Application of a resampling scheme to solve the divergence in the pathwise filter
World Congress, Volume # 16 | Part# 1
Alexsandro Machado Jacob; Takashi Yoneyama
Digital Object Identifier (DOI)
nonlinear filtering,pathwise filter,Monte Carlo approximation,simulation of SDEs
A Monte Carlo-based approach to filtering for nonlinear systems based on the Pathwise theory was proposed by M. H. A. Davis in 1981. The discrete-time Markov chain used to compute the solution of a Fokker-Planck equation whose coefficients were determined by the observed process was here replaced by the simulation of an equivalent stochastic diffierential equation in order to make the filter implementation more clear and with less computational cost. This paper shows that the Pathwise filter for an one-dimensional Ornstein-Uhlenbeck state process with saturation in the observation has an interesting characteristic of divergence in this estimates when the signal-to-noise ratio on the state equation is low. Rewriting the filtering solution in terms of observation-based weights, it is presented that the low performance of the filter can be preliminary explained by the sudden increase of the weight variances. To solve this problem, a resampling scheme using the effective number of particles was used to smooth or, at least, maintain the weight variance controlled.
 Clark, J. M. C. (1978). Communication Systems and Random Process Theory. Chap. The design of robust approximations to the stochastic differential equations of non-linear filtering, pp. 721-734. Sijthoff and Noordhoff.  Clark, J. M. C. and D. Crisan (2005). On a robust version of the integral representation formula of nonlinear filtering. Probability Theory and Related Fields. (to appear).  Davis, M. H. A. (1981a). New approach to filtering for nonlinear systems. IEE Proc. D, Control Theory and Appl. 128(5), 166-172.  Davis, M. H. A. (1981b). Stochastic systems: the mathematics of filtering and identification and applications. Chap. Pathwise non-linear filtering, pp. 505-528. D. Reidel.  Del Moral, P. and L. Miclo (2000). Séminaire de Probabilités XXXIV. Chap. Branching and interacting particle systems approximations of Feynman-Kac formulae with applications to non-linear filtering, pp. 1-145. Vol. 1729 of Lecture Notes in Mathematics. Springer-Verlag. Berlin.  Doucet, A. (1998). On sequential simulation-based methods for bayesian filtering. Technical Report CUED/F-INFENG/TR-31. University of Cambridge. Cambridge.  Jacob, A. M., E. M. Hemerly and T. Yoneyama (2004). Improvements on Davis' Monte Carlobased nonlinear filtering. In: Congresso Brasileiro de Automática (CBA). Gramado.  Kallianpur, G. and H. Striebel (1968). Estimations of stochastic processes: Arbitrary system process with additive white noise observation errors. Ann. Math. Statist. 39, 785- 801.  Kloeden, P. E. and E. Platen (1999). Numerical solution of stochastic differential equations. 3rd ed.. Springer. Berlin.  Kushner, H. J. (1977). Probability methods for approximations in stochastic control and elliptic equations. Academic Press. New York.  Le Gland, F. (1984). Monte carlo methods in nonlinear filtering. In: Proceedings of the 23rd IEEE Conference on Decision and Control. Las Vegas.  Lipster, R. S. and A. N. Shirayaev (1977). Statistics of randon processes I: general theory. Springer-Verlag. New York.  O'Loghlen, J. W. and G. C. Wright (1982). Computer results for new nonlinear filtering algorithm. IEE Proc. D, Control Theory and Appl. 129(2), 70-71.  Pardoux, E. (1981). Stochastic systems: the mathematics of filtering and identification and applications . Chap. Non-linear filtering, prediction and smoothing, pp. 529-557. D. Reidel.  Souza, E. T. de (1992). Métodos numéricos para filtragem não-linear. Ph.d. thesis. Instituto Tecnológico de Aeronáutica. São José dos Campos.  Zakai, M. (1969). On the optimal filtering for diffusion process. Z. Wahrsch. Verw. 11, 230- 243.