Research On The Resource Allocation Of Edge Computing And Communication For IoT

With the vigorous development of Internet of Things(Io T)technology,the number of mobile terminals connected to the network and the data traffic increases sharply constantly challenging the capacity limit of wireless networks.Meanwhile,a large number of new applications are emerging with the development of 5G technology,such as autonomous driving and augmented/virtual reality.These new applications not only have high bandwidth requirements,but also have high requirements for data storage,computing and latency.However,limited storage and computing resources and battery capacity of mobile devices are difficult to meet the requirements of high reliability and low delay for the computation-intensive tasks of new applications.To solve these problems,mobile edge computing(MEC)proposes to extend cloud services to the edge of the network,reduce the delay of task offloading,and provide computing and locally processing capabilities for mobile devices.In the MEC,many mobile devices share the resources of the edge server.However,the computing resources and communication resources of the edge server are limited.Each user needs to make a decision on whether to perform computing offloading,and reasonably allocate the computing and communication resources in the network to give full play to the advantages of MEC network.Secondly,the mobile devices offload the computing task to the edge server over the wireless link.Due to the uncertainty of the wireless propagation environment,the benefits of MEC cannot be enjoyed if the task cannot be effectively offloaded to the edge server.From the perspective of communication,IRS technology is combined with MEC network to further deliver the advantages of MEC by improving the communication quality of wireless offloading link.Therefore,we investigate the resource allocation problems of deep learning edge computing,IRS-assisted secure edge computing,IRS-assisted wireless powered cooperative edge computing and double-IRS-assisted MISO communication.The main research contents of this thesis are as follows:(1)Deep-learning-based resource allocation of edge computing and communication for the Io TIn MEC networks,for small-scale compute-intensive tasks,they are often highly integrated and can only be executed locally or on the edge servers.For the atomic problem,we first propose the computing offloading model.Then,the joint optimization problem of offloading decision and resource allocation for atomic tasks is modeled as a non-convex mixed integer programming problem with the target of minimizing the weighted-sum energy consumption of all users.Then,a multi-user MEC model based on unsupervised deep learning is proposed to transform a constrained mixed integer programming problem into an unconstrained deep learning problem.And a joint training network is designed.The teacher network and the student network can be trained iteratively and the student network can obtain the lossless gradient information.The problem of gradient disappearance in the back propagation process can be effectively solved.Simulation results show that the trained neural network can achieve the mapping from the channel gain to the offloading decision and resource allocation with low complexity,and the binary offloading scheme can effectively reduce user energy consumption and improve system performance(2)Resource allocation of the IRS-assisted secure edge computing and communicationIn MEC networks,mobile devices reduce task execution latency and energy consumption by offloading tasks to edge servers for execution.However,due to the adverse propagation environment,the offloading signal suffers from fading and attenuation during transmission,severely inhibiting the performance of MEC networks.In addition,due to the broadcast characteristics of signals,the offloading data in MEC networks may be eavesdropped by eavesdroppers,resulting in information leakage which seriously threat to the security of the offloading data.To address the above issues,we propose an IRS assisted secure computing offloading scheme from the perspective of improving the communication quality.Under the constraints of edge computing resources and IRS phase-shifting,the joint optimization problem of offloading ratio,computing resource allocation of edge servers,multiple user detection(MUD)matrix and IRS phase-shifting parameters is modeled as the minimization problem of multi-user weighted delay sum.Then,the original problem is decoupled into a computational design subproblem and a communication design subproblem.And,the Lagrangian dual method is used to optimize the offloading decision and edge server computing resource allocation.The weighted minimum mean square error method and the Riemann conjugate gradient method are used to solve active beamforming and passive beamforming,respectively.Finally,simulation results show that the proposed scheme can further improve the performance of MEC networks and reduce the weighted delay sum of multiple users by improving the quality and security of wireless offloading links compared with the IRS-free scheme and the IRS random phase scheme.(3)Resource allocation of IRS-assisted wireless powered cooperative edge computing and communicationThe development of applications such as autonomous driving,environmental detection,and intelligent grazing,requires the deployment of a large number of sensor devices in the Internet of Things to collect data.These sensor devices usually have limited battery capacity,and frequent battery replacement or charging consumes a lot of manpower and material resources.At the same time,the large amount of data collected by sensor devices usually requires online processing.Due to the limitations of hardware conditions,the computing,storage,and communication capabilities of sensor devices are limited.It is difficult to rely on their own resources to complete the tasks.In addition,intelligent devices are now densely deployed in wireless networks.Due to the sudden nature of wireless traffic,many devices are idle.In order to fully utilize idle computing resources around and provide continuous power to low-power devices,we first propose an IRS-assisted wireless charging cooperative computing offloading model.Users can perform local computing or offloading through wireless charging,and far users can expand their task processing capabilities through the cooperation of near users.Secondly,based on this model,the joint optimization problem of the passive beamforming of the IRS during charging and information transmission,local CPU frequency,transmit power and transmit time is modeled as a non-convex maximization problem of computational bit number.In order to solve this problem,the original optimization problem is transformed into four subproblems using alternating optimization algorithms with iteratively updating.Finally,simulation results verify the effectiveness of the proposed scheme.(4)Resource allocation and cooperative beamforming design for double-IRS-assisted MISO communication systemAs one of the key technologies of 6G,intelligent reflecting surface(IRS)can economically and effectively reconfigure the wireless propagation environment.Most existing works of IRS focuse on the optimization and performance enhancement of passive beamforming,without considering the cooperation between multiple IRSs,and does not fully exploit the advantages of multiple IRS-assisted wireless communication.Hence,we studied the design of active and passive beamforming for doubleIRS assisted multiple input single output(MISO)downlink communication for users at dead area without the direct link.Considering the cooperation of multiple reflection links and single reflection links,the active beam at the base station and the passive beam at the two IRSs are jointly optimized to maximize the user weighted rate sum under the constraint of transmitting power.To solve this problem,a double-IRS assisted closed fractional programming block coordinate descent method is proposed.First,the original problem is reconstructed into a manageable problem using a closed fractional programming method,and then the suboptimal solution is found using approximately linear block coordinate descent and continuous convex approximation techniques.Finally,simulation results demonstrate that the proposed double IRS assisted wireless communication scheme can further improve system performance and channel capacity.