Since there is no difference between uplink and downlink subframes in the IEEE 802.16 mesh mode, both downlink and uplink traffics are transmitted within a single time frame. Hence, in most scheduling methods, only one scheduling algorithm is used for both uplink and downlink traffics. However, because of different characteristics of uplink and downlink traffic, different scheduling methods should be used for each of the traffic types. In this paper, we focus on the mesh centralized scheduling of downlink flows in the data subframe. After comprehensive analysis on the characteristics of downlink traffic, we propose a new algorithm called Traffic-aware Scheduling (TAS). In this algorithm, the downlink traffic distribution is tuned for maximum concurrent transmission rate. This goal is achieved by choosing different link selection criteria such as maximum demandant sender, farthest receiver and least interfered path based on the downlink traffic analysis results. The simulation results show that our algorithm outperforms existing methods in terms of scheduling length, link concurrency, and throughput (about 13.7 % in average) for the downlink traffic. Moreover, the proposed algorithm is scalable. In particular, in average, 3 % improvement is achieved in terms of scheduling length in higher traffic loads and number of nodes.