Abdur Rauful Jamil Nibir

Deep Reinforcement Learning to Employ Deep Neural Network for UAV Supported Vehicular Networks

Modelling a UAV network is very challenging because of various dynamic network requirements such as determining a coverage area for the connected UAVs, fluctuating energy levels, dynamic antenna positioning, services types and traffic requirements. . Prioritizing service delivery to UAVs based on their geo-locations is an important challenge in such stochastic environment which requires an intelligent decision making mechanism to manage their movements.

The Markov decision process (MDP) can be formulated to maximize the output with the restraints of the complete transmission. Deep deterministic policy gradient (DDPG) is applied to the deep reinforcement learning technique which is capable to solve the MDP problems. It can have drastic improvement over the transmission control and the 3D flight control of the UAVs. At last, broad simulation results to show the viability of the proposed arrangements differentiated along with two benchmark plans.