Filtration of stochastic processes for continuous and discrete observations with memory. II: Synthesis of filters (Q1287249)

The authors use the main equation of nonlinear filtration derived in Part I [ibid. 56, No. 9, Pt. 1, 1241-1249 (1995), resp. ibid. 1995, No. 9, 49-59 (1995)] of this paper. They solve the problem of filter synthesis, using the a posteriori Gaussian distribution of filtered processes. They consider the case of discrete-continuous channels of observations and the memory of arbitrary multiplicity which is denoted as \(N\). A case of rare discrete observations is considered, when the form of differential equation and the realization of the filter are essentially simplified. As a particular case, the efficiency of discrete channel with \(N=2\) is investigated in comparison with the channel having memory of multiplicity \(N=1\).