The security of information systems plays an important role in national and defence securities. The research of secure operating systems, being the "base" of information systems, has been regarded as a strategic problem. In this paper, some concepts of computer systems' security are firstly proposed, and the technique of access control is analyzed in detail. Then, the security mechanism of Linux is elaborated and some insecure factors in Linux are pointed out. Several approaches are provided to construct secure operating systems. The paper mainly discusses diverse security policies in operating systems. By the GFAC framework, we implement several security policies in Linux kernel version 2.2.14, including HRU model based fine grain discretionary access control, multi-level security policy based mandatory access control and Biba model based integrity access control. Finally, the paper tries to introduce artificial intelligence technology into the construction of secure operating systems. Rough sets and vague sets based security rule acquisition models are presented. They can be used to construct the secure operating system kernel flexiblely. |