UK unis, McAfee collude to beat collusion attacks
EPSRC splashes cash at security
The UK's Engineering and Physical Sciences Research Council (EPSRC) is backing research designed to improve detection of “collusion” between malicious apps on the Android platform.
Collusion attacks use malicious apps with different levels of permissions to bypass Android access controls. For example, one app might request permission to access personal data, but not ask for Internet access. Instead, the user might be encouraged to install a second app that has communication access.
In that scenario, the app with access to data would then pass information to the app with Internet access for transmission back to an attacker.
According to the EPSRC's announcement, this is a relatively little-studied attack model, so researchers at Royal Holloway University of London, City University London, and Coventry and Swansea Universities are sharing £3 million of grants to study the issue (some of the money is also going to teams looking at the UK's cyber-security, the announcement notes).
Intel's McAfee business is also taking part in the research, providing a library of safe apps and assistance in analysing results.
The work is being led by Royal Holloway information security lecturer Dr Lorenzo Cavallaro. The announcement states that his team “will study the behaviour of apps on Android operating systems and develop novel techniques to spot malicious apps, which of course, are designed to remain hidden. They will use this information to enrich or enhance devices to counteract attacks.”
As this paper from ETH Zurich notes, collusion attacks can be hard to spot. “Existing security products … that are used for the analysis of application permissions, analyze and report the permissions independently for each individual application and therefore do not take into consideration application collusion,” it states.
According to that paper, collusion attacks are a consequence of per-app permission models. They undermine the assumption that applications can be independently restricted in their access to resources. ®
Sponsored: Data Loss Prevention & Data Theft Prevention