MEASUREMENT OF THE CHARACTERISTICS OF STOCHASTIC PROCESSES ON THE BASIS OF EFFECTIVE ESTIMATES.

The problem of determining the characteristics of a stochastic process (SP) from an observed realization of the SP is considered. The model of the SP to be processed is taken to be a Gaussian nonstationary SP (NSP). Optimum filtration methods are used to obtain a recursive computational algorithm for ML estimates that is capable of being implemented in digital computers when SP are processed in stochastic control systems. A flow chart of the algorithm and examples are given. To enhance the operating speed for the case of processing SP in real time, the feasibility of implementing the algorithm by an analog-to-digital device is examined.