Design And Implementation Of Algorithm And Platform About Multi Aerial Base Station Deployment And Computation Offloading

With the advent of the 5G network era and the popularity of various intelligent mobile terminals,the demand for communication and computing from intelligent services such as augmented reality and virtual reality has increased significantly.This growth places high demands on computing power,requiring more computing power to ensure the availability and service quality of various high computing power intelligent services for users.Mobile edge computing(MEC),as an emerging computing model,can transfer computation and storage resources from the cloud to edge nodes closer to user devices,reducing data transmission latency and network load,improving network response speed and efficiency,and ensuring the real-time and stability of applications.Aerial base stations equipped with MEC servers are a type of freely movable edge node that is more adaptable to diverse application scenarios and have advantages that traditional ground base stations and MEC servers do not have.Aerial base stations equipped with MEC can provide fast communication and local computing power enhancement services in hotspots or natural disasters where there is a shortage of computing facilities.When traditional ground base stations cannot meet explosive computing power demands or are unavailable,deploying aerial base stations is a quick and effective solution to achieve localized computing power enhancement,as users can offload computing tasks to the aerial base stations.However,due to the limited endurance,computing power,and communication range of the unmanned aerial vehicles(UAVs)acting as aerial base stations,uneven user location distribution,and the fact that the locations of aerial base stations directly affect the channel conditions between aerial stations and users,which in turn affects the offloading strategy,the deployment of aerial base stations and computation offloading problem are strongly coupled.Therefore,studying the problem of multi aerial base station deployment and computation offloading is of great significance.Firstly,various models for multi aerial base station deployment and computation offloading scenarios,including models for delay,energy consumption,communication,and offloading are researched in this paper.The research status and existing problems in this field are also studied both domestically and internationally.Based on research above,a decisionmaking algorithm for multi aerial base station deployment and computation offloading was proposed,which was based on differential evolution and multi-agent reinforcement learning.To improve the practicality of the algorithm and increase its convergence speed,a predeployment of all aerial base stations is first carried out based on differential evolution algorithm.Then,the deployment locations are optimized using a neural network to generate an offloading strategy.Simulation results show that the proposed algorithm can make reasonable decisions for multi aerial base station deployment and computation offloading.Computation task delay,energy consumption,and throughput metrics are effectively improved.Based on the algorithm proposed,this paper designs and implements a multi aerial base station deployment and computation offloading experimental platform.Firstly,various requirements in aerial base station computation offloading scenarios are analyzed,and then the overall architecture of the experimental platform is designed,and the detailed design of the sub-modules are provided,followed by coding and implementation based on the design plan.Finally,the functional testing of the experimental platform is conducted to verify that the system realizes the computation offloading and UAV deployment functions,can effectively visualize the algorithm results,and evaluate the algorithm performance,which is beneficial to the relevant research in aerial base station deployment computation offloading.In summary,this article proposes a method for the deployment and computation offloading of multiple aerial base stations based on differential evolution and MADQN(Multi Agent Deep Q Network),which improves the quality of service for users.A multi-aerial base station deployment and computation offloading experimental platform is designed and implemented,which can effectively support relevant algorithm research in this field,making aerial base stations better adapted to various complex practical application scenarios and providing important technical support for better meeting users’ computational needs.