Research On Routing Scalability In Named Data Networking

The universal network layer(IP)is the center of the hourglass architecture of the current Internet,and the minimum necessary functions of global interconnection are realized through the network layer.However,in recent years,with the popularization of 4G and the large-scale use of 5G,the information is more concerned by people than the location of the information,and IP networks are inefficient and difficult to optimize in large-scale distribution of content.In order to solve the problems existing in the traditional IP network,it is urgent to explore a new network architecture.Named data networking(NDN)is considered to be a new type of network architecture that can better meet the needs of users for information transmission.However,NDN routes and forwards packets based on names.Compared with fixed-length IP addresses,names using hierarchical naming will increase the number of the forwarding information base(FIB)entries exponentially in large-scale networks.As a result,the load of the node is increased,and a hidden danger of collapse is existed.At the same time,problems such as low FIB lookup efficiency and large memory usage are brought by the large number of FIB entries,which makes it difficult for NDN to be applied to large-scale networks,and scalability problems are existed.Therefore,the scalability of named data networking from the routing level is studied in this thesis.The main work and innovations of this thesis are as follows:(1)Aiming at the routing scalability problems of NDN in large-scale networks,such as the huge number of FIB entries,the excessive load and overhead of nodes,etc.,from the perspective of reducing the number of FIB entries and reducing the load of nodes in the network,a solution for routing scalability based on name classification is proposed.Two different forms of content announcement packages are released actively by producer to store content names of different popularity in a decentralized manner,so that interest packages are processed in a decentralized manner according to the popularity of the content names requested by them.At the same time,the Non-global name concentration point is introduced to share the storage pressure of global name concentration point.Finally,compared with the tracking-based scheme,the processing delay of interest packets is reduced in this scheme,and the number of interest packets processed by the router are increased by about 8%-13%,thereby the routing scalability of the network is improved.(2)Aiming at the exponential explosive growth of name entries in FIB,the large memory usage,and the slow name lookup speed caused by the expansion of the network scale of NDN,from the perspective of speeding up FIB name lookup and reducing FIB memory usage,a name lookup scheme based on popularity and CDT is proposed.FIB is divided into counting bloom filter(CBF),popular FIB,conflict-split degraded trie(CDT),and auxiliary FIB in this scheme.CBF is used to quickly filter out name prefixes that are not in FIB,popular FIB is used for fast forwarding of highly popular name prefixes,the number of nodes and the depth of the tree are reduced by CDT through the conflict split mechanism,and auxiliary FIB is used to assist the update and replacement of name prefixes in popular FIB and the rapid positioning of nodes in CDT.Finally,compared with the BF-PDT(Bloom Filter + Popularity Table + Degraded Trie)scheme,optimized effects of this scheme are presented on creation time,lookup time,and memory usage,which proves that the storage and name lookup performance of FIB is improved in NDN.As a result,the routing scalability of the network is further improved.