Paper 2011/031

Efficient Unconditional Asynchronous Byzantine Agreement with Optimal Resilience

Ashish Choudhury and Arpita Patra

Abstract

We present an efficient and optimally resilient Asynchronous Byzantine Agreement (ABA) protocol involving n = 3t+1 parties over a completely asynchronous network, tolerating a computationally unbounded Byzantine adversary, who can control at most t parties out of the n parties. The amortized communication complexity of our ABA protocol is O(n^{3} \log \frac{1}{\epsilon}) bits for attaining agreement on a single bit, where \epsilon (\epsilon > 0) denotes the probability of non-termination. We compare our protocol with the best known optimally resilient ABA protocols of Canetti et al.(STOC 1993) and Abraham et al.~(PODC 2008) and show that our protocol gains by a factor of O(n^{8} \log \frac{1}{\epsilon}^{3}) over the ABA protocol of Canetti et al. and by a factor of O(n^{5} \frac{\log{n}}{\log \frac{1}{\epsilon}}) over the ABA protocol of Abraham et al. in terms of the communication complexity. To design our protocol, we first present a new, optimally resilient statistical asynchronous verifiable secret sharing (AVSS) protocol with n = 3t+1, which significantly improves the communication complexity of the only known optimally resilient statistical AVSS protocol of Canetti et al. Our AVSS protocol shares multiple secrets simultaneously and incurs lower communication complexity than executing multiple instances of an AVSS protocol sharing a single secret. To design our AVSS protocol, we further present a new asynchronous primitive called asynchronous weak commitment (AWC), which acts as a substitute for asynchronous weak secret sharing (AWSS), which was used as a primitive for designing AVSS by Canetti et al. We observe that AWC has weaker requirements than the AWSS and hence can be designed more efficiently. The common coin primitive is one of the most important building blocks for the construction of an ABA protocol. The best known common coin protocol of Feldman et al. requires multiple instances of an AVSS protocol sharing a single secret as a black-box. Unfortunately, this common coin protocol does not achieve its goal when the multiple invocations of AVSS sharing a single secret is replaced by a single invocation of an AVSS protocol sharing multiple secrets simultaneously. Therefore in this paper, we extend the existing common coin protocol to make it compatible with our new AVSS protocol (sharing multiple secrets). As a byproduct, our new common coin protocol is much more communication efficient than the existing common coin protocol.