Design and formal analysis of security protocols, an application to electronic voting and mobile payment

Overview & Synopsis

The last decade has seen the massive democratization of smart devices such as phones, tablets, even watches. In the wealthiest societies of the world, not only do people have their personal computer at home, they now carry one in their pocket or around their wrist on a day to day basis. And those devices are no more used simply for communication through messaging or phone calls, they are now used to store personal photos or critical payment data, manage contacts and finances, connect to an e-mail box or a merchant website... Recent examples call for more complex tasks we ask to such devices: Estonia voting policy allows the use of smart ID cards and smartphones to participate to national elections. In 2017, Transport for London launched the TfL Oyster app to allow tube users to top up and manage their Oyster card from their smartphone. As services grow with more complexity, so do the trust users and businesses put in them. We focus our interest into cryptographic protocols which define the exchanges between devices and entities so that such interaction ensure some security guarantees such as authentication, integrity of messages, secrecy… Their design is known to be an error prone task. Thankfully, years of research gave us some tools to improve the design of security protocols, among them are the formal methods: we can model a cryptographic protocol as an abstract process that manipulates data and cryptographic function, also modeled as abstract terms and functions. The protocol is tested against an active adversary and the guarantees we would like a protocol to satisfy are modeled as security properties. The security of the protocol can then be mathematically proven. Such proofs can be automated with tools like ProVerif or Tamarin. One of the big challenge when it comes to designing and formally proving the security an “industrial- level” protocol lies in the fact that such protocols are usually heavier than academic protocols and that they aim at more complex security properties than the classical ones. With this thesis, we wanted to focus on two use cases: electronic voting and mobile payment. We designed two protocols, one for each respective use case and proved their security using automated prover tools. The first one, Belenios VS, is a variant of an existing voting scheme, Belenios RF. It specifies a voting ecosystem allowing a user to cast a ballot from a voting sheet by flashing a code. The protocol’s security has been proven using the ProVerif tool. It guarantees that the vote confidentiality cannot be broken and that the user is capable of verifying their vote is part of the final result by performing a simple task that requires no technical skills all of this even if the user’s device is compromised – by a malware for instance. The second protocol is a payment one that has been conceived in order to be fully scalable with the existing payment ecosystem while improving the security management and cost on the smartphone. Its security has been proven using the Tamarin prover and holds even if the user’s device is under an attacker’s control