Secure multi-party computation on incomplete networks.

Varsha Dani, Valerie King, Mahnush Movahedi +2 more

Yasuaki Nishitani, Yoshihide Igarashi

Yilei Wang, Zhe Liu, Hao Wang +1 more

SUMMARY Rational party is a new kind of parties who behave neither like honest parties nor like malicious adversaries. The crux point of rational party is the definition of the utility function, as rational parties only care about how to maximize their utility. In other words, rational parties choose the strategies, which can bring them the highest utilities. In rational secure two‐party computation protocol, the main task is how to boost mutual cooperation to complete the protocol. Social rational secure multi‐party computation (SRSMPC) means that in a social network, some distributed and rational parties with reputation properties want to jointly compute a functionality. The seemingly simple task becomes tough under three conditions. The first condition is that the network composed by parties may not be complete. That is, two parties may not be neighbors and they are connected through other parties. The second is that the network may be not secure. That is, messages may be tempered by malicious parties. The third condition is that parties may run the protocol under incomplete information scenario. That is, parties may have types and each type has a corresponding utility function. Under the first and second conditions, parties need to consider how to securely transmit messages between two parties who are not neighbors. Under the third condition, we propose the Tit‐for‐Tat strategy and prove that mutual cooperation is a sequential equilibrium between two parties. In this paper, we construct an SRSMPC protocol by using mechanism design under incomplete information to facilitate the implementation of the SRSMPC protocol within constant rounds. Meanwhile, newcomers are allowed to participate in the protocol. To the best of our knowledge, this is the first social rational secure computation protocol for multi‐party under an incomplete information scenario and an incomplete network. Copyright © 2013 John Wiley & Sons, Ltd.

Vaya

Yilei Wang, Qiuliang Xu, Zhe Liu

In traditional rational multi-party computation, parties are assumed to have utility functions and they act according to the utility assumptions. In economics and political scenarios, parties in multi-party computation may have some other properties such as reputation which plays an important role on boosting mutual cooperation in multiparty computation. In this paper we put forward a new kind of parties satisfying reputation assumption and utility assumptions. Meanwhile we construct a social rational multiparty computation protocol under incomplete information and incomplete network to facilitate the implementation of the protocol within constant rounds.

Jincheng Zhang

Secure Multi-Party Computation (SMPC) aims to enable computation on sensitive data without revealing the individual inputs of the participants. Traditional SMPC protocols often suffer from significant computational overhead, particularly in terms of communication complexity. This paper proposes a novel approach to SMPC leveraging tensor network techniques inspired by quantum computation. We demonstrate that tensor networks can efficiently represent and manipulate the necessary data structures for SMPC protocols, leading to a substantial reduction in communication overhead and improved scalability. Our method utilizes the inherent structure of tensor networks to minimize the number of exchanges required during computation, effectively addressing a key bottleneck in existing SMPC implementations. We outline the theoretical foundations and present a framework for designing SMPC protocols using tensor networks. This work represents a significant step towards practical and efficient SMPC solutions, particularly for large-scale datasets and complex computations. The key contribution lies in the application of quantum-inspired techniques to solve classical computational problems, specifically within the context of secure computation.