Research On Optimization Techniques For Cooperative Computing Based UAV-MEC Networks

Currently,the rapid development and commercial deployment of mobile communication technology are driving continuous changes in mobile technology,giving rise to many emerging applications such as virtual reality,target recognition and tracking,and autonomous driving.These computation-intensive and latency-sensitive applications have spurred the rise of mobile edge computing(MEC)technology.However,traditional MEC faces significant challenges in addressing the growing network computing demands due to its inflexible resource deployment.On this background,the unmanned aerial vehicle(UAV)enabled MEC technology emerges at the right moment and receives a wide attention.Compared to the traditional MEC technology,UAV-enabled MEC technology utilizes the UAV’s mobility and controllability,and provides communication and computation services for the network terminals,by combining the technologies of UAV communication and edge computing.According to the network structures,UAV-enabled MEC can be classified into two types,i.e.,the cooperative computing and the non-cooperative computing.Compared to the non-cooperative UAV-MEC computing networks,the computation resource of the cooperative UAV-MEC computing networks is dispersedly deployed on multiple computing nodes.This manner provides a more flexible and strong computational force.Although the UAV-enabled MEC owns many advantages,it is still facing the following crucial problems:1)the payload of the UAV and the energy supply of the terminal devices are limited,which result in the scarce system’s energy;2)the two key performance indicators,i.e.,computation capacity and energy consumption,are hard to be balanced,which result in the uncontrollability of the network energy efficiency;3)the computation task offloading suffers from the complex transmission environment and the random channels,which limites the network’s computing ability;and 4)the network sevices are generally latency-sensitive and the computation latency problem is severe,especially for the case with one single UAV and multiple users,under which the strict requirement for the latency is difficult to be guaranteed.The dissertation carries out the research on joint optimization techniques for cooperative computing based UAV-MEC networks,which focuses on solving the aforementioned several crucial problems and challenges,by building the different cooperative computing networks and proposing the corresponding optimization algorithms.The main research work and innovation of this dissertation are presented as follows:(1)As for the energy consumption problem of the UAV-enabled MEC networks,an air-ground cooperative computing network is established,and an air-ground cooperative energy-controlled joint multi-dimensional resource and trajectory optimization algorithm is proposed.In this scenario,the UAV and the ground computing node are considered,and the UAV is used to be a computing node and mobile relay,which means that the UAV can deliver the comutation task to the remote ground access point where the MEC computing resource is deployed,while executing the local computing.The models for the UAV’s mobility and energy consumption are built,and the causality condition of the system for data receiving,forwarding,and computing is given.The goal of the minimum system energy consumption is achieved by designing the task offloading strategy of the ground terminals,the UAV’s relaying strategy,and the UAV trajectory.(2)As for the energy efficiency problem of the UAV-enabled MEC networks,an air-ground cooperative computing network with multiple ground computing nodes is established,an air-ground cooperative energyefficient joint multi-dimensional resource and trajectory optimization algorithm is proposed.In the proposed network,the UAV cooperates with nultiple ground MEC computing nodes,and the ground terminals can choose any computing nodes for task offloading.The computational capacity model and energy consumption model of the network are established,and the task offloading strategy in each time slot is designed.Also,the procedure of computational results being sending back is considered.The goal of the maximum system energy efficiency is achieved in the premise of eneuring the minimum computation requirement for the ground terminals,by determining the offloading power of the ground terminals,time slot division,bandwidth allocation,computation allocation of the computation nodes,and the UAV trajectory design.(3)As for the complex transmission environment problem of the UAV-enabled MEC networks,a reconfigurable intelligent surface(RIS)based air-ground cooperative computing network is established,and an airground cooperative environment-reconfigured joint multi-dimensional resource and trajectory optimization algorithm is proposed.In the network,the UAV simultaneously acts as a computation node and a mobile relay to assist task computing and data forwarding for the ground terminals.Besides,the task transmission qulity is enhanced by exploiting the RIS technology to reconfigure the wireless transmission environment.The RISbased transmission model is built,and the communication protocol in time slots is designed.The goal of maximizing the minimum computational capacity of the ground terminals is achieved,by determining the UAV’s proactive beamforming,RIS’s passive beamforming,time slot division,and UAV trajectory.In order to evaluate the performance of the proposed algorithm,a direct algorithm is designed by analyzing the problem characteristics,based on which the superiority of the proposed algorithm is verified.(4)As for the computation latency problem of the UAV-enabled MEC networks,a multi-UAV cooperative computing network is established,and a multi-UAV cooperative joint multi-dimensional resource and trajectory optimization algorithm is proposed.In the network,to overcome the disadvantages of the traditional fixed total time slots and slot length,an optimized design approach for the total time slots and the time slot length is proposed.In addition,both the partial offloading mode and the binary offloading mode are considered.Under the two modes,the optimization problems are formulated,with the goal of task completion time minimization.Considering the different problem structures under the two modes,the problem of partial offloading mode is firstly solved.On the basis,the penalty method is introduced to tackle the 0-1 constraint in the problem of the binary offloading mode.Finally,the corresponding optimization algorithms are proposed,and the optimal number of time slots,time slot length,offloading coefficient,computation resource allocation,and UAV trajectory are obtained,which significantly reduce the system’s task completion time.