Using Feedforward Neural Networks for Parameter Modeling of a 4G Link for Unmanned Aerial Vehicles

A prediction of link parameters based on positional information contributes to a safe operation of Unmanned Aerial Vehicles (UAVs). For logistical and economical reasons, it makes sense to establish the wireless link via the fourth generation mobile standard Long Term Evolution (LTE) as it offers an already existing and stable infrastructure. The usual non line of sight (NLOS) connection in such a scenario poses challenges in parameter modeling, for which artificial neural networks are shown to be a suitable solution in this paper. The connection quality between transmitter station and UAV is mainly described by the parameters Reference Signal Received Power (RSRP) and Reference Signal Received Quality (RSRQ), which are tried to be predicted via an Artificial Neural Network (ANN) in this work. This approach takes the UAVs position (either Cartesian or polar coordinates) as input and maps them to the respective parameter.