Towards Efficient Content Discovery in NDN with Enhanced Component-Based Bloom Filters

Abstract

Variable-length hierarchical names of Named data networking (NDN) content bring about intuitive issues for both content routing and large-scale forwarding. Bloom filters can accelerate hash-based NDN content discovery with a false positive probability. In this paper, we propose ECBF, a component-based bloom filter scheme to enhance the reliability of bloom filters. We exploit the hierarchical characteristics of NDN content names and store them into different bloom filters in a component-based way, and then we present a correlation validation policy to check whether the names exist. Finally, we introduce an ECBF-based content discovery scheme BCD. ECBF and the correlation validation value are stored in a fast on-chip memory, and cryptographic check values of content names and related information are respectively stored in an off-chip memory of intermediate routers. This design not only improves the effectiveness of content discovery, but also reduces the additional memory throughput. We experimentally demonstrate that BCD can provide a more than twice reduction of false probability compared with the primary bloom filter methods. Besides, BCD can economize about 40% memory consumption.