A particle filter using SVD based sampling Kalman filter to obtain the proposal distribution

In this paper, we propose a novel particle filter (PF), which uses a bank of singular-value-decomposition based sampling Kalman filters (SVDSKF) to obtain the importance proposal distribution. This proposal has two properties. Firstly, it allows the particle filter to incorporate the latest observations into a prior updating routine and, secondly it inherits advantage of having good numerical stability from the singular-value-decomposition (SVD). The convergence results of the numerical simulations we made confirm that the proposed PF method outperforms the standard bootstrap PF as well as other local linearization based PFs.