian Processes (GP) we are able to incorporate model knowledge and empirically measured data, with correct uncertainty handling and Bayesian parameter estimation. This approach leads to a hybrid localization technique, that outperforms several other procedures. We evaluate our method on two huge datasets, while focusing on measurement gaps in the available data. This provides a realistic and challenging scenario, compared to randomly selected missing data. We show that we are able to significantly reduce the localization error especially for increasingly sparse data sets.