Monday, February 8, 2010

Smart Grid Security Performance Standards

I have just gone through the somewhat laborious process of reviewing

February 2010 DRAFT NISTIR 7628

Smart Grid Cyber Security Strategy and Requirements

which I found through Smart Grid News (

Section 3.1 touches on crypto, and the general tone of the document would suggest that logical (software) security is the method of choice due to the performance hit a system must take when implementing more secure levels of crypto (i.e. secure microcontrollers). Since performance is so important, one would logically conclude that hardware based security introduces challenges. Correct me if I am wrong, but that is how I interpret this.

That being said, what seems to be lacking is any sort of reference for performance. Since hardware based security is, in many ways, superior to logic based security (not always, but the BEST hardware based security chips are far more secure than the best logic based counterparts), then we really need a frame of reference here. Granted, no hardware based security solution will ever be able to match the performance of a logic based system (in fact, you can get maximum performance by simply using buzz words to describe a part of the system - like 256 bit encryption), but the best secure IC's deliver some pretty good performance while offering some very solid security. After all, banks rely on hardware based security (i.e smart card based security) for their most critical systems, and system availability and reliability are directly tied to the very high performance requirements vendors must adhere to in order to sell to the banking industry.

The one issue that does come into focus, of course, is budgetary constraints. Vendors of AMI systems must compete to sell their products, and the increased cost of implementing secure microcontrollers that deliver the requisite level of performance cuts into everyone's bottom line (which ultimately is the TRUE deciding factor). Logic based security can be implemented for anywhere from $0 to fractions of a cent, while high performance and high security hardware costs more.

Nonetheless, if cyber security is such a major concern (as it should be) in the implementation of the Smart Grid, then we should perhaps seek to create some target objectives for vendors of hardware based security, including performance and cost. We should also view the total cost in a systemic manner, taking into account the risk of relying on logic based security, and the cost of failure.

After all, if we are building an infrastructure that is expected to remain a part of our critical energy infrastructure for MANY years to come before replacing it with the next best thing, we should probably create solid, tactical objectives as well as higher level objectives.

It really takes both to succeed.

No comments: