Generalized UAV Deployment Design for UAV-Assisted Wireless Networks

To improve unmanned aerial vehicles (UAVs)-assisted wireless networks, the appropriate deployment of UAVs is critical. However, as existing deployment approaches are commonly based on specific models that cannot be easily generalized, we in this paper propose a generalized deployment approach for UAV-assisted wireless networks by combining the deep neural network (DNN) based surrogate model with a zeroth-order optimization (ZOO). The design of the ZOO is presented. Furthermore, since the accuracy of the surrogate model is critical, we discuss its design and update approaches. We conduct practical simulations to evaluate our proposed approach. Results show that our proposed approach can significantly outperform all the reference schemes.