V. CONCLUSION
In this paper, we firstly point out that trust of platform
hardware is fundamental to a trusted computing platform.
Then, we propose the notion of Hardware Integrity
Measurement, whose purpose is to measure and collect the
trust information of platform hardware devices. Finally, a
protocol is presented.
In [6], the authors show how to design and implement a
malicious processor, and then use the malign processor to
attack a system. This kind of hardware-based attack can
easily bypass traditional access control and, therefore, is hard
to detect. The authors in [6] further discuss the possibilities
of using testing, reverse engineering and fault-tolerance to
defend against the hardware-based attacks. However, they
draw the discouraging conclusion that even the seeming
effective fault-tolerance cannot be directly applied into
practices, because the cost may be too high. So, our notion of
measuring hardware integrity can be regarded as its
supplement.
Of more importance, as existing work shows, hardwarebased
attacks like [6] can easily defeat the traditional security
models, because these models are software-based without
taking hardware into consideration. In the future, we will try
to improve the traditional security models by involving trust
of platform hardware.
V. CONCLUSIONIn this paper, we firstly point out that trust of platformhardware is fundamental to a trusted computing platform.Then, we propose the notion of Hardware IntegrityMeasurement, whose purpose is to measure and collect thetrust information of platform hardware devices. Finally, aprotocol is presented.In [6], the authors show how to design and implement amalicious processor, and then use the malign processor toattack a system. This kind of hardware-based attack caneasily bypass traditional access control and, therefore, is hardto detect. The authors in [6] further discuss the possibilitiesof using testing, reverse engineering and fault-tolerance todefend against the hardware-based attacks. However, theydraw the discouraging conclusion that even the seemingeffective fault-tolerance cannot be directly applied intopractices, because the cost may be too high. So, our notion ofmeasuring hardware integrity can be regarded as itssupplement.Of more importance, as existing work shows, hardwarebasedattacks like [6] can easily defeat the traditional securitymodels, because these models are software-based withouttaking hardware into consideration. In the future, we will tryto improve the traditional security models by involving trustof platform hardware.
การแปล กรุณารอสักครู่..
