A bandwidth computation model for covert storage channels and its applications

A Markov model for bandwidth computation and its application to Secure Xenix are presented. The model can be used for computing the bandwidth of both individual channels and aggregated channels (i.e. serial and parallel aggregation). Using this model, a tool has been built and experiments conducted to determine the factors that affect the bandwidth of covert storage channels (noise, scheduling delays, load, etc.). The tool can be used to compute the minimum delays for each channel under various loads and program behavior. Thus, it allows the placement of dynamically adjustable delays in multiprogrammed systems, which guarantees minimum performance impact. >