Trusted code remote execution through trusted computing and virtualization

Abstract

Given two machines A and B, A has to commit job codes to be executed on B and get the results back. How can A believe that the job codes are executed correctly on B and related privacy information is not revealed? We propose the attack model and the security goals during the remote execution process. To address this research problem, we propose a new method based on Trusted Computing and Virtualization. We extend trusted computing by providing property-based attestation, job keys generating and migrating securely in TPM. We extend virtualization by using TPM and MAC in hypervisor to isolate job VM. We examine the method to the security goals and it shows that this method can solve the problem effectively. © 2007 IEEE.