2014/04/14 by Leila Pishdad, Pishdad, Leila, Fabrice Labeau +1
Computer Science · Engineering · #FOS: Electrical engineering #Indoor and Outdoor Localization Technologies #Structural Health Monitoring Techniques #Systems and Control (eess.SY) #Target Tracking and Data Fusion in Sensor Networks #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1404.3638
openalex publication_date 2014/04/14 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
In this work we propose an approximate Minimum Mean-Square Error (MMSE)\nfilter for linear dynamic systems with Gaussian Mixture noise. The proposed\nestimator tracks each component of the Gaussian Mixture (GM) posterior with an\nindividual filter and minimizes the trace of the covariance matrix of the bank\nof filters, as opposed to minimizing the MSE of individual filters in the\ncommonly used Gaussian sum filter (GSF). Hence, the spread of means in the\nproposed method is smaller than that of GSF which makes it more robust to\nremoving components. Consequently, lower complexity reduction schemes can be\nused with the proposed filter without losing estimation accuracy and precision.\nThis is supported through simulations on synthetic data as well as experimental\ndata related to an indoor localization system. Additionally, we show that in\ntwo limit cases the state estimation provided by our proposed method converges\nto that of GSF, and we provide simulation results supporting this in other\ncases.\n