Filtering of stochastic processes in the case of continuous-discrete observation channels with memory (Q1095101)

The author considers a nonlinear filtering problem where both discrete- time and continuous-time observations are available; moreove, he supposes that the observation at time t involves the value of the signal process x at time t and also its value at time \(t-t^*\) for some fixed delay \(t^*\). He derives the equation satisfied by the optimal filter: it is a stochastic partial differential equation with jumps at times of discrete observations. When the coefficients of the system are linear with respect to the signal process, a finite-dimensional equation for the conditional mean of \(x_ t\) is proved to hold and in a particular case, the behaviour of systems with and without memory are compared.