Computation Offloading And Multi-dimensional Resource Optimization In Digital Twin-assisted Industrial Internet Of Things

With the rapid development of the Internet of Things,the demand for efficient production and intelligent management in traditional industries has surged,and therefore the field of Industrial Internet of Things is booming.However,the shortage of resources in the Industrial Internet of Things has restricted its transformation and upgrading to intellectualization and digitalization.Multi-access Edge Computing,as a new computing paradigm deployed at the edge of the networks,provides a promising solution to overcome the problem that the limited computing resources of devices can not satisfy the computing requirements of industrial applications.During the process of offloading the computation tasks to the edge,non-orthogonal multiple access technology opens up a new way to solve the problem of limited transmission resources in the Industrial Internet of Things by improving the channel capacity.At the same time,in order to deal with the contradiction between the large-scale access of edge devices and limited computing resources,digital twin technology improves the computing performance of edge computing servers through the condition monitoring and real-time prediction.Therefore,this thesis focuses on the digital twin-assisted edge computing architecture under the Industrial Internet of Things scenario.By empowering computation offloading with non-orthogonal multiple access technology,this thesis proposes a multi-dimensional resource optimization framework based on the partial and full offloading strategies,respectively.The main research results are summarized as follows:(1)A digital twin-assisted edge computing architecture based on the Industrial Internet of Things scenario is proposed.In order to achieve high-fidelity twinning services,a real-time twinning pipeline model of edge server is designed.The digital twin server,as the virtual representation of the edge server,is able to assist the edge computing when the edge server is busy,which effectively improves the reliability of the edge computing system in the complex industrial environments.(2)A multi-dimensional resource optimization framework based on the partial offloading strategy is proposed.Aiming at the delay-sensitive issue of computing tasks in the Industrial Internet of Things,the total task completion delay of all industrial devices is minimized by jointly optimizing the subchannel allocation,the computing capacity allocation,the edge association,and the transmit power allocation.After proving the non-convexity of the mixed integer programming problem,the non-convex problem is then transformed into four solvable sub-problems based on the coordinate descent algorithm.In order to reduce the complexity of the algorithm,a global alternating optimization algorithm is further proposed to solve the sub-problems separately until convergence.Simulation results show that compared with the traditional offloading scheme,the proposed scheme reduces the total task completion delay by7.6% ～ 18.8%,and also increases the offloading percentage of tasks by 10.3% ～ 21.4%.(3)A multi-dimensional resource optimization framework based on the full offloading strategy is proposed.Aiming at the challenge of large-scale task transmission due to insufficient system throughput,the average throughput of the system is improved by jointly optimizing communication and computing resources.Based on the constraint of total task completion delay,the original problem is decomposed and solved by the Lagrangian dual method to obtain the optimal computing capacity allocation.The communication resources are optimized step by step through the alternating iterative method under the limitation of the total transmission power of industrial devices and the transmission speed of a single industrial device.On this basis,a clustering algorithm for the industrial device clustering is proposed to improve the transmission performance of the system.Compared with the traditional scheme,the average system throughput of the proposed scheme is increased by 28.2%,and the transmission energy consumption is reduced by 9.1%～23.2%.