Information Flow and Execution Policy for a Model of Detection without False Negatives

In this paper, we propose an extension of an intrusion detection system, implemented at the operating system level. This model is based on a flow control policy, expressed at the scale of the system objects. The extension presented here takes the execution of processes mechanisms into account and improves the expressivity of the security policy. The model thus becomes usable to restrain information flows realized by processes depending on the user and/or the code of the process. Besides, we prove that this model does not produce false negatives : all the violations of the security policy raise an alert.