Randomized digital optimal control

In sampled data systems the sampling periods are almost always assumed to be deterministic, i.e., known in advance. However, in practice, the sampling periods should often be conceived as stochastic variables. This so called stochastic sampling phenomena occurs in many areas of sampled data systems such as in biological control systems or when a human being selects the sampling instants. Data may be absent at sampling instants when the sampling is governed by a stochastic process like a radar or sonar echo, or when the sampling mechanism fails. Also technical imperfections in the instrumentation may cause stochastic sampling. Stochastic sampling may also be applied intentionally for instance when, for economy, a digital computer is time-shared in a stochastic manner, or to prevent jamming in communications. Other applications of intentional stochastic sampling are elimination of hidden oscillations between sampling instants, decreasing the influence of intelligent disturbances, and increasing stabilizability [1].