Abstract | ||
---|---|---|
Many-core microprocessor architectures are quickly becoming prevalent in data centers, due to their demonstrated processing power and network flexibility. However, this flexibility comes at a cost; co-mingled data from disparate users must be kept secure, which forces processor cycles to be wasted on cryptographic operations. This paper introduces a novel, secure, stream processing architecture which supports efficient homomorphic authentication of data and enforces secrecy of individuals' data. Additionally, this architecture is shown to secure time-series analysis of data from multiple users from both corruption and disclosure. Hardware synthesis shows that security-related circuitry incurs less than 10% overhead, and latency analysis shows an increase of 2 clocks per hop. However, despite the increase in latency, the proposed architecture shows an improvement over stream processing systems that use traditional security methods. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1145/3075564.3075592 | Conf. Computing Frontiers |
Keywords | Field | DocType |
stream processing, secure processing, network-on-a-chip, homomorphic MAC | Homomorphic encryption,Authentication,Computer science,Cryptography,Latency (engineering),Parallel computing,Microprocessor,Network on a chip,Real-time computing,Stream processing,Cloud computing | Conference |
Citations | PageRank | References |
0 | 0.34 | 10 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jeff Anderson | 1 | 23 | 4.05 |
tarek elghazawi | 2 | 697 | 84.30 |