Last month it was pointed out that the storm long brewing over the Federal Information Processing Standard (FIPS) security requirement in the DCI specification was coming to a head. It has been known for a few years that National Institute of Standards and Technology, known as NIST, planned to obsolete the FIPS 140-2 specification, for which compliance is required by the DCI specification. The new specification, predictably called FIPS 140-3, is to go into effect 1 Jan, 2011.
No one has been noticeably panicked about this, although there are plenty of reasons to do so. Consider Moore’s Law, which predicts that available computation power will double every 18 months. Accordingly, security standards must also advance, since an increase in computational power means it takes less time to crack the safe.
However, this is a problem for digital cinema, both long term and short term. NIST is doing the right thing and evolving its standards to stay ahead of Moore’s Law. But the industry is not going to retire all of those shiny new digital cinema systems because NIST says it’s now time to up the security ante. In the long term, Moore’s Law will win, older security systems will be phased out, and newer security systems phased in. (Shhh…if exhibitors were smart, they’d start planning for this now, and figure out how they’re going to get the studios to pay for it.) In the short term, DCI needs to decouple from NIST and preserve investment by keeping the status quo for digital cinema. But this is easier said than done.
First, it’s not possible to continue to specify FIPS 140-2. Being the clever agency it is, NIST revised FIPS 140-2 in January to bring it closer to 140-3. Compliance with either the revised 140-2 or 140-3 would create havoc in digital cinema. But NIST also removes the links to older documents from its web site, complicating other work, such as that of DCI and SMPTE, when attempting to reference older NIST documents. One of the few sites with such links can be found at http://www.mkpe.com/isdcf#nist.
Under the hood, there are several reasons why the change by NIST will create havoc. More than one set of procedures in digital cinema will be impacted. The obvious problems have to do with a new rule requiring ‘one key, one use.’ The DCI specification requires multiple use of the media blocks public key: it must encrypt the AES symmetrical key in the KDM, it must be used to verify the digital signature of security logs, and it must be used to conduct TLS sessions. This multi use of the one key is, at least in regards to digital signatures, forbidden in the revised NIST standards. To do this properly under new rules requires the use of more than one certificate in the media block. The additional keys would have to be tracked in addition to the public key used for KDM encryption. (And to think that one key-pair already causes the industry more trouble than it can deal with.)
In addition, NIST plans to discontinue use of weaker security algorithms, and require use of stronger ones. The potential impact of this has not yet been quantified.
Digital cinema was not designed to support the ‘one key, one use’ rule, and if a change in algorithms is due, it would further impact interoperability. Clearly, implementation of the updated NIST standards in digital cinema is simply not feasible.
Unfortunately, the path forward is not clear cut. NIST complicates the ability of other organizations such as DCI and SMPTE to reference its older work. FIPS 140 testing agencies will no longer honor the older documents in their tests. It is possible for newer equipment to meet the new FIPS standards, but this isn’t a very good solution, either. If equipment were to pass the revised FIPS 140 standards, it would have to be operated in “non-NIST mode” to be compatible with the digital cinema standards. In short, equipment manufacturers would have to jump over expensive, high hurdles to obtain FIPS compliance under new rules, but would have to dumb it down to non-FIPS-compliant operation for digital cinema use. As ridiculous as this idea sounds, it is sadly a plausible one.
Oracles on high say that the studios, the people who got us into this mess, have an elegant answer. It has the ring of an oxymoron. But given the ridiculous alternative, there is certainly no harm in waiting to learn the proposal. Much good would take place if the oracles speak the truth. But the long term problem will likely persist. Be prepared for the industry to revisit this problem again and again.