Abstract
Name-based forwarding plane is a critical but challenging component for Named Data Networking (NDN), where the hash table is an appealing candidate for data structure utilized in FIB on the benefit of its fast lookup speed. However, the hash table is flawed that it does not naturally support the longest-prefix-matching (LPM) algorithm for name-based forwarding. To support LPM in the hash table, besides the linear lookup, the random search (such as binary search) aims at increasing the lookup speed by reconstructing the FIB and optimizing the search path. We propose a composite data structure for random search based on the combination of hash table and trie; the latter is introduced preserve the logical associations among names, so as to recycle memory and prevent the so-called backtracking problem, thus enhancing the lookup efficiency. The experiment indicates the superiority of our scheme in lookup speed, the impact on memory consumption has also been evaluated.
Highlights
Named data networking (NDN; Zhang et al, 2010) is a networking paradigm concentrating on the content itself rather than its endpoints
In our Forwarding Information Base (FIB) design, besides the hash table, we introduce a component granularity trie structure to record the logical relationships among names: Figure 1 illustrates the primary data structure of our design: The hash table is utilized for fast name lookup, the search key in each table entry (n, e) is the interest name n, while the value is the corresponding trie node e
We can not obtain a realistic Named Data Networking (NDN) FIB with sufficient size and logical relationships among names, based on which we could numerically analyze the influence of false negative error
Summary
Named data networking (NDN; Zhang et al, 2010) is a networking paradigm concentrating on the content itself rather than its endpoints Instead of their destination addresses, packets are forwarded based on hierarchically structured names like URL to facilitate multicasting and in-network caching. Since the NDN FIB is keyed by hierarchically-structured content names, as in IP, the Interest packets are forwarded based on the longest prefix matching (LPM) results, with the requested names as the lookup keys. As some parts of the name cannot be informed or inferred beforehead, the consumer sends Interests carrying only a prefix of the content name to the network, and any Data under this prefix can be returned by NDN protocol. Same as IP protocol, the name matching algorithm in NDN’s Forwarding Information Base (FIB) is the Longest Prefix Matching (LPM), performed with the Interest name as the search key. NDN forwarding is challenging in scalable applications for several reasons: (1) an NDN name has unbounded variable length, making a lookup operation timeconsuming; (2) The amount of NDN FIB entries is larger than that of IP by orders of magnitude; (3) NDN FIB has to be updated more frequently as a result of content publishing and deletion, making the forwarding plane design a difficult but essential task
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have