1: \begin{abstract}

2: This paper proposes a dynamic primal-dual type algorithm to solve

3: the optimal scheduling problem in wireless networks subject to

4: uncertain parameters, which are generated by stochastic network

5: processes such as random packet arrivals, channel fading, and node

6: mobilities. The algorithm is a generalization of the well-known

7: max-weight scheduling algorithm proposed by Tassiulas \emph{et al.},

8: where only queue length information is used for computing the

9: schedules when the arrival rates are uncertain. Using the technique

10: of fluid limits, sample path convergence of the algorithm to an

11: arbitrarily close to optimal solution is proved, under the

12: assumption that the Strong Law of Large Numbers (SLLN) applies to

13: the random processes which generate the uncertain parameters. The

14: performance of the algorithm is further verified by simulation

15: results. The method may potentially be applied to other applications

16: where dynamic algorithms for convex problems with uncertain

17: parameters are needed.

18: \end{abstract}