In this paper, we derive the parameters of an optimal leaky-bucket shaper for the transmission of constrained traffic over a wireless fading channel. The analysis is based on an efficient stochastic min-plus system theory for the derivations of second-order statistics, as a wireless fading channel is a stochastic rate-latency server. The shaper is optimal in the sense that it minimizes the end-to-end delay and buffer requirements under the constraints dictated by the lower layer characteristics, in particular the wireless channel. Our model is used to plot the analytic delay and buffer sizes in comparison to simulation results for various values of shaper's parameters, from which the optimal or sub-optimal solution is calculated.