| Distributed protocols, e.g. consensus, leader election, atomic commit, are the buiding blocks of all network applications, from everyday file-sharing systems to complex grid applications for solving grand challenges like protein folding, financial modeling etc. They have been widely studied and well understood. Traditionally, the studies mostly regard computation-intensive tasks and the main concerns are correctness, fault-tolerance and efficiency. Since the growth of internet and mobile communication infrastructures, the scope of network applications has extended towards service-oriented tasks: e-commerce, electronic voting, auctions, content-exchange systems, mobile ad-hoc applications etc. Here new concerns, like guaranteeing privacy, fairness, or authenticity, play an essential role. These issues are generated by possible malicious behaviour in the network and they do not replace, but come in addition to the old correctness, fault-tolerance and efficiency problems. This project aims at a systematic study of both theoretical and practical aspects of designing and analysing such secure distributed protocols, taking as starting point the well-developed theory and practice of distributed protocols. We wish to redefine existing theoretical failure models to accommodate malicious behaviour and thus obtain new attack models. These models will then be used as setting for deriving theoretical impossibility and complexity results that will clarify the borders of what is achievable by security protocols. Furthermore, these models will be also used as framework for formally analysing existing security protocols and designing new ones. |
