Research On Migration Mechanism Of 5G Service Function Chaining

As technology evolves,a range of new vertical use cases emerges.But traditional mobile networks employ a one-size-fits-all approach to providing services,regardless of the diverging requirements of vertical services.Service function chaining(SFC)is regarded as an important technology for the 5G networks not only because it can flexibly tackle diverse usage scenarios,but also can provide users with customized services.However,due to SFC requests’ lifecycle and resource adjustment,the distribution of the remaining physical resources may become unbalanced,which brings potential negative effects to subsequent SFC requests as well as network operators,such as reducing the acceptance ratio of subsequent services,reducing utilization of physical resources,etc.And more than 80% of network traffic is generated by less than 10% of top-K flows.To improve the acceptance ratio of subsequent services and the utilization of physical resources,operators need to migrate key SFC requests in the 5G network.Aiming at the above problems,this paper intends to first study how to effectively identify the top-K flows in 5G networks,and then study the problem of SFC migration.Aiming at the identification of top-K flows in 5G networks,this paper first formulates the problem as a Combinatorial Multi-Armed Bandit(CMAB)model and then revises the CMAB model according to the characteristics of emerging network technologies(such as software defined network,network function virtualization,etc).Finally,this paper designs an effective algorithm based on greedy thoughts to dynamically identify the top-K flows in the network.Simulation experiments based on real network traffic data show that in the Abilene network,the identification performance of the algorithm designed in this paper is 65.42%,32.05%higher than two existing top-K arms identification strategies,respectively.In the Geant network,the above numbers vary to 40.48%,and 8.43%,respectively.For the SFC migration problem,this paper uses the integer linear programming model to formulate the problem under the condition of comprehensively considering the constraints of the SFC request during migration,the distribution of physical resources,and the migration cost.Then,this paper designs two different migration strategies,a conservative migration strategy and an aggressive migration strategy,to accomplish the migration task of SFC.Compared with the conservative migration strategy,the aggressive migration strategy focuses on the balance of physical resource distribution,thus causing more migration cost.Finally,simulation experiments show that after SFC migration using a conservative strategy,when two different strategies are used to allocate resources to subsequent SFC requests,the request acceptance ratio increases by an average of 1.81%and 2.54%,respectively,and the long-term profit of operators increases by an average of 3.05% and 4.62%,respectively.After using the aggressive strategy for SFC migration,the above numbers change to 11.75%,6.84%,9.49%,and 7.14%,respectively.In addition,in many real application scenarios,the change of network traffic will have obvious diurnal phenomenon,that is,the resource requirements of the services during the day will be significantly higher than that at night.To tackle this phenomenon,this paper optimizes the SFC migration strategy based on advanced time series prediction techniques.The simulation experiments show that when two different strategies are used to allocate resources to subsequent SFC requests in combination with the resource prediction mechanism,the acceptance ratio of requests for the conservative migration strategy is further increased by 16.66% and 11.02%respectively,and the long-term profit of operators is further increased by14.33% and 11.37% respectively.For the aggressive migration strategy,the above numbers change to 9.28%,8.54%,8.52%,and 9.57%respectively.