We have two accepted articles at the IEEE Workshop on Statistical Signal Processing (SSP ’18), 10-13 June 2018, Freiburg, Germany. Namely,
E. Arias-de-Reyna, D. Dardari, P. Closas, P. M. Djurić, “Estimation of Spatial Fields of NLOS/LOS Conditions for Improved Localization in Indoor Environments”
Abstract: A major challenge in indoor localization is the presence or absence of LOS. Many models account for the effect of LOS on localization of objects because the absence of LOS, denoted as NLOS, directly affects the accuracy of any localization algorithm. The estimation of LOS/NLOS fields in indoor environments remains a major challenge. In this paper, we propose a novel crowd-based Bayesian learning approach to the estimation of ranging bias caused by LOS/NLOS conditions. The proposed method is based on the concept of Gaussian processes and exploits numerous measurements. The performance of the method is demonstrated with extensive experiments.
J. Vilà-Valls, P. Closas, M. F. Bugallo, J. Míguez, “A Probabilistic Approach for Adaptive State-Space Partitioning”
Abstract: The multiple Bayesian filtering approach is based on the partitioning of the state-space in several lower dimensional subspaces, combined with a set of parallel filters that characterize the marginal subspace posteriors. This solution has been shown to perform well and solve some of the problems typically suffered by standard Bayesian filters, such as the curse-of-dimensionality, in some scenarios. An inherent problem in the application of multiple Gaussian filters (MGF) and multiple particle filters (MPF) proposed in the literature is how to partition the state-space. A closed answer does not exist because this is an application-dependent problem. In this contribution we further elaborate on the multiple filtering approach, and propose a probabilistic adaptive state-partitioning strategy based on the cross-correlation computed at each filter.