UAV-based Resource Allocation Algorithm For D2D Communication

5G wireless communication system completely commercialized today,D2 D communication as an important technology in the system,it allows the user to distance is close without forwarding data,the BS to communicate directly,greatly down the time delay and up the throughput,the BS thus received extensive attention of the researchers.However,when D2 D users reuse the cellular channel,it will bring the same frequency interference to the cellular users and the base station.At the same time,the communication scenarios and requirements are increasingly diversified,and the traditional communication scenarios cannot meet the communication requirements under some special circumstances,such as the base station damaged by disasters or subjected to strong ground electromagnetic interference.In the face of the above problems,it is necessary to adopt a reasonable resource allocation algorithm to reduce the influence of interference on communication quality,ensure the communication quality of cellular users and maximize the system throughput.The base station of unmanned aerial vehicle(UAV),with its advantages of flexible and convenient deployment and significant line-of-sight link(LOS)ratio,which is not easily subject to ground interference,expands the scope of application scenarios of D2 D communication,and is of great significance to improve the system performance.Based on the UAV base station model,this paper constructs two different application scenarios,designs different resource allocation schemes respectively,discusses the influence of model and algorithm parameters and draws corresponding conclusions.The specific work is as follows:Firstly,the network architecture of UAV base station and the mode selection and resource reuse method of D2 D communication technology are deeply studied,and the channel gain model of UAV base station is analyzed with emphasis,and the system model is established under this application background.Secondly,the optimization resource allocation problem of optimization algorithm is proposed for the scenarios in which the system has few users and high demand for stability.Under the condition that a small number of users are randomly distributed in a single cell,the algorithm compares the users in the system to the "particles" in the bird population.During the iteration process,the speed and position are constantly updated,and the optimal solution of the system is obtained after the termination decision condition is satisfied.The base station makes centralized decision according to the system state.The simulation results show that compared with the ground base station,the throughput of the UAV-BS can be significantly improved.Thirdly,a strategy scheme based on reinforcement learning Q-learning algorithm is designed in view of the frequent location change of system users and the need for long-term optimization.In the particle swarm optimization algorithm,the optimal solution of the system is only for the current moment.When the environment of the system changes dynamically,the optimal solution needs to be recalculated,and each recalculation will bring additional state switching cost.When there are more users in the system,the system’s overhead will increase significantly.The introduction of reinforcement learning maximizes the long-term throughput of the system.Finally,through simulation analysis and adjustment of system parameters for comparison,we show that compared with the ground BS,the throughput of the UAV base station is significantly improved.After the introduction of reinforcement learning and many time tries,the system can up to the certain solution,Q-learning can make the throughput up in a long-term way when the system environment changes.