Now throughout all of these standards and frameworks, they each embody different kinds of trusted computing principles. And the ones that you see on the screen now: controllability, security, privacy, interoperability, portability, and ease of use, tend to be shot throughout all of them in various potentially unique ways but each framework standardizes how it talks about each of these areas, so that it has a common definition throughout the structure of that particular framework and one that it usually shares with almost every other standard framework. And here you see the trusted computing principles, the results of us adopting all of these standards and their common criteria across all different kinds of systems and software development.

We have resistance, which is the product is built to withstand attempts to subvert normal operations within predetermined design limits; robustness, essentially the ability to withstand different kinds of impacts without a complete failure; resilience, which means that it has the flexibility of functionality such that it can continue even after attack or an error; recoverability, of course, is critical given that it has the features and structure necessary to facilitate trusted recovery; that it has a certain level of redundancy within it; and that it will perform in a reliable and predictable manner, protecting the qualities of trust and assurance. One way that it does it is through the protection ring model, which has come to be part of many systems' designs.

Now the ring model itself was conceptualized in the Multics Operating System under Project Guardian at MIT. In it, they developed four different rings beginning with ring zero surrounding the OS Kernel. And it rings one, two, and three various portions of the systems, the assets that are within it, the command structure, program execution, features, and user rights are contained in increasingly low levels of control as it moves further out from ring zero, which encompasses the innermost kernel. This four-ring model separated and implemented defense in-depth with its four rings by what each one included. The innermost ring could not be, for example, attacked or accessed directly. Anyone having the privilege to be able to do so would have to go through rings three, two, and one in that order before getting to ring zero. So having various interfaces, various pathways described and various gateway mechanisms in between the layers were absolute requirements for any access of anything beyond the ring that an entity was in at the moment in order to be put through multiple levels of controls before they were able to reach their destination.

Now what we have in current use through most of our systems today is a two-ring version of a ring model also showing defense in depth with rings three and zero adapted from the four rings that you saw on the previous slide. The functions and assets that you saw in four rings have been put into these two, basically with ring zero and all of the included elements within it being for the highest privilege mode or administrator or cis admin or pseudo with all privileged and non-privileged instructions possible to someone with the proper authorization. Then ring three, notice the application level, the lowest level of privilege or the user level and non-privileged instructions only would be possible.

A concept that was first discussed in the orange book, the idea of a trusted computing base captured the various features that you see here. Now, the trusted computing base is a term that is coming back into regular usage. It combines all the elements of hardware, software, firmware, and the controls within it to ensure that all of the elements within the TCB are at the most carefully managed and controlled assets within a system environment. In it, we have protected objects which may be known but cannot be accessed directly. They can only be accessed through a trusted process governed by policy by an authorized subject. Part of this was the trusted platform module, which was a specific module designed to store cryptographic keys and authentication materials for various portions of a system.

Now, the TCB is surrounded by a security perimeter, which separates it from all non-TCB components. And across the perimeter, there are gateway portions and interfaces through which only authorized and cleared elements can pass. That doesn't mean that they're classified going in and unclassified coming out. It means that there are rules, permissions, and authorizations granted to any subject coming into the TCB from outside crossing that perimeter. coming into the TCB from outside crossing that perimeter. And that is to make sure that the TCB itself is preserved secure and that its integrity is kept intact. With regard to the trusted computing base, there is a mechanism within that ensures the enforcement of the security policy in the specific system. And this is known as the reference monitor. Now in all systems and in all access control modes, the reference monitor must be found to be working under all operational conditions. In fact, it is a requirement of the system in which it works that if it is not running, the system is not running because the reference monitor is at the very heart of the security policy in terms of its rules and its enforcement.

The reference monitor is one of the very few programs for which exhaustive testing in all possible modes and access methods is required. It has to mediate 100% of all of the subject object interactions that are possible within its system context. And, as I said, it must be running 100% of the time that the system itself is in operation. And as you see in the diagram we have the reference monitor, essentially an abstract or a virtual machine that does this particular job of policy enforcement. It mediates the interactions between the subjects and objects. It has a security kernel in which the rules governing those interactions are housed and it has an audit file that it will use to record every single activity that it enforces. So when the subject attempts access to an object, the reference monitor refers to the rules database, allows or denies the access attempt to the object, and writes the transaction and its record to the audit file, so that there is a way for any security officer, or any privacy officer, or any auditor to be able to review the records to establish accountability and trace it back to the subject that made the attempt.