Robust and Private Distributed Shared Atomic Memory in Message Passing Networks

Paper i proceeding, 2015

We study the problem of privately emulating shared memory in message passing networks. The system includes $N$ servers, and at most e semi-Byzantine servers that can deviate from the algorithm by sending corrupted data. Moreover, at most f servers can fail and stop. The focus is on coded atomic storage (CAS) algorithms. We present a variant that ensures no information leakage by letting the servers store their data as secret shares. Our enhancement to CAS uses ⌈(N+k+2e)/2⌉-size quorums and Reed-Solomon codes. This enhancement preserves the algorithm ability to function in asynchronous system settings. To the best of our knowledge, we are the first to address the privacy issue when emulating shared memory in message-passing systems.