1: \begin{abstract}

2: Motivated by applications arising in networked systems,

3: this work examines controlled regime-switching systems that stem from

4:  a mean-variance formulation.

5: A main point is that

6: the switching process is a hidden Markov chain. An additional

7: piece of information, namely, a noisy observation of switching process corrupted by

8: white noise is available.

9: We focus on

10:  minimizing the variance subject to

11: a fixed terminal expectation. Using  the Wonham filter, we convert

12: the partially observed system

13:  to

14: a completely observable one first.

15: Since closed-form solutions are virtually impossible be obtained, a Markov chain approximation method is used to devise

16: a computational scheme. Convergence of the algorithm is obtained.

17: A numerical example is provided to demonstrate the results.

18: 

19: \vskip 0.2 true in

20: \para{Key Words.} Mean-variance control, numerical method, Wonham filter.

21: \end{abstract}

22: