| Description |
Covert communications are useful when the communicating parties would like to hide the presence of any secret communications. In this thesis, we explore the area of designing hard-to-detect covert systems. Detection work has typically focused on anomaly-based detection. We nd that current methods to thwart anomaly detection, which are based on calculating a conservative estimate for the probability of detection, are not as practical, generically applicable, and sufficiently accurate as would be necessary in real world applications. We propose a new method which allows us to quantify the detectability of a covert system to meet these needs. Furthermore, to thwart brute-force style detection attacks, we propose a new design paradigm for covert systems, one that uses multiple colluding receivers. Our results show that multiple colluding receivers in a covert system has a signi cant e ect on the detectability of the system. To validate our methods, we implement a multiple receiver covert system and run a number of experiments. We nd that our detection quanti cation is much more accurate than previous work, and our multiple receiver design paradigm makes brute-force detection much more difficult for the adversary and also aids in thwarting anomaly based detection. |
