Multi-level Information Security in Information Flow Tracking
Industries that require security (trusted platforms, secure storage, network devices, etc.) and/or integrity (real-time operating systems, critical embedded system controllers, etc.) stand to gain the most from this technology. Even automobile OEMs will require secure systems, now that all new vehicles are controlled by dozens of embedded microcontrollers. A very high demand for this technology is anticipated, as it is the first tool in the industry to formally validate security and integrity properties spanning across hardware and software, enabling more efficient solutions while maintaining system integrity.
To allow full account for information flow security in critical systems, researchers have proposed Gate-Level Information Flow Tracking (GLIFT). GLIFT monitors all digital information flows by tracking individual bits through Boolean gates. At such a low level of abstraction, GLIFT is able to capture all transition activities including register to register timing. As a result, all digital information flows are made explicit, including timing channels that are inherent in the underlying hardware implementation but invisible to programmers.Previous work has illustrated the employment of GLIFT for building verifiably information flow secure high-assurance systems. GLIFT has been shown to be effective in detecting timing channels in bus protocols such as I2C and USB. Although GLIFT provides an effective approach for enforcing information flow security, the existing GLIFT method targets a two-level linear security lattice and thus only considers two-level security labels, e.g., trusted < untrusted or, the dual, unclassified < confidential. However, most systems benefit from or require multi-level security (MLS). For example, data objects are usually classified into at least four security levels, namely Top secret, secret, confidential and unclassified in military systems. A two-level linear security lattice simply cannot be used for modeling such a policy. In addition, many systems tend to be interested in non-linear lattices for modeling security policies.For example, it is often desirable to have a policy which requires isolation of the highest security level (Top Secret) from several incomparable entities (e.g., Secret US and Secret UK). That is, the model specifies that Secret US and Secret UK are at the same level but represent two different objects. More specifically, Top Secret might be the label for a data encryption process which requires that Secret US and Secret UK learn nothing other than the cipher-text while it is perfectly secure for processes SecretUS and Secret UK to learn information about one another. Thus, there is a need to expand GLIFT to more general security lattices in order to adapt to a wider range of systems.Given here is an advancement of the GLIFT binary security model, enabling multi-level security for information flow tracking systems built into the hardware level.This model classifies data objects in a system into different security levels, tracks the flow of information between security domains, and enforces a specific security policy such as non-interference. While non-interference is a strong and useful security policy, it requires tight information flow control (IFC) to prevent unintended interactions between different system components resulting from harmful flows of information.
20160026801
Tech ID/UC Case 23222/2013-209-0 Related Cases 2013-209-0
USA
