In time varying wireless networks, the state of communication channels are subject to random variations, and need to be estimated for satisfactory rate allocation. Yet, the estimation process reduces the resources available for the rate allocator, and also possesses inaccuracies in its channel state estimates. The objective of this work is twofold: to understand the optimal rate allocation under probabilistically defined uncertainties of given estimators; and to include the cost of estimation into the design of joint estimation-rate allocation policies. We will present our preliminary results in addressing these two points.