Injecting Heterogeneity Through Protocol Randomization

In this paper, we argue that heterogeneity should be an important principle in design and use of cryptographic protocols. We use automated formal analysis tools to randomly generate security protocols as a method of introducing heterogeneity. We present the results of simulations for the case of two party authentication protocols and argue that choosing protocols randomly out of sets numbering in the hundreds of millions is practical and achievable with an acceptable overhead. To realize the simulation, we implemented a highly efficient protocol verifier, achieving approximately two orders of magnitude improvement in performance compared to previous work.