Tolerating generalized mobile adversaries in secure multiparty computation

We study a distributed adversarial model of computation in which the faults are non-stationary and can move through the network (like viruses) as well as non-threshold (there is no specific bound on the number of corrupted players at any given time). We show how to construct multiparty protocols that are perfectly secure against such generalized mobile adversaries. The key element in our solution is devising non-threshold proactive verifiable secret sharing schemes that generalize the secret sharing schemes known in the literature.