Paper Review: Reliable Group Rekeying: A Performance Analysis
Reviewer: Kenneth Chin
This is a neat paper in the sense that the authors introduced a group key management system which has a satisfiable reliability, scalablility and security. Keygem basically consists of three major components: registration, rekey encoding and rekey transport. It is obvious that the authors have a very profound knowledge in areas like tree, forward error correction and so forth. The authors successfully incorporated different decent elements such as proactive FEC, batch processing concept and keytree into the group key management system to make it scalable and reliable.
I am a little bit skeptical about the scalability of Keygem; Keygem is not encouraged to has a huge group of users as this has an impact on bandwidth. Therefore, I think it is only locally scalable but not universally scalable. Nevertheless, Keygem should be working very well with a small group of users. For large number of users, the solution would be using multiple distributed Keygems. However, there would be a lot of challenges to design an efficient distributed group management system. The second thing I would like to comment on is that batch rekeying is based on the assumption of delayed service. It inherently imposed a limit on the kinds of applications that Keygem can support. Finally, using a single group key is very dangerous because it creates a single point of failure. Rekeying also generates a lot of keys and messages. Thus, in that sense, would public-private key pair be a better solution?
One of the most important thing i learned from this paper is the R-BFA algorithm. It is essentially a balance between DFA and BFA.
This paper is an interesting paper and the analysis results are very useful. It is a 3rd-grade paper.