A conceptual model of self-monitoring multi-core systems


The paper discusses a conceptual model of building a self-monitoring mechanism in a multi-core system such that the health of the entire system can be monitored securely and in an implicit manner. Accordingly, a many-core system is vertically partitioned (from hardware to application level) so that a small set of processors (security core) is separated from other user data processing cores to build a security subsystem. Then the model delegates security handling tasks to this isolated subsystem and uses a set of privileged instructions to monitor the other part (user side) and interact using a secure message exchange protocol. Simulation of the proposed security mechanism using several off-the-shelf software tools will be performed in order to demonstrate the proof-of-concept. © 2010 ACM.

Publication Title

ACM International Conference Proceeding Series