Conditional Disclosure of Secrets: Amplification, Closure, Amortization, Lower-bounds, and Separations Benny Applebaum Barak Arkis Pavel Raykov Prashant Nalini Vasudevan
Conditional Disclosure of Secrets [GIKM00] π: 0,1 π Γ 0,1 π β {0,1} πΊ -Correctness: If π π¦, π§ = 1, then for any π‘ , Randomness π Pr π· π¦, π§, π π΅ , π πΆ = π‘ > 1 β Ξ΄ π¦ π§ A B Secret π‘ π -Privacy: If π π¦, π§ = 0 , then for any π‘ , π π΅ π πΆ Ξ πππ π¦, π§ ; π π΅ , π πΆ < π Communication: π π΅ + |π πΆ | C π¦, π§ Randomness: |π |
Connections and Applications β’ Attribute-Based Encryption. [Att14,Wee14] β’ Secret-sharing for certain graph-based access structures. β’ Light-weight alternative to zero-knowledge proofs in some settings. [AIR01] β’ Data privacy in information-theoretic PIR. [GIKM00] β’ A minimal model of multi-party computation.
What Was Known Earlier Upper bounds: β’ Communication 2 π( π log π) for any predicate on π -bit inputs. [LVW17] β’ Communication π(π) for predicates with size- π branching programs or span programs. [IW14,AR16] Lower bounds: β’ Explicit predicate that requires Ξ©(log π) bits of communication. [GKW15] β’ Same predicate requires Ξ© π bits for linear CDS. [GKW15]
CDS and Statistical Difference Randomness π πΊ -Correctness: π§ π¦ A B If π π¦, π§ = 1, then for any π‘ , Secret π‘ Pr π· π¦, π§, π π΅ , π πΆ = π‘ > 1 β Ξ΄ π π΅ π πΆ 0 1 β‘ Ξ π π΅ , π πΆ π¦,π§ ; π π΅ , π πΆ π¦,π§ > 1 β 2π C π¦, π§ π -Privacy: If π π¦, π§ = 0 , then for any π‘ , Distribution of (π π΅ , π πΆ ) : Ξ πππ π¦, π§ ; π π΅ , π πΆ < π 0 β’ input (π¦, π§) , π‘ = 0 : π π΅ , π πΆ π¦,π§ β’ input (π¦, π§) , π‘ = 1 : π π΅ , π πΆ π¦,π§ 1 0 1 β‘ Ξ π π΅ , π πΆ π¦,π§ ; π π΅ , π πΆ π¦,π§ < 2π
Separations Explicit function ππ·ππ: 0,1 4n log π Γ 0,1 2n log π β 0,1 that has: β’ CDS complexity: π(log π) β’ Randomized communication complexity: Ξ©(π 1/3 ) β’ Linear CDS complexity: Ξ©(π 1/6 ) Inspired by oracle separations between SZK and other classes [Aar12], and the Pattern Matrix method [She11].
Collision Problems π log π β π¨ : 0,1 log π β 0,1 log π π¨ β β β β π¨ π = π π’β block in π¨ log π π blocks β β π¨ (π) is uniformly distributed 0 if β π¨ is 1βtoβ1 π·ππ π¨ = α 1 if β π¨ is 2βtoβ1 β β π¨ (π) is far from uniform
Collision Problems 4π log π π¦ β β β 0 0 1 0 1 0 1 0 0 1 1 1 1 0 0 1 π§ ππ·ππ π¦, π§ = π·ππ(π¦ π§ ) 3 2 2 β β β 4 π¦[π§] 1 0 1 β β β 1 π log π π ππ·ππ > Ξ©(π 1/3 ) linCDS ππ·ππ > Ξ©(π 1/6 ) ([Amb05,Kut05] + [She11]) (left + [GKW15])
Collision Problems π¦ A 0 0 1 0 1 0 1 0 0 1 1 1 β β β 1 0 0 1 π¦, π§ π‘ π C π§ B 3 2 2 β β β 4 Use PSM [FKN94] to send: β’ β π¦ π§ (π) if π‘ = 0 π¦[π§] β’ π β 0,1 log π if π‘ = 1 1 0 1 β β β 1 log π If ππ·ππ π¦, π§ = 0 , both are π blocks the same distribution, else they are far apart.
Closure β - Boolean formula over 0,1 π of size π CDS for each of CDS for π 1 , β¦ , π β(π 1 , β¦ , π π ) π Comm: π’ 1 , β¦ , π’ π Comm: π β ππππ§(π’ π , π π ) Rand : π 1 , β¦ , π π Rand : π β ππππ§(π’ π , π π ) Construction uses transformations for Statistical Difference [SV03,Oka96], and PSM protocols [FKN94].
Amplification CDS for π CDS for π Single-bit secret π -bit secret Corr: 2 βΞ©(π) Corr: 0.1 Priv: 2 βΞ©(π) Priv: 0.1 Comm: π’ Comm: π(ππ’) Construction uses constant-rate ramp secret-sharing schemes [CCGdHV07]. Incomparable version follows from the Polarization Lemma [SV03].
Lower Bound There exists a predicate π: 0,1 π Γ 0,1 π β {0,1} for which any perfect (single-bit) CDS requires communication at least 0.99π . Proven by reduction to the PSM lower bound of [FKN94]. Earlier bound was explicit, Ξ©(log π) bits. [GKW15]
Amortization For any predicate π: 0,1 π Γ 0,1 π β {0,1} and π > 2 2 2π , there is a perfect CDS protocol for π with π -bit secrets with communication complexity π(ππ) . Proven using techniques from the amortization of branching programs [Pot16]. π -fold repetition of best known general protocol [LVW17]: π β 2 π( π log π)
Summary We prove the following properties of CDS: β’ Lower Bounds: Non-explicit, Ξ©(π) . β’ Separation: From insecure communication and linear CDS. β’ Amortization: π(π) per bit of secret, if there are more than 2 2 2π bits. β’ Closure: Under composition with formulas. β’ Amplification: Of correctness and privacy from constant to 2 βΞ©(π) with π(π) blowup. To note: β’ Connections with Statistical Difference and SZK. β’ Barriers to PSM lower bounds.
Recommend
More recommend