Multi-Domain Orchestration Scheme For Satellite-Terrestrial Integrated Networks

With the rapid development of the Internet,satellite networks with various advantages such as wide coverage,full-time span,and high robustness have become the focus,and the Satellite-Terrestrial Integrated Network(STIN)will also play an important role in the future Internet.However,the STIN also face some challenges in network management,such as dynamic topology,non-negligible transmission delay,and multi-dimensional constraints.In addition,the differentiated service requirements bring difficulties to service deployment and scheduling.In this paper,network virtualization and service function chain(SFC)are introduced into the STIN,the model of STIN is designed in detail,and two service orchestration schemes are proposed.Finally,we simulate the satellite network to complete the experiment for verification.The specific research content can be divided into the following three parts:Ⅰ.Various factors should be considered in the model design of the STIN.In this paper,a research scenario is established based on MEO/LEO satellite network,and multi-dimensional resource constraints in the STIN are included in the network model.The dynamic correlation of satellite topology is carried out based on time aggregation graph,and the service requests are classified according to differentiated requirements.It provides the foundation for the implementation of the subsequent scheme.Ⅱ.The diverse business requirements in STIN increase the complexity of service orchestration.Considering the limitation of network resources and different types of services,this paper designs a resource-adapted virtual network function(VNF)resource scheduling scheme.A VNF deployment algorithm based on the weighted bipartite graph is designed,and balances the load and sharing degree of functional nodes to avoid frequent adjustment and resource insufficiency.In order to solve the problem of frequent reconfiguration caused by topology changes,a SFC orchestration strategy for differentiated services and a dynamic SFC migration algorithm are proposed.Finally,the simulation verifies that the proposed algorithm can well adapt to the STIN scenario,reduce the remapping cost,and improve the resource utilization rate and service acceptance rate by considering the multi-service optimization objective.Ⅲ.Although the centralized orchestration scheme can uniformly schedule and globally coordinate resources,it will incur additional overhead in topology collection and decision delivery,and not be fast enough to respond to sudden failures.Therefore,this paper designs the domain management model of STIN,and takes the intra-domain management cost and the request distribution as the domain parameters.The alliance game theory is introduced to realize the distributed domain partitioning algorithm,and the multi-domain service orchestration algorithm is designed based on the two-step embedding algorithm.Finally,we verified that the domain partitioning algorithm could converge,and the proposed multi-domain service orchestration algorithm could realize the rapid response of the domain request as well as flexible network resource management and inter-domain coordination.