Token-Leakage Tolerant and Vector Obfuscated IPE and Application in Privacy-Preserving Two-Party Point/Polynomial Evaluations

Abstract

In mobile environments, data stored in nodes are subject to side-channel attacks such as power analysis, emitted signal, detected radiation, etc. In this work, we propose a leakage-resilient inner-product encryption that the decryption will succeed if and only if the decryption attribute vector (generate the token) meets the orthogonal encryption attribute vector (obfuscated encryption policy), that is, the match holds that the inner product of two vectors is zero. Propose scheme supports the security of attribute-hiding and leakage-resilient in the standard model. The adversary cannot only issue any token reveal query on non-match vector, but also can request at most |$\ell $|-bit information on the token-leakage query even if the queried vector matches the challenge vector. We prove the security by the technique of dual system encryption in the orthogonal subgroups, to be strongly leakage-resilient and adaptively attribute-hiding. We also deploy our scheme as a building block to devise a secure two-party point/polynomial evaluation protocol in mobility environments, in which two parties cooperate to evaluate a polynomial in the sense that their sensitive inputs of both point and polynomial are fully preserved. Finally, we assess the performance of leakage resilience including the leakage bound and the leakage fraction (LF). Analysis shows that the leakage bound is approximate |$(n-1)\log {\pi _2}$| and the LF is about |${1}/{2(1+\omega _1+\omega _3)}$|⁠, where |$n$| is the length of vector, |$\pi _2$| is the order of subgroup |$\mathbb {G}_{\pi _2}$| and |$\omega _1,\omega _3$| are the constants. We can obtain optimized LF |$1/2-o(1)$| by varying the sizes of subgroups.