论文部分内容阅读
In 2020,mobile communications is considered to be one of the fastest growing parts of the communications industry.At present,the society is about to enter the stage of full popularization of 5th generation mobile networks(5G)from business to daily life.With the advent of the 5G network era and the popularity of various user terminals,image,audio,video,and enhancement for large-scale users Data services such as Augmented Reality(AR)and Virtual Reality(VR)have grown dramatically,and mobile data traffic will explode.This phenomenal growth requires that the capacity of mobile networks needs to be greatly increased to ensure the availability of data to users.Quality of communication services.When the existing traditional ground base stations are insufficient to meet the burst traffic demand or are unavailable,deploying aerial base stations is a fast and effective solution to achieve network capacity enhancement.How to plan the optimal 3-Dimension(3D)position of the aerial base station according to the user’s business needs and service scenarios is a key issue that needs to be urgently solved.At present,the conventional optimization algorithms that solve this problem have high time complexity and it is difficult to utilize past experience.The deep reinforcement learning model can be trained through historical experience feedback to quickly iterate to obtain the optimal solution,so it is suitable for solving the best 3D location planning problem of aerial base stations.Based on this problem,this paper proposes a location planning method of aerial base station based on Deep Q-Network(DQN)algorithm.Firstly,the aerial to ground path loss model and resource allocation model are established.Secondly,the location planning optimization problem of aerial base station is modeled.Then,the location planning method based on dqn is designed.Then,multiple aerial base stations are planned The deployment of base station puts forward the design scheme.Finally,the simulation experiment is carried out by Python+ Tensorflow.After the aerial base station is deployed in the planned location,the average spectrum efficiency of the whole system is effectively improved.At the same time,this topic also builds a SpringBoot-based aerial base station location planning experimental platform.First,the simulation system is analyzed for multi-dimensional requirements.Secondly,the overall architecture of the simulation system is designed.Then the sub-modules are given.The detailed design scheme of the simulation system.Finally,a functional test was performed on the simulation system to verify that the system realized the generation of simulation scenarios.It could effectively visualize the algorithm results and evaluate the effects of the algorithms,which was helpful for related research.Authors verify their own algorithms on the system,which helps to quickly establish a stable and reliable aerial base station communication network.In short,this topic proposes an aerial base station location planning method based on the DQN algorithm,which realizes the establishment of an aerial base station location planning simulation system,which can effectively improve the actual utility of the aerial base station and the overall network performance,and make the aerial base station better adapt to various Complex practical application scenarios form an important technical support for the construction of 5G agile networks and self-optimizing networks.