An Efficient Name Look-up Architecture Based on Binary Search in NDN Networking

With the rapid increase in network traffic, IP architecture, which is based on the host-client mode, is becoming inefficient in coping with massive data transmission. Due to the in-network caching characteristic, named data networking (NDN), touted as the next-generation Internet architecture, has been proposed and broadly explored to improve the data forwarding efficiency of various methods, such as multicasting and collaborative caching. The key challenge is how to improve the query speed from the large-scale forwarding information base (FIB) tables. Binary search with hashing can reduce the query time fruitfully. However, a virtual prefix is required to ensure the look-up path, which will cause performance degradation and space occupation. In this paper, we proposed a high-speed query method based on two vectored-Bloom filters (VBFs) and a traditional Bloom filter to optimize the performance. In particular, a Level-VBF stores the longest level of each first component in the FIB tables. A Face-VBF returns the output face of the input prefix. Moreover, the empty node-BF programs the virtual prefix to indicate the correct binary search path. The performance evaluation results demonstrate that the LBS-BF achieves a more than 17% loop-up speed increase compared with that of the hashing binary search algorithm while occupying much less storage space.