In this article, we describe how a fleet of unmanned aerial vehicles (UAVs) can optimize communication performances by having its members independently change their orientations. This distributed solution, based on a hill-climbing approach, relies on information available locally at each node, namely the reception power of the received frames. The solution is evaluated using the ns-3 network simulator, whose source code is modified to be able to deal with nonisotropic antennas in the context of Wi-Fi networks, as well as simulate angular movements. As isotropic antennas are only theoretical objects, this step is mandatory in order to increase the realism of network simulations. The results, obtained using realistic antenna models, highlight that controlled mobility, in particular controlled orientation needs to be considered in order for UAV networks to provide better performances.