Accountable Storage

Giuseppe Ateniese, Michael T. Goodrich, Vassilios Lekakis, Charalampos Papamanthou, Evripidis Paraskevas, Roberto Tamassia

Abstract: We introduce Accountable Storage (AS), a framework enabling a client to outsource n file blocks to a server while being able (any time after outsourcing) to provably compute how many bits were discarded or corrupted by the server. Existing techniques (e.g., proofs of data possession or storage) can address the accountable storage problem, with linear server computation and bandwidth. Instead, our optimized protocols achieve $$O(\delta \log n)$$ complexity (where $$\delta $$ is the maximum number of corrupted blocks that can be tolerated) through the novel use of invertible Bloom filters and a new primitive called proofs of partial storage. With accountable storage, a client can be compensated with a dollar amount proportional to the number d of corrupted bits (that he can now provably compute). We integrate our protocol with Bitcoin, supporting automatic such compensations. Our implementation is open-source and shows our protocols perform well in practice.