We contend that reasoning about knowledge is both natural and pragmatic for verification of electronic voting protocols. We present a model in which desirable properties of elections are naturally expressed using standard knowledge operators, and show that the associated logic is decidable (under reasonable assumptions of bounded agents and nonces).

No abstract

An important problem in the analysis of security protocols is that of checking whether a protocol preserves secrecy, i.e., no secret owned by the honest agents is unintentionally revealed to the intruder. This problem has been proved to be undecidable in several settings. In particular, [11] prove the undecidability of the secrecy problem in the presence of an unbounded set of nonces, even when the message length is bounded. In this paper we prove that even in the presence of an unbounded set of nonces the secrecy problem is decidable for a reasonable subclass of protocols, which we call context-explicit protocols.

In [22], we extend the Dolev-Yao model with assertions. We build on that work and add existential abstraction to the language, which allows us to translate common constructs used in voting protocols into proof properties. We also give an equivalencebased definition of anonymity in this model, and prove anonymity for the FOO voting protocol. AnonymityFormal verification of security protocols often involves the analysis of a property where the relationship between an agent and a message sent by him/her needs to be kept secret. This property, called "anonymity", is a version of the general unlinkability property, and one of much interest. There can be multiple examples of such anonymity requirements, including healthcare records, online shopping history, and movie ratings [20]. Electronic voting protocols are a prime example of a field where ensuring and verifying anonymity is crucial.It is interesting to see how protocols are modelled symbolically for the analysis of such properties. In the Dolev-Yao model [10], one often requires special operators in order to capture certain behaviour. Many voting schemes employ an operation known as a blind signature [8]. A blind signature is one where the underlying object can be hidden (via a blinding factor), the now-hidden object signed, and then the blind removed to have the