1. Usable Security Lab Crypto Lab Efficiency Comparison of Various Approaches in E-Voting Protocols Oksana Kulyk, Melanie Volkamer

2.  Many e-voting protocols exist  Election settings widely differ  Different security requirements  Different non-security requirements  Different ballots, voting rules, and number of voters  No „one-size-fits-all“ solution  Research question: How to choose the most appropriate one? Motivation of our research project Oksana Kulyk | Voting - FC 2016 Workshop2

3. Proposal: Decision support system Oksana Kulyk | Voting - FC 2016 Workshop3 Module Database Decision Support System Specification of election setting E-voting protocols comparison result

4.  Many e-voting protocols exist, focus on mix net and homomorphic tallying based  Election settings widely differ  Different security requirements  Different non-security requirements, focus on efficiency  Different ballots, voting rules, and number of voters  No „one-size-fits-all“ solution  Research question: How to choose the most appropriate one? Focus of this paper Oksana Kulyk | Voting - FC 2016 Workshop4

5. Setup Election key generation Pre- computations Vote casting Encryption (Client) Well-formedness proof (Client) Verify proofs (Server) Vote anonymisation Mixing (Trustees) Zero-knowledge proof of shuffle validity Tallying Decryption (Trustees) Result decoding Verify proofs Protocol overview Oksana Kulyk | Voting - FC 2016 Workshop5 Computing a homomorphic sum

6.  Mix-net  Terelius-Wikström 2010, Wikström 2014 (Implementation)  Homomorphic tallying focusing on different ballot types  Helios v.4 (Implementation)  Groth 2004  Joaqim 2014 Vote anonymisation Oksana Kulyk | Voting - FC 2016 Workshop6

7.  Used regardless of anonymisation approach  Performed by one or several trustees  Variants  Single key holder  Non-threshold key distribution  Threshold key distribution  Verifiable threshold distributed decryption  Pedersen 1992 Tallying – Decryption Oksana Kulyk | Voting - FC 2016 Workshop7

8. HeliosGrothJoaqim Approval kMin...kMax of LYesNoYes 0...L of LYes k...k of LYes Divisive (T,T) of L (use all T votes)Yes No (t,T) of L (use all T votes)YesNo (t,T) of LYesNoYes Ranking k of L (k < L)No Yes L of LNoYes Ballot types Oksana Kulyk | Voting - FC 2016 Workshop8 Wikström (Mix) Yes

9. SetupVote casting Vote anonymisation Tallying Considering different phases Oksana Kulyk | Voting - FC 2016 Workshop9

10. Setup How much data can be precomputed? Vote casting How long does it take for a voter to cast the vote? How long does it take to verify the well-formness of all cast votes? Vote anonymisation How long does it take to anonymise the votes? Tallying How long does it take to decrypt the result? How long does it take to decode the result? Relevant questions for each phase Oksana Kulyk | Voting - FC 2016 Workshop10

11.  Count modular exponentiations for each approach  Assumptions  Optimized algorithms are in place  Fixed-base  Multi-exponentiations  Pre-computations are done by the server during setup Methodology Oksana Kulyk | Voting - FC 2016 Workshop11

12.  Number of voters (expected and maximal)  Number of candidates  Number of trustees  Ballot type  Other ballot type specific parameters  Device performance (client and server) Relevant parameters Oksana Kulyk | Voting - FC 2016 Workshop12

13. Tool implementation Oksana Kulyk | Voting - FC 2016 Workshop13

14.  Approval voting  Swiss referendums  Estonian elections  Norway elections (simplified version)  IACR elections  Boardroom voting  Divisive voting  Local elections, state of Hesse  Borda ranking  Victoria state elections Case studies Oksana Kulyk | Voting - FC 2016 Workshop14

15. Swiss referendum Oksana Kulyk | Voting - FC 2016 Workshop15

16. Estonia elections Oksana Kulyk | Voting - FC 2016 Workshop16

17. Norway elections Oksana Kulyk | Voting - FC 2016 Workshop17

18. IACR elections Oksana Kulyk | Voting - FC 2016 Workshop18

19. Boardroom voting Oksana Kulyk | Voting - FC 2016 Workshop19

20. Local elections, State of Hesse Oksana Kulyk | Voting - FC 2016 Workshop20

21. Victoria state election Oksana Kulyk | Voting - FC 2016 Workshop21

22.  Mix net shuffling  Complex ballots  Large number of voters  Small number of trustees  Helios  Average ballot complexity  Average number of voters  Large variations in number of voters  Groth 2004  Very simple ballots Conclusions Oksana Kulyk | Voting - FC 2016 Workshop22

23.  More complex election rules  Implementation-specifics  Communication load  Number of available servers  Storage load  Trade-off with other criteria Future work Oksana Kulyk | Voting - FC 2016 Workshop23