Resource Allocation Of Service Function Chain In Multi-access Edge Computing Integrated Networks

5G technology has brought a leap in the communication capabilities of the mobile Internet of things(Io T).Still,it will continue to be difficult to cope with the rapidly increasing demand for network services with the demand for intelligent connections.In the future,the 6G communication system will break through the limitations of the surface terrain,and expand communication to natural spaces such as space,air,land,and sea.It can improve the overall resource efficiency through the coordinated transmission of multiple access methods and the unified management of multiple system resources.A space-air-ground integrated network can achieve seamless global coverage in a true sense.By introducing multi-access edge computing(MEC)into the space-air-ground integrated network,cloud infrastructure and service capabilities are relocated from the central cloud to the edge of Internet.It can realize local distribution and processing,thereby avoiding the high latency and bandwidth bottlenecks of non-terrestrial networks.The integrated and converged network involves multiple communication systems,with many and complex nodes.Even more,some access nodes have high mobility.It is necessary to design a lowlatency,high-efficiency network structure,and a flexible service deployment plan.Resource allocation is one of the key issues in service deployment.This dissertation studies the resource allocation problem of the service function chain(SFC)under the MEC-enabled space-air-ground integrated network.It is significant to design an efficient,green,and reliable SFC resource allocation plan for a heterogeneous integrated network with diversified quality of service(Qo S)requirements(delay requirements,security requirements,etc.)and resource constraints(computing resources,memory,bandwidth,etc.).The aim is to reduce the service deployment cost or energy consumption of service providers.The main contributions are as follows.First an adaptive SFC resource allocation scheme for cloud data centers is proposed.The issue of resource allocation is one of the critical issues of cloud data center virtualization.It is also the research foundation of space-air-ground integrated network resource allocation.The immense scale and complex node types of cloud data centers exacerbate the complexity of the problem.Energy-related costs account for an increasing proportion of total data center expenditures while infrastructure upgrades will worsen energy efficiency issues.This research proposes an adaptive SFC resource allocation algorithm considering the diverse nodes and the balance between resource cost and energy efficiency cost.This algorithm can reduce energy consumption in light-load network scenarios and save resource costs in heavy-load network scenarios.We propose a resource allocation algorithm for virtual network functions(VNFs)based on association rules.The shortest distance is used as the association rule to ensure that the physical server hosting the VNF is as close as possible by keeping the distance between the switches shortest.The centralized deployment of VNFs reduces the number of active physical servers,provides a good foundation for the resource allocation of switch VNFs and link requests.In this way,the costs are reduced.The joint resource allocation method of the switch VNF and link request is expressed as an integer linear programming(ILP)model,which can automatically adjust the resource cost and energy efficiency optimization weights in the objective function according to the current resource status of the cloud data center.We also investigate the resource utilization and time efficiency of the two breadth first search(BFS)-based link resource allocation algorithms.These two algorithms have the different starting node and direction when searching for available physical resources.Second a SFC resource allocation scheme for the terrestrial-MEC network is proposed.Cloud data center is the primary computing power support of the traditional terrestrial network.As the supplement to cloud computing,MEC can push computing to the network's edge and provide lower latency services.However,MEC's physical resources are very limited,which is difficult to process some service requests that are both computationally sensitive and time-delay sensitive.The frequent transmission of data between the memory and the CPU during operation also causes idleness and waste of resources.This research aims at the balance between computing power and timeliness in the terrestrial-MEC network.We propose an resource allocation mechanism to provide better computing support for improved MEC server which applied general-purpose computing on graphics processing units(GPGPU)and processing in Memory(PIM)technology.We construct a mathematical model of the network for direct communication between terminal devices and MEC,characterizing the association and flow between user service requests and SFC resource allocation in MEC to reduce SFC resource allocation's complexity in various MEC converged networks.We construct a mathematical model of a cloud-MEC network(terrestrial-MEC network),and concretized relay management,communication coordination,and resource scheduling between different physical environments into mathematical expressions.We formulate the SFC resource allocation problem as an ILP model,design algorithm based on column generation,and discuss the advantages and disadvantages of optimization theories versus heuristic algorithms.We compare the acceptance ratio and cost between terrestrial-MEC network and non-converged network,improved MEC server and ordinary MEC server from the transient and steady-state level.Last a SFC resource allocation scheme for the MEC-enabled integrated network with three-dimensional communication is proposed.Non-terrestrial networks,which have rich physical resources and large delay,can provide remote areas,airspace,and oceans with communication,although the latency is higher.It is difficult to respond quickly to delay-sensitive services.Therefore,ground-based networks and MEC are required to assist non-terrestrial networks.Remote areas have unique geographical,population,and resource conditions,so the fault tolerance of three-dimensional communication services is lower,putting forward higher requirements for service security.This research studies the delay and security issues of the three-dimensional communication MEC network.We construct the high altitude platform station(HAPS)-cloud-MEC network mathematical model,which characterizes the relationship between the ground-based network and the air-based network communication characteristics and resource attributes.A SFC resource allocation ILP model that balances resource requests and delay constraints is proposed.The model's objective function can measure the resource and communication status of HAPS,cloud,and MEC and ensure that SFC meets the delay constraint,while optimizing resource costs.A two-stage optimization model of SFC resource allocation in the shipboard MEC network is proposed.In response to the uncertainty of SFC's security requirements,the sliding time window sampling method is used to improve the accuracy of the empirical probability distribution.A pre-deployment algorithm for network defense measures based on distributionally robust optimization(DRO)is proposed.This algorithm can control the accuracy of the decision by using a different confidence level of the uncertainty set.In this way,the cost and security balance can be controlled by service providers.