A central question in the realm of cryptographic protocols is whether security can be retained when many instances of a protocol are executed concurrently. This question has received a considerable amount of attention, resulting in a pretty good understanding of what can be done when a majority of protocol participants behave honestly.
Of special interest is the important case of two-party protocols, where a majority of honest participants cannot be assumed, and for which progress has been significantly slower. The slow progress could be attributed to the fact that the most natural way of defining security (i.e. the simulation paradigm) runs into trouble when extended to the concurrent setting.
In this talk I will present a new definition of security for two-party protocols that without trusted set-up:
1) handles an arbitrary number of concurrent executions; and
2) is implementable based on standard complexity assumptions.
No previous solution has been shown to achieve both properties simultaneously. In contrast to previous definitions of secure computation, ours is not simulation-based.
Joint work with Silvio Micali and Rafael Pass.
Alon Rosen is a faculty member in the School of Computer Science at the Herzliya Interdisciplinary Center. Before that he spent two years as a postdoc in the Cryptography Group of MIT's Computer Science and AI Lab, and two years as a postdoc in the Center for Research on Computation and Society at Harvard's department of Electrical Engineering and Computer Science.
He did his Ph.D. at the Weizmann Institute of Science, under the supervision of Oded Goldreich and Moni Naor. He graduated in 2003.
His main fields of interest are Cryptography and Computational Complexity