Trade-off between security and performance has become an intriguing area in recent years in both the security and performance communities. As the security aspects of security protocol research is fully- edged, this thesis is therefore devoted to conducting a performance study of these protocols. The long term objective is to translate formal de nitions of security protocols to formal performance models automatically, then analysing by relevant techniques. In this thesis, we take a preliminary step by studying five typical security protocols, and exploring the methodology of construction and analysis of their models by using the Markovian process algebra PEPA. Through these case studies, an initial framework of performance analysis of security protocol is established. Firstly, a key distribution centre is investigated. The basic model su ers from the commonly encountered state space explosion problem, and so we apply some efficient solution techniques, which include model reduction techniques and ordinary di fferential equation based fluid flow analysis. Finally, we evaluate a utility function for this secure key exchange model. Then, we explore two non-repudiation protocols. Mean value analysis has been applied here for a class of PEPA models, and it is compared with an ODE approximation. After that, an optimistic nonrepudiation protocol with off-line third trust party is studied. The PEPA model has been formulated using a concept of multi-threaded servers with functional rates. The nal case study is a cross-realm Kerberos protocol. A simplified technique of aggregation with an ODE approximation is performed to do efficient cient analysis. All these modelling and analysis methods are illustrated through numerical examples.
展开▼
