Network coding techniques are usually applied upon network layer protocols to improve throughput in wireless networks. However, the packet scheduling algorithm has long been neglected. With proper scheduling, bottleneck nodes can clear their packet queue faster than a vanilla round-robin scheduling algorithm. Therefore, a higher Coding+MAC gain can be achieved. In this paper, we formulate the scheduling problem and propose a search algorithm to find the optimal solution. A pruning rule and an approximation scheme are proposed to reduce the amount of search branches, and hence to improve the time performance. Results show that the algorithm can improve the throughput gain as compared to round-robin scheduling. The throughput gain is most conspicuous when the network is almost but not saturated.