一个支持可信主体特权最小化的多级安全模型

A Multi-Level Security Model with Least Privilege Support for Trusted Subject

对已有多级安全模型的可信主体支持进行回顾和分析,提出了DLS(离散标记序列)多级安全模型.该模型将可信主体的生命周期分解为一系列非可信状态,对每一个状态赋予一个敏感标记.可信主体的当前敏感标记等于当前非可信状态的敏感标记,非可信状态的切换由预定义的可信请求事件触发.从而可信主体的当前敏感标记可以根据其应用逻辑而动态调整.同时给出了模型保持系统安全性的安全状态和规则.与Bell模型等可信主体敏感标记范围模型相比,该模型在多级安全的策略范围内实现了可信主体的特权最小化.

The trusted subject supports of the existing multi-level security models are reviewed and a new model called DLS （discrete label sequence） is proposed. It decomposes the lifecycle of a trusted subject into a sequence of untrusted states （US）. Each untrusted state is associated with a certain current security label, and only the predefined trusted request events （TRE） can trigger the transition from one US to the other. Thus, the current security level of a trusted subject is dynamically changed according to its application＇s logic. Definitions of secure states and rules to preserve security are also presented. Compared with the trusted subject implemented by security level range, this model gives a better support of least privilege and achieves the support within the MLS policy framework.