TR2021-090

Mixed-Integer Linear Regression Kalman Filters for GNSS Positioning


    •  Greiff, M., Berntorp, K., Di Cairano, S., Kim, K.J., "Mixed-Integer Linear Regression Kalman Filters for GNSS Positioning", IEEE Conference on Control Technology and Applications (CCTA), August 2021.
      BibTeX TR2021-090 PDF
      • @inproceedings{Greiff2021aug,
      • author = {Greiff, Marcus and Berntorp, Karl and Di Cairano, Stefano and Kim, Kyeong Jin},
      • title = {Mixed-Integer Linear Regression Kalman Filters for GNSS Positioning},
      • booktitle = {IEEE Conference on Control Technology and Applications (CCTA)},
      • year = 2021,
      • month = aug,
      • url = {https://www.merl.com/publications/TR2021-090}
      • }
  • MERL Contacts:
  • Research Areas:

    Control, Dynamical Systems, Signal Processing

Abstract:

In this paper, recursive filters are formulated for the mixed-integer GNSS receiver estimation problem, where the integer variables come from the ambiguities in the carrier-phase measurements. Insights from the linear setting illustrate pitfalls in designing optimal recursive filters, motivating a relaxation of the original optimization problem and a departure from conventional methods. A set of filters are developed for sequential nonlinear mixed-integer estimation based on statistical linearization, entertaining two estimate densities and taking the time-evolution of the ambiguities into account by adapting the process noise covariance based on a statistical model. Numerical examples illustrate the efficacy of the proposed algorithms.