| With the continuous expansion of network scale,providing caching service based on content delivery network(CDN)is an important way to improve quality of service,reduce the flow into the core network and reduce the impact of burst traffic on the source server.However,due to the limitations of network location and hardware resources,the traditional CDN has exposed some serious disadvantages,such as high deployment cost,low resource utilization,poor reliability and so on.In view of the above disadvantages,Smart Integration Identifier Network carries out multi-dimensional coordination of network resources through "control forwarding separation" and "identity location decoupling",which provides effective support for the realization of flexible deployment,resource efficient,safe and reliable CDN services.Therefore,this paper designs a caching mechanism based on virtual content delivery network(v CDN)in Smart Integration Identifier Network,in order to improve the effectiveness and reliability of v CDN based caching service in Smart Integration Identifier Network.The contribution of this paper can be summarized into the following three aspects:Firstly,the v CDN service is implemented based on the container and the cache mechanism for Smart Integration Identifier Network is designed.The caching mechanism based on v CDN in Smart Integration Identifier Network includes virtualization module,resource status monitoring module,global load balance module,local load balance module,cache decision module,secure transmission module and cache control module.All modules work together to realize flexible deployment,cache control and load balance of v CDN service.Secondly,a proactive cache mechanism based on federated deep learning is constructed,combined with cache replacement strategy to achieve better performance.What’s more,load balancing strategy is used to further optimize cache services.In this paper,a federated convolutional neural network algorithm for predicting content popularity is proposed to solve the problem of training data dispersion in v CDN system,and the prediction results is used to guide cache control.The combination of proactive cache control and LRU cache replacement strategy improves the cache hit rate and network resource utilization.On this basis,combined with the load balancing mechanism based on polling resolution and node migration,the cache results are optimized to further improve the utilization of network resources and enhance the reliability of v CDN system in the face of burst traffic.Finally,this paper builds a prototype system based on Smart Integration Identifier Network architecture to perform functional testing and performance testing of the cache mechanism.The test results show that compared with the traditional CDN caching mechanism,the caching mechanism based on v CDN proposed in this paper can reduce the resource occupation by up to 40% on the premise of maintaining the cache hit rate of about 78.7%,which effectively improve the utilization of network resources.By combining the load balancing mechanism,the average load rate of v CDN nodes is reduced by about 58.7% compared with that without load balancing,which effectively enhances the reliability of cache service based on v CDN. |
