Localization of unmanned aerial vehicle (UAV) and unmanned water vehicle (UWV) base stations to provide maximum coverage on terrestrial and underwater networks

Nowadays, due to technological developments, unmanned vehicles have started to be used in specific applications in many areas such as telecommunication, underwater research, civilian, and defense. In these applications, it is aimed to keep the losses that may occur in data transmission between devices (sensor, mobile device, data storage center, etc.) at a minimum level. Therefore, the use of unmanned vehicles with high mobility as a base station has been an alternative solution. In this study, the placement of Unmanned Aerial Vehicle Base Station (UAVBS) and the placement of Unmanned Water Vehicle Base Station (UWVBS) were investigated to cover the maximum number of users or underwater sensors for the improvement of both terrestrial and underwater network. In the first part of our study, machine learning approach is proposed for multiple UVABS's placement problem. Required data that are used on machine learning models are generated by using a Mixed Integer Non-linear Programming (MINLP) optimization model of multi UAVBS placement problem. In the machine learning approach; Convolutional Neural Networks (CNN), Residual Neural Networks (ResNet), Deep Neural Networks (DNN) and K-Means algorithms were used respectively. The results are compared with the MINLP model in terms of user coverage. It is observed that the ResNet model gives better results than other machine learning models. In the second part of our study, we defined a MINLP optimization model for a single UWVBS placement to maximize the sensor coverage. The results of the optimization model were compared with the Benchmark Location in terms of a different number of underwater sensors and transmission power levels. For multiple UWVBS's deployment scenario, the optimization model is re-evaluated and compared with KMeans algorithm.