Technology Of Multidomain Virtual Network Embedding Algorithms For Various Qos Requirements

With the rapid development of cloud computing,Internet of things,5G and other emerging technologies,the traditional Internet architecture has gradually become rigid.At present,the network service that people pursue is oriented to the quality of service(QoS).Internet architecture only provides "best effort" delivery and cannot serve users well.In order to solve the above problems,it is a reliable way to design an efficient virtual network embedding(VNE)scheme under the background of network virtualization(NV)technology.Cross domain network embedding is the focus of VNE algorithm.Based on the virtual network technology framework of software defined network(SDN),in the face of differentiated QoS requirements,this thesis proposes a new cross domain VNE algorithm based on demand awareness.At present,the differentiated QoS requirements in cross domain VNE mainly include cost,delay and bandwidth.The cross domain VNE algorithm for differentiated QoS is divided into cost aware cross domain VNE algorithm,delay aware cross domain VNE algorithm and bandwidth aware cross domain VNE algorithm.The above algorithm uses the cooperative working model of global controller and local controller to research the cross domain embedding algorithm innovatively.A cost-aware cross-domain virtual network embedding algorithm(CA-VNE)is proposed to maximize operator revenue and reduce VNE cost.Aiming at minimizing the embedding cost,this algorithm firstly uses the virtual network request shard algorithm to shard the virtual network requests received by the global controller.Then,the network topology is abstracted and the candidate physical nodes are sorted by the proposed estimated average embedded cost formula.In the in-domain embedding stage,the candidate physical nodes are determined by particle swarm optimization(PSO)and the in-domain embedding is completed under a local controller.Finally,the cross-domain resource allocation is carried out.In this stage,the inter-domain link mapping algorithm completes the link mapping between each physical domain,and finally realizes the cost sensing cross-domain VNE process.Simulation results show that the proposed algorithm can greatly reduce the embedding cost and has good stability and scalability.In this thesis,a DA-VNE method for low latency multi-domain VNE is studied.In this algorithm,the global controller is used to partition the virtual network requests.Then,in the candidate node selection algorithm,a delay unit DelayUnit is defined to measure the delay attribute of the candidate physical node,and finally the physical node with the smallest delay unit DelayUnit is selected as the candidate physical node.Then,with the support of the information uploaded by the local controller,the global controller uses the PSO algorithm to pre-map the virtual network.Finally,Floyd algorithm is used to map inter-domain links under the control of the global controller.The simulation results show that with the increase of the number of virtual network requests,the method is more stable in the total delay,node delay and link delay of virtual network requests,and is better than the multi-objective optimization algorithm and the delay estimation algorithm.In this thesis,a bandwidth-aware multi-domain virtual network embedding(BWA-VNE)is proposed to solve the problem of high-bandwidth multi-domain VNE.Firstly,the local controller selects candidate nodes according to the predicted embedding cost formula and the link with large bandwidth value.Then the global controller uses the virtual network pre-embedding algorithm to pre-map the node according to the information uploaded by the local controller.Then the local controller completes the in-domain embedding according to the link bandwidth value.Finally,the interdomain link mapping algorithm is implemented by the global controller.The simulation results show that the average bandwidth of the selected link is higher than the average bandwidth of the link in each domain.