1: \begin{abstract}

2: We consider the optimal power management in

3: renewable driven smart building MicroGrid under noise corrupted

4: conditions as a stochastic optimization problem. We

5: first propose our user satisfaction and electricity consumption

6: balanced (USECB) profit model as the objective for optimal power

7: management. We then cast the problem in noise corrupted

8: conditions into the class of expectation maximizing in stochastic

9: optimization problem with convex constraints. For this task, we

10: design a Bregemen projection based mirror decent algorithm

11: as an approximation solution to our stochastic optimization

12: problem. Convergence and upper-bound of our algorithm with

13: proof are also provided in our paper. We then conduct a broad

14: type of experiment in our simulation to test the justification of

15: our model as well as the effectiveness of our algorithm.

16: \end{abstract}

17: