Edge Caching Research In POF-ICN Architecture

Content transmission has gradually become the main mode of network traffic.With the commercial application of 5G and development of ICT technologies,the scale of content in the network is growing larger,and Internet users demand a higher speed of content delivery.The traditional Internet architecture,which core is IP address,has exposed many deficiencies in content distribution.The new network architecture has become a research hotspot in academia.The information-centric network(ICN)takes information/content as the core of the network,and has natural advantages in terms of mobility support and content distribution,but also facing problems that being incompatible with current network equipment and difficult to deploy.The software-defined network(SDN)dedicates to decouple the control plane and forwarding plane of the network,thereby achieving agile and flexible network management and control,and providing technical support for the implementation of ICN.The combination of SDN and ICN has gradually become an important research direction of the future network.POF-ICN(Protocol Oblivious Forwarding)is one of the representative architecture.Based on the POF-ICN network architecture,this thesis builds a collaborative cache to achieve efficient,rapid and stable large-scale content distribution in the edge network.First,in order to adapt the routing and forwarding of ICN,which is based on content name,to the separation mechanism of SDN control and forwarding,this thesis draws on the idea of content addressable network(CAN)for content addressing and portioning,and maps edge POF content switches into self-organized collaborative caching and forwarding network over the edge network.Through the centralized orchestration by the controller,cache access tables and forwarding flow tables are constructed on the POF content switch nodes,realizing routing and forwarding function based on content name in forwarding plane.Compared to the native ICN caching strategy,edge caching can improve the cache hit rate while ensuring the stability of content distribution.Aiming at the limitations of edge caching,this thesis further proposes cache filtering mechanisms,designs a cache quick lookup flow table based on Bloom filter,and studies the cache decision filtering mechanisms based on content popularity.First we improve the cache decision filtering based on LRU and hash table by introducing the request interval,then we propose a cache decision filtering mechanism that maintains an approximate frequency of recently requested content items by Count-Min sketch and reset method.The experimental results show that the proposed caching decision filtering mechanism can improve the deficiencies of uniform caching in edge caching,further enhance the performance of the cache system and reduce unnecessary cache evictions,thus achieving efficient and stable edge network content distribution.