Stochastic processes
On estimation of discrete processes under multiplicative and additive noise conditions
Information Sciences: an International Journal
Simultaneous optimum detection and estimation of signals in noise
IEEE Transactions on Information Theory
Optimal recursive estimation with uncertain observation
IEEE Transactions on Information Theory
Brief paper: Finite-state, discrete-time optimization with randomly varying observation quality
Automatica (Journal of IFAC)
Automatica (Journal of IFAC)
Correspondence item: Authors' reply
Automatica (Journal of IFAC)
On identification and adaptive estimation for systems with interrupted observations
Automatica (Journal of IFAC)
| Hi-index | 22.15 |
This paper is concerned with determining an optimal control policy in situations where at any time it is possible that information concerning the state vector of the system may or may not be contained in the observations-''interrupted stochastic control problem''. The interrupted observation mechanism is formulated in terms of two-state Markov chain with states 0 and 1. A Separation Theorem is established for discrete-time linear system with interrupted observations and an expected quadratic cost. The optimal policy is realized by cascading a nonlinear estimator, which computes the conditional mean of the state vector, with the optimal feedback gain matrix in which all uncertainties are removed. This does not require the use of dynamic programming and has much computational advantage. The nonlinear estimator consists of a weighted sum of Kalman type filters. The performance measure is evaluated and it is pointed out that the probability of interruption enters into the expected cost only as a function of the estimation error covariance matrix.