A0510
Title: Centralized fusion prediction in hypercomplex systems with random packet dropouts under properness conditions
Authors: Rosa Maria Fernandez-Alcala - University of Jaen (Spain) [presenting]
Jose Domingo Jimenez-Lopez - University of Jaen (Spain)
Jesus Navarro-Moreno - University of Jaen (Spain)
Juan Carlos Ruiz-Molina - University of Jaen (Spain)
Abstract: The centralized fusion prediction problem is investigated for multi-sensor stochastic systems affected by multiple random packet dropouts. The packet dropouts are described as a sequence of independent Bernoulli random variables with known probabilities. The problem is tackled in the tessarine domain and examined under specific properness conditions linked to the vanishing of certain pseudo-correlation functions. Concretely, assuming that the state and the observations are jointly Tk proper, a linear least-mean-square centralized fusion prediction algorithm is devised for correlated state and observations noises. The proposed methodology leads to a dimensionality reduction in the processes involved, resulting in significant computational savings. A numerical example is given to show the implementation of the presented algorithm as well as its better performance over its counterpart in the quaternion domain in different uncertainty scenarios.
