Publicly Verifiable Non-Interactive Arguments for Delegating Computation OMER PANETH AND GUY ROTHBLUM Meet Ran Computational task: Review submission #123 Accept \ Reject Delegating Computation Review Submission #123 Accept \ Reject + Proof Delegating Computation π π₯ =? Worker Delegator π π₯ = π¦ + Proof Proof System Properties Prover Proof that π π₯ = π¦ (Worker) Verifier (Delegator) β’ Standard completeness and soundness β’ Efficiency: β’ Verifier - π π₯ β ππππ¦πππ (ππ ) β’ Prover - ππππ¦(ππ ) The Best Possible Proofs Trusted party Prover CRS Proof that π π₯ = π¦ Verifier Publicly-Verifiable Non-Interactive 1. Computational soundness (arguments) Arguments for Delegating Computation 2. Trusted common reference string (CRS) (or Publicly-Verifiable Delegation) Constructions from Strong Assumptions 1. In the Random Oracle model [Micali 94] 2. In the plain model based on βknowledge-of-exponentβ assumptions [G10, L12, DFH12, GGPR13, BCCT13, BCIOP13] Can delegate non-deterministic computations (evidence for necessity of non-falsifiable assumptions) [Gentry-Wichs 11] Privately-Verifiable Delegation [Kalai-Raz-Rothblum 14] Challenge Prover π π₯ = π¦ + Proof Challenge +π‘ Verifier πππ(π‘, Proof) Based on (sub-exponentially) secure FHE (or PIR) Do publicly-verifiable delegation exist under natural falsifiable assumptions? Results Publicly-verifiable delegation for all poly-time computations assuming Graded Encodings satisfying a natural hardness assumption. Delegation for bounded depth computations with adaptive soundness based on a milder assumption. Concurrent Work Publicly-verifiable delegation through Obfuscation [Gentry-Lewko-Sahai-Waters 14] + [Lin-Pass-Bitansky-Garg-Telang 14, Canetti-Holmgren-Jain-Vaikuntanathan 14, Koppula-Lewko-Waters 14] 1. non-interactive delegation for poly-time computation from sub-exponential Subgroup Elimination. 2. two-message delegation for bounded-space from polynomial Subgroup Elimination. Next 1. High-level Overview of the first protocol. 2. Graded encodings and the assumption. 3. More details on the first protocol. Not today : second protocol. PCPs to Privately-Verifiable Delegation [Aiello-Bhatt-Ostrovsky-Rajagopalan 00] πΈπ π1 π1 , β¦ , πΈπ ππ ππ Prover PCP Proof π π1 , β¦ , ππ π π1 , β¦ , π ππ Verifier πΈπ π1 π(π1 ) , β¦ , πΈπ ππ π(ππ ) Donβt know how to prove security with arbitrary PCPs [Dwork-Langberg-Naor-Nissim-Reingold 04] Secure with no-signaling PCPs [Kalai-Raz-Rothblum 14] From Private to Public Verifiability πΈπ π1 π1 , β¦ , πΈπ ππ ππ Prover πΈπ π1 π(π1 ) , β¦ , πΈπ ππ π(ππ ) Verifier Our Approach exploit structure - no generic transformation. [Kalai-Raz-Rothblum 14]: Any No-Signaling PCP + Any FHE Privately-Verifiable Delegation This work: A Specific PCP (Sum-Check Protocol) + A Specific Encryption (from Graded Encoding) Publicly-Verifiable Delegation Curve Encryption β High Level Proof: a low degree polynomial π: πΉ π β πΉ. Query: an Input π β πΉ π . Answer: the evaluation π π . π Curve Encryption β High Level Proof: a low degree polynomial π: πΉ π β πΉ. βEncryptedβ query: πΎ π A random low degree curve πΎ: πΉ β πΉ π though π βhiddenβ among other points Curve Encryption β High Level Proof: a low degree polynomial π: πΉ π β πΉ. βEncryptedβ answer: πΎ π the univariate restriction π(πΎ(β )). Example: π πΎ β β‘0 β π π =0 Curve Encryption β High Level Proof: a low degree polynomial π: πΉ π β πΉ. Encrypted query: πΎ π βencodedβ curve πΎ Encrypted answer: βencodedβ restriction π(πΎ(β )) Completely hidden Graded Encodings [Garg-Gentry-Halevi 13] Encoding πΌ 1. πΌ β 2. πΌ β1 ± π½ β × π½ 3. ππ πΌ β β of element πΌ β πΉ under level β β 0, β¦ , π β πΌ±π½ β2 β β πΌ×π½ β1 +β2 if β1 + β2 β€ π β 1 if and only if πΌ = 0 Given polynomial π: πΉ π β πΉ and encodings πΌ1 1 , β¦ , πΌπ β’ If π is of degree β€ π, can test if π πΌ1 , β¦ , πΌπ = 0 β’ Hard for π of degree > π 1 Encoding Curves A curve πΎ: πΉ β πΉ π of degree π: π ππ β π‘ π , πΎ π‘ = ππ = (ππ,1 , β¦ , ππ,π ) β πΉ π π=0 The level-β encoding of πΎ is: πΎ β β ππ,π β πβ 0,π ,πβ[π] Curve Encryption For π β πΉ π let πΎπ be a random degree π curve though π: βπ‘ β πΉ: πΎπ π‘ = π. Plaintext: π β 0,1 π, Ciphertext: πΎπ , Security: for every π0 , π1 β 0,1 Secret key: π‘ 0 . 1 π: πΎπ0 1 βπ πΎπ1 . 1 The Hardness Assumption Intuitively: Hard to evaluate an encoded polynomial on encoded input. Given graded encoding with maximal level π: π πΌπ β π‘ π π‘ 1 , π0 1 , β¦ , πΌπ 1 , π=0 βπ 1 Where π‘, πΌ0 , β¦ , πΌπ , π§ are random in πΉ. π‘ 1 , π0 1 , β¦ , πΌπ 1 , π§ 1 Plausibility of the Assumption π‘ 1 , π0 1 , β¦ , πΌπ 1 , π π=0 πΌπ β π‘π 1 βπ π‘ 1 , π0 1 , β¦ , πΌπ 1 , π§ 1 . β’ Reduction from curve encryption use re-randomization β’ Recent attack [GGH13, CHLRS 14, GHMS14, BWZ14, CLT14] β’ No candidate are known. β’ Alternative assumption: curve encryption is secure (no known attacks in all candidate graded encodings) Back to Delegation Arithmetization of Computation Can write any time π computation as an assignment: π: 0,1 π β 0,1 for π = π(log π) The assignment π is globally consistent if: βπ1 , π2 , π3 β 0,1 π : π π1 , π2 , π3 , π π1 , π π2 , π π3 =0 Where π: πΉ 3π+3 β πΉ is a low-degree arithmetic circuit Protocol Outline πΎ1 1, β¦ , πΎ π Prover π β πΎ1 1 π , β¦ , π β πΎ π π Verifier Proof π π: 0,1 π level- π encoding of the coefficients 1 β . of the univariate restriction π πΎ β 0,1 - assignment π: πΉ π β πΉ β multi-linear extension of π (π is multi-linear and agrees with π on 0,1 π ) Security Proof Outline Prover convinces verifier to accept locally consistent assignment [Kalai-Raz-Rothblum 14] There exists a globally consistent assignment Locally Consistent Assignment [Kalai-Raz-Rothblum 14] An π-local assignment generator is a PPT algorithm: π1 , β¦ , ππ β π΄πΊ(π1 , β¦ , ππ ) 1. Everywhere local consistency: for all π = π1 , β¦ , ππ : Pr πβπ΄πΊ(π) βπ, π, π β π : π ππ , ππ , ππ , ππ , ππ , ππ β 0 β€ ππππ . 2. No-signaling Constructing an Assignment Generator βπ β π sample πΎππ πΎπ1 , β¦ , πΎππ 1 Prover 1 and π‘π π΄πΊ 0 π1 , β¦ , ππ β 0,1 1 π β πΎπ1 , β¦ , π β πΎππ If proof is rejecting: βπ β π obtain [ππ = π qπ = π(πΎππ (π‘π ))] π1 , β¦ , ππ β 0,1 π Protocol Details Recall: π is the multi-linear extension of π. Let π: πΉ 3π β πΉ be the low-degree polynomial: π π1 , π2 , π3 β π π1 , π2 , π3 , π π1 , π π2 , π π3 Goal: verify that: 1. π is consistent with π 2. π π§ = 0 for all π§ β 0,1 3π Only over the CRS curves , (1) The Prover Strategy Given CRS = πΎ 1 1 , β¦ , πΎ π 1 , and computes π, π. 1. βπ β π : ππ π β π πΎ π π 2. βπ, π, π β π : ππ,π,π π1 , π2 , π3 β π πΎ π π1 , πΎ π π2 , πΎ π π3 Proof contains: ππ π , ππ,π,π 3π The Verifier Strategy Given: CRS = πΎ1 1, β¦ , πΎ π 1 , Proof = ππ π , ππ,π,π 3π Test (1) over every three curves π, π, π β π : π π1 , π2 , π3 = π π1 , π2 , π3 , π π1 , π π2 , π π3 , ππ,π,π π1 , π2 , π3 β‘ π πΎ π π1 , πΎ π π2 , πΎ π π3 , ππ π1 , ππ π2 , ππ π3 (1) Back to the Assignment Generator βπ β 3 sample πΎππ 1 and π‘π π΄πΊ 0 π1 , π2 , π3 β 0,1 πΎπ1 , πΎπ2 , πΎπ3 Prover π1 , π2 , π3 , π1,2,3 If proof is accepting: βπ β 3 obtain ππ = ππ π‘π π1 , π2 , π3 β 0,1 π Everywhere Local Consistency π1,2,3 π1 , π2 , π3 β‘ π πΎ 1 π1 , πΎ 2 π2 , πΎ 3 π3 , π1 π1 , π2 π2 , π3 π3 ππ,π,π π‘1 , π‘2 , π‘3 = π πΎ 1 π‘1 , πΎ 2 π‘2 , πΎ 3 π‘3 , π1 π‘1 , π2 π‘2 , π3 π‘3 Follows from missing tests 0 = ππ,π,π π‘1 , π‘2 , π‘3 = π π1 , π2 , π3 , π1 , π2 , π3 Achieving Local Consistency Everywhere Verifier Random Everywhere accepts Local consistency Localconsistency In standard PCPs: Follows from low-degree test and SchwartzβZippel lemma In our protocol: Follows from security of curve encryptions THANKS!
© Copyright 2026 Paperzz