Secure Multiparty Computation (MPC) allows a set of mutually distrusting parties to jointly perform a computation on their private inputs in a way no information about their inputs is revealed, except the output of the computation. The use of MPC is promising in real-life situations that demand both privacy and computation at the same time. The aim of my research is to advance our understanding of the feasibility of tasks related to MPC, and to construct efficient MPC protocols in various adversarial and network settings. More specifically, I am interested in in the following three research directions:

Resource-Efficient MPC. One of the fundamental metrics to analyze efficiency of MPC protocols is communication complexity, that measures the total communication of the protocol across all parties. However, this standard metric ignores the practically important aspect of load-balancing. We explore the notion of Bottleneck Complexity (defined as maximum communication complexity of any party in the protocol) that addresses this issue.

Realistic Adversaries in MPC. As per the standard security definition in MPC, a protocol is considered secure if an adversary learns nothing about honest parties’ inputs (other than what can be derived from the output). We observe that this definition permits honest parties to learn other honest parties’ inputs, which is clearly undesirable in real-life. This motivates the notion of ‘friends-and-foes’ (FaF) security, that addresses this issue.

You Only Speak Once (YOSO) MPC. The above two research directions — resource efficiency, and modelling realistic adversaries — seem to conflict since achieving security against more powerful adversaries is, intuitively, easier when the protocols use more resources. However, the large-scale distributed setting of YOSO MPC manages to strike a balance between security and efficiency.