Research On Converged Networking And Service Mechnism Of Access Networks

Access network is widely deployed with various technologies,and fiber access and wireless access are two typical access technologies.In order to meet the increasing service requirements and to provide high-capacity and anytime-anywhere access capabilities,fiber and wireless access networks(FiWi access networks)will provide high-bandwidth and low-interference of fiber access converged networking with the high-flexibility and seamless coverage of wireless access,aiming at achieving complementary advantages.With the emergence of the Internet of Things(IoT),Industrial Internet and Tactile Internet,new technical challenges have been brought to the access networks.New network performance requirements in terms of bandwidth,delay and energy consumption are raised.Especially,the emerging services such as VR/AR,industrial control,and intelligent driving have more stringent demands on low latency.In order to meet these demands,edge computing brings content and computing capability to the edge of the access network with the proximity to users.It provides services closing to the users or data source at the edge of networks.Edge computing provides new ideas about converged networking of access networks at the same time.This thesis is mainly supported by the National Natural Science Foundation of China.We focus on the converged networking and service mechanism of the heterogeneous FiWi access networks combined with edge computing by using network virtualization.The main work and contributions are as follows:1)We propose an integrated networking scheme of FiWi access networks combined with edge computing.And we design and implement a FiWi access network device(called FiWi node)embedded with computing resources.And the test results verify the feasibility of the proposed integrated networking scheme.Firstly,the converged networking scheme for FiWi access networks and edge computing includes unified resource scheduling and allocation,global view of network and connection control,and service orchestrating and management functions,which makes full use of SDN and NFV to elimate the difference of heterogeneous physic networks and computing devices.Secondly,we apply edge computing capability to achieve data forwarding,enhancing the connectivity of the access network,which reduces the delay,and improves the resource utilization rate.Lastly,we has designed and developed a FiWi node embedded with computing resources,and build a testbed.Extensive experiments are conducted,and the test results verify the feasibility of the proposed integrated networking scheme.2)We propose an edge computing server placement scheme in the FiWi access network that minimizes the cost,which jointly optimizes the edge computing server placement and task offloading according to different service requirements.Firstly,we propose an edge computing server placement to minimize the weighted sum of end-to-end delay and energy consumption as the cost,where the delay includes access,transmission,and processing delay,and the energy consumption includes transmission and computational energy consumption.Secondly,in order to meet the various demands of resources for different types of services,we propose a placement algorithm based on user preferences.By deploying the edge computing server in the site location with the highest information entropy weight,the network delay and energy consumption are reduced.Thirdly,the placement algorithm based on user preference is compared with the random placement algorithm and the enumeration based placement algorithm.The simulation results show our proposed algorithm in terms of delay and energy consumption outperforms the other two algorithms.The deployment number and locations of the edge computing servers are obtained based on the minimum delay,energy consumption,and their weighted sum,respectively.3)We propose a load balancing service mechanism of service function chain(SFC)in the converged access network,and present two VNF multipath deployment algorithms.Firstly,the collaborative edge computing and the SDN coordinated with NFV of the SFC service maechanism to support the synergic SFC orchestration framework.The framework integrates distributed edge computing resources into a unified edge cloud,and expands the network service orchestration originally by the SFC into a unified orchestration of edge computing applications oriented cloud services and network services for edge computing services.Secondly,for the differentiation of service flow characteristics,we focus on the load balancing VNF deployment problem,leveraging the virtual network functions(VNFs)with multiple instances deployed on edge computing nodes(ECNs)and multipath capacity in the FiWi access networks.Thirdly,the VNF deployment problem is modelling as a mixed integer nonlinear programming(MINLP)problem.And we propose two optimization deployment algorithms,greedy-bisection multipath deployment algorithm(GBMP)and k-shortest multipath deployment algorithm(KSMP).Extensive simulations are conducted in two service scenarios.The numerical results show that our proposed algorithm outperforms single-path and ECMP-based deployment algorithms in terms of load balancing indicator,service acceptantance ratio and network utilization.