姚燕青新_基于非完美随机源的隐私问题的详解.pptVIP

姚燕青 北京航空航天大学 基于非完美随机源的隐私问题的 安全性研究 CACR 2015 Oct. 23~24, 2015 Latest Results Yevgeniy Dodis,?Yanqing Yao: Privacy with Imperfect Randomness.?CRYPTO (2)?2015:?463-482. Yanqing Yao,? Zhoujun Li: Achieving Differential Privacy with Bias-Control Limited Source.?IACR Cryptology ePrint Archive?2015:?360?(2015). Submitted. group elements from DH key exchange physical sources Perfect Randomness Ideality biometric data secret with partial leakage Reality Imperfect Sources Non-Extractable Sources (e.g., [SV86,CG88,Dod01]) Can cryptography be based on imperfect random secrets? Extractable sources (e.g., [vN51,CFG+85,Blu86]) allow for deterministic extraction of nearly perfect randomness are good for (almost) anything possible with perfect randomness Non-extractable sources (e.g., SV sources) sufficient for simulating probabilistic polynomial-time algorithms (e.g., [VV85, SV86, CG88, Zuc96, ACRT99]). sufficient for authentication applications (e.g., MACs [MW97, DKRS06], signature schemes [DOPS04, ACM+14]) Are they sufficient for privacy applications Imperfect Sources Prior work about privacy SV sources are not sufficient for unconditionally secure symmetric encryption [McInnes-Pinkas, CRYPTO 1990] SV sources are not sufficient for computationally secure encryption,commitment, zero-knowledge,secret sharing, etc. [Dodis-Ong-Prabhakaran-Sahai, FOCS 2004] Suggesting traditional privacy requires an extractable source [Bosley- Dodis, TCC 2007] Even if SV source is efficiently samplable The above negative results hold [Austrin-Chung-Mahmoody-Pass-Seth, CRYPTO 2014]. SV sources are provably sufficient for achieving Differential Privacy [Dodis-(López-Alt)-Mironov-Vadhan, CRYPTO 2012]. There is a qualitative gap between traditional differe