Conditionally optimal iterative filters are proposed which are built v
p on the basis of the theory of conditionally optimal discrete filters
intended for filtering random processes described by nonlinear differ
ence equations. As in the calculations involved with the theory of con
ditionally optimal filtering, here all the calculations for finding th
e gains of iterative filters rely only on the use of a priori informat
ion on both the structure of the difference equations that determine r
andom processes and a class of admissible filters. The calculations do
not use the results of current observations and thus they can be perf
ormed beforehand in designing the filters. The filtering process in it
self can be carried out in real time. An example is given.