New approach to filtering for nonlinear systems
Michael Harold Davis · IEE Proceedings D Control Theory and Applications · 1981
The paper describes a recently developed reformulation of the optimal-filtering equations for a noisily observed diffusion process and discusses the computational implications. The computation involved is the solution of a parabolic partial differential equation whose coefficients are determined by the observed process. A Monte Carlo method of solution is proposed and given in detail as an example. It is argued that nonlinear filtering is now a practical proposition.