A few recent publications, which are not (co-)written by STREWS, but may be of interest to the STREWS community.

Study on cryptographic protocols – by ENISA

ENISA (the EU agency for network and information security) published a report on cryptographic protocols. The protocols studied are lower-level protocols, such as TLS, SSH, UMTS and Bluetooth. Higher level protocols, such as HTTP, ofen pass over connections established by such lower-level protocols.

The study especially looks at the way cryptographic methods are used in those protocols, because cryptography that is mathematically strong can easily be applied incorrectly: The same algorithms that works in one protocol can thus fail to protect data in another, or provide only weak protection. The error may be due to bugs in implementations (e.g. the heartbleed bug), but the report looks more specifically at design errors in the protocol (e.g., the padding weakness in SSLv3).

IAB Statement on Internet Confidentiality – by the IAB

The IAB (Internet Architcture Board) published a statement in which it talks about the threats of pervasive monitoring and recommends that all new protocols, at all levels, should use encryption. Encryption combined with authentication is best, but even without authentication, encryption already increass the cost for eavesdroppers. The danger of not taking that route is that the trust of people in networks, already degraded, will degrade even more.

This statement is of course an echo of the STRINT workshop (the joint W3C/IAB workshop organized by STREWS in February 2014), one of the conclusions was exactly that: a recommendation to standards organizations such as IETF and W3C to adopt a policy of encryption everywhere.

Security Collapse in the HTTPS Market – by A. Arnbak et al.

This article (also as PDF) in ACM Queue of September looks at some problems with the certificate system underlying TLS, and thus HTTPS.

Those problems aren't new: The article discusses security breaches that all occurred before 2012 and risks that have been known in the IETF and security research communities for some time. One such problem is the fact that any Certificate Authority (CA) can issue certificates for any domain, even if another CA already issued one earlier. Thus, a single bad CA can put the whole Web at risk.

But the fact that the issues are known doesn't mean the underlying weaknesses in the system have been solved. (That is also the reason they came up again at the STRINT workshop last February.)

The article lists the social and economic reasons why the system is difficult to change. It looks at various attempts by governments to regulate CAs and at some proposed technological solutions, such as Google's Certificate Transparency, but it concludes that the system resists change. The risks to Web security remain high. They even got higher, after the Snowden revelations.

(For some reactions on & criticism of the paper, see the comments on Bruce Schneier's blog.)