Multi-packet reception (MPR) has recently received attention for use in military wireless ad hoc networks because of the potential to greatly improve spatial reuse. However, MPR typically requires complicated hardware that makes it prohibitive to implement in many current military platforms. An alternative, cross-layer approach is to use an overlapped transmission technique, in which multiple transmissions can occur in the same geographical area because the communicators can use their knowledge of some of the interfering packets to recover the desired information. Examples of overlapped transmission techniques include overlapped carrier-sense multiple-access (OCSMA), physical-layer network coding, analog network coding, and Katti et al.'s COPE protocol. These techniques have much lower signal processing demands than MPR techniques and have also stimulated a lot of interest for use in military networks. Much of the previous research on overlapped transmission focuses on the physical-layer: how transmissions should be allowed to overlap, how the signals should be designed and detected, and the error probabilities under different conditions. These works tend to make optimistic assumptions about the availability of packets to perform overlapped transmission: often performance is evaluated for a simple three-node network under the assumption that traffic is always available at each of the edge nodes. In this paper, we explain why these protocols may suffer under real-world traffic conditions because of directional traffic flows and congestion control at the transport layer. We focus on how transmission control affects the performance of the OCSMA protocol. We propose modifications to the TCP parameters and the OCSMA protocol to improve the performance. We also show that OCSMA offers better fairness over the IEEE 802.11 MAC protocol in several scenarios.