A New Bloom Filter Architecture for FIB Lookup in Named Data Networking

Round  1

Reviewer 1 Report

Authors propose a novel bloom filter based approach for FIB implementation in NDN routing. The strategy described in the relies on the translation of a prefix tree into a pair of functional bloom filters, the first providing the level in the suffix tree, and second indicating which interface should be used for forwarding. The idea is originals most other approaches rely on a single filter to provide this functionality. Authors provide a detailed analysis of their algorithm, which results are confirmed by simulation. 

Overall the paper is good quality work with detailed introduction of the various algorithms involved and a progressive description of the technique.

However the writing is a bit unclear sometimes, mainly because of language. I suggest careful read-proofing with an native English speaker. 

output face : output interface 

- P.5 L 174 : "When no more nodes to follow" : when there are no more nodes to follow

Also  Related Works/ Evaluation: A rapid search yielded several papers that provide a comparable approach, and should be included in related works, such as:  the evaluation is only done with a comparison to classic prefix-tre based appproach. Comparing other approaches in related works should be considered. For example:

- Cristina Muñoz, Liang Wang, Eduardo Solana, Jon Crowcroft: I(FIB)F: Iterated bloom filters for routing in named data networks. NetSys 2017: 1-8

Author Response

Author Response File: Author Response.pdf

Reviewer 2 Report

This research presents using Bloom filter for rapid search of NDN applications.   Though the similar idea has been used off-chip searching, the author's proposal for on-chip search is novel. Especially it creates a low and high priority tries for achieving the bloom filter. The results are also interesting. 

However, looks like the presentation of results is somehow confusing. Tables 5 shows the performance, but the total performance of the proposed algorithm is larger than the traditional methods. 

Also, Figure 4 does not have the proposed method results.

There is an apparent gap between theoretical and simulation (FIgure 5 and 6), please explain more why such gap exists.

Some grammar errors still exist such as: The INDET result occurs when every accessed cell of the l-BF has a conflict value.  Please fix.

Author Response

Author Response File: Author Response.pdf