Research Of PIT And FIB On Forwarding Plane In Named Data Networking

With the rapid development of society and technologies,the contents including HD videos and big data have become the core for Internet services.The users except to share information,no longer only need end-to-end conmmunication.However,the lack of IP addresses,lack of mobile support and NAT transformation exposed by IP network make it difficult to meet the increasing demands.To deal with these problems,a future network architecture named Named Data Networking(NDN)is proposed,which is content-oriented and supports sharing.NDN has excellent foresight,but it is also faced with the challenges need to be solved,especially the fast lookup and compression for hierarchical names in the forwarding plane.On the base of MBF that is a derived structure of Bloom filter,the thesis proposed an improved index structure called Bitmap-Mapping Bloom filter(B-MBF)to solve the problems.Owing to deploy Bitmap properly,B-MBF can achieve dynamic allocation for offset addresses of elements,thus avoiding the waste of the storage in the off-chip memory.Meanwhile,B-MBF optimizes the lookup process by only utilizing a hash function to reach multiple hash mappings,which improves the lookup speed.Moreover,for the forwarding plane of NDN,a PIT structure named B-MaPIT and a FIB structure named B-MaFIB are proposed based on B-MBF,respectively.And the processing algorithms about Interest and Data are introduced in detail.In B-MaPIT and B-MaFIB,B-MBF and Packet Store are properly deployed in the on-chip memory and off-chip memory,to reduce the memory consumption and promote the lookup speed.In order to evaluate the performance of B-MaPIT and B-MaFIB,a series of experiments are performed in a computer.The experimental results indicate,compared to the existing researches,B-MaPIT and B-MaFIB have more excellent performance in the memory consumption,building speed and throughput,which is important for the forwarding performance of forwarding plane in NDN.