Scalable Certificate Revocation Schemes for Smart Grid AMI Networks Using Bloom Filters

Rabieh K., Mahmoud M. M. E. A. , Akkaya K., TONYALI S.

IEEE TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING, vol.14, no.4, pp.420-432, 2017 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 14 Issue: 4
  • Publication Date: 2017
  • Doi Number: 10.1109/tdsc.2015.2467385
  • Page Numbers: pp.420-432
  • Keywords: Certificate revocation, public key cryptography, smart grid security, Public key infrastructure, AMI


Given the scalability of the advanced metering infrastructure (AMI) networks, maintenance and access of certificate revocation lists (CRLs) pose new challenges. It is inefficient to create one large CRL for all the smart meters (SMs) or create a customized CRL for each SM since too many CRLs will be required. In order to tackle the scalability of the AMI network, we divide the network into clusters of SMs, but there is a tradeoff between the overhead at the certificate authority (CA) and the overhead at the clusters. We use Bloom filters to reduce the size of the CRLs in order to alleviate this tradeoff by increasing the clusters' size with acceptable overhead. However, since Bloom filters suffer from false positives, there is a need to handle this problem so that SMs will not discard important messages due to falsely identifying the certificate of a sender as invalid. To this end, we propose two certificate revocation schemes that can identify and nullify the false positives. While the first scheme requires contacting the gateway to resolve them, the second scheme requires the CA additionally distribute the list of certificates that trigger false positives. Using mathematical models, we have demonstrated that the probability of contacting the gateway in the first scheme and the overhead of the second scheme can be very low by properly designing the Bloom filters. In order to assess the scalability and validate the mathematical formulas, we have implemented the proposed schemes using Visual C. The results indicate that our schemes are much more scalable than the conventional CRL and the mathematical and simulation results are almost identical. Moreover, we simulated the distribution of the CRLs in a wireless mesh-based AMI network using ns-3 network simulator and assessed its distribution overhead.