200 likes | 348 Views
Transitive Signatures based on Factoring and RSA. Mihir Bellare (University of California, San Diego, USA) Gregory Neven (Katholieke Universiteit Leuven, Belgium). Standard digital signatures. SKG. ( spk , ssk ). 1 k. SSign. ssk. σ. M. SVf. spk. accept / reject. M. σ’. σ 2,3.
E N D
Transitive Signatures based on Factoring and RSA Mihir Bellare(University of California, San Diego, USA) Gregory Neven(Katholieke Universiteit Leuven, Belgium)
Standard digital signatures SKG (spk,ssk) 1k SSign ssk σ M SVf spk accept /reject M σ’
σ2,3 σ4,5 σ1,2 4,5 2,3 1,2 2 σ1,2 σ2,3 1 3 4 5 σ4,5 Transitive signatures [MR02] • Message is pair of nodes i,j TKG (tpk,tsk) 1k • Signing i,j = creating and authenticating edge {i,j} TSign tsk σi,j • An authenticated graph grows with time i,j TVf tpk accept /reject i,j σ’i,j
1,2,3 σ1,3 σ1,3 σ1,2 σ2,3 Transitive signatures [MR02] • Additional composition algorithm TKG (tpk,tsk) 1k • Authenticated graph is transitive closure of directly signed edges TSign tsk σi,j i,j TVf tpk accept /reject i,j σ’i,j 2 Comp σ1,2 σ2,3 tpk i,j,k σi,k 1 3 σi,j σj,k 4 5 σ4,5
tpk 1,2 F 2 σ1,2 σ2,3 σ1,2 1,2 ║ 2,3 ║ 4,5 2,3 TSigntsk(·,·) 1 3 TSigntsk(·,·) σ1,3 σ2,3 σ1,2,σ2,3,σ4,5 σ1,4 4 5 4,5 σ4,5 σ4,5 {1,4}, σ1,4 Security of transitive signatures • Standard security definition of [GMR] doesn’t apply: composition allows forgery to some extent • New security goal [MR02]: • computationally infeasible to forge signatures not in transitive closure of the edges signed directly by the signer • even under “chosen-edge” attack
Why transitive signatures? Applications? Micali and Rivest suggest • military chain-of-command (directed) • administrative domains (undirected) Compelling application yet to be found But a cool concept! J
1,y1 2,y2 2,y2 1,y1 3,y3 2,y2 1,y1 x2 ,y2 2 σ1,2 ← Z* R Signer assigns to each node i: N 1 3 x1 ,y1 x3 ,y3 • secret label xi i,yi • public label yi ← xie mod N • node certificate Verification of ( , , δ1,2): To sign edge {1,2}: • check node certificates • edge label δ1,2 ← x1·x2-1 mod N • check δ1,2 = y1·y2-1 mod N • signature σ1,2 = ( , , δ1,2) e RSATS-1: RSA based scheme [MR02] Assume standard signature scheme with • key pair (spk,ssk) • message M signed under ssk M tpk = (spk, N, e) tsk = ssk
2,y2 3,y3 1,y1 2,y2 3,y3 1,y1 σ1,3 σ1,3 = ( , , δ1,3) where δ1,3 = δ1,2·δ2,3 modN = (x1·x2-1)(x2·x3-1) modN =x1·x3-1 modN xi are kept in signer’s state Composition in RSATS-1 To compose signatures σ1,2 and σ2,3: σ1,2 = ( , , δ1,2) where δ1,2 =x1·x2-1 modN x2 ,y2 2,y2 2 σ1,2 σ2,3 σ2,3 = ( , , δ2,3) where δ2,3 =x2·x3-1 modN 1 3 x1 ,y1 x3 ,y3 1,y1 3,y3
Non-adaptive security of RSATS-1 RSATS-1 can be proven transitively secure against forgery under non-adaptive chosen-edge attack if • RSA is one-way • underlying standard signature scheme is secure under chosen-message attack Is RSATS-1 secure under adaptive attack? • Neither proof nor attack known • Might rely on stronger properties of RSA than one-wayness • We consider security under one-more inversion [BNPS01]
A is successful iff • xie = yi mod N for i=1..m • n < m Assumption: this problem is hard [BNPS01] z1 Used before • by [BNPS01] to prove security of Chaum’s blind signatures • by [BP02] to prove security of GQ identification scheme z1d mod N … zn znd mod N x1,…,xm RSA under one-more inversion N,e A Chall y1 Z* R yi … N ym RSA-1N,e(·)
Adaptive security of RSATS-1 Theorem: RSATS-1 is transitively secure against forgery under adaptivechosen-message attack if • the one-more RSA-inversion problem is hard • the underlying standard signature scheme is secure under chosen-message attack.
y2 y5 2 5 σ2,3 σ5,6 yi σ1,2 σ1,3 3 6 {1,2} {1,3} {2,3} y3 1 y6 y1y2-1 y1 y1 σ1,2 σ1,4 σ4,6 σ2,3 σ1,3 δ1,2 x1 4 σ1,4 y4 x1,…,x6 n1 nodes n2 nodes If A would know x3: (remember δi,j=xi·xj-1) n1-1 queries n2-1 queries x2 ← δ2,3·x3 y2y3-1 x1 ← δ1,2·x2 (n1-1)+(n2-1)+1 = n1+n2-1 queries δ2,3 < n1+n2 decrypted challenges Proof idea for RSATS-1 N,e (spk,N,e) Chall A F RSA-1
1,y1 1,y1 2,y2 3,y3 2,y2 2,y2 3,y3 1,y1 1,y1 x2 ,y2 Signer assigns to each node i: 2 σ1,2 σ2,3 • secret label xi 1 3 • public label yi ← xi2 mod N σ1,3 x1 ,y1 x3 ,y3 • node certificate ← Z* R N Signature σ1,2 = ( , , δ1,2) with δ1,2 = x1·x2-1 mod N i,yi Verification of σ1,2 : • check signatures on , • check δ1,2 = y1·y2-1 mod N 2 Composition of σ1,2 and σ2,3: σ1,3 = ( , , δ1,3) with δ1,3 = δ1,2·δ2,3 modN FBTS-1: Factoring based scheme tpk = (spk, N); tsk = ssk
Theorem:FBTS-1 is transitively secure against forgery under adaptive chosen-message attack if factoring N is hard the underlying standard signature scheme is secure under chosen-message attack. Security of FBTS-1 Proof idea: • with probability 1/2, forgery gives second square root • signatures might leak information about known root → information-theoretic lemma needed
x2 ,y2 • chooses secret label xi σ1,2 σ2,3 • computes public label yi = f(xi) • creates node certificate σ1,3 x1 ,y1 x3 ,y3 Signature σ1,2 = ( , , δ1,2) where δ1,2 = g(x1,x2) Composition of σ1,2 and σ2,3: σ1,3 = ( , , δ1,3) where δ1,3 = h(δ1,2,δ2,3) 2,y2 2,y2 Scheme f(xi) g(xi,xj) h(δi,j,δj,k) 1,y1 1,y1 1,y1 i,yi 3,y3 3,y3 RSATS-1 xie mod N xi·xj-1 mod N δi,j·δj,k mod N FBTS-1 xi2 mod N xi·xj-1 mod N δi,j·δj,k mod N Node certification paradigm For each node i, the signer: 2 1 3
For each node i, signer lets: • public label yi ← Htpk(i) • secret label xi← “inversion” of yi(using trapdoor information in tsk) y2=Htpk(2) ,x2 RSATS-1 and FBTS-1, but not MRTS 2 σ1,2 σ2,3 Signature σ1,2 = δ1,2 where δ1,2 = f(x1,x2) 1 3 σ1,3 y1=Htpk(1) ,x1 y3=Htpk(3) ,x3 Composition of σ1,2 and σ2,3: σ1,3 = δ1,3 where δ1,3 = g(δ1,2, δ2,3) Eliminating node certificates Let Htpk be a public hash function
FBTS-2: Modifications needed because public labels have to be squares mod N Theorem:FBTS-2 is transitively secure against forgery under adaptive chosen-message attack if • factoring N is hard • HN: {0,1}*→ZN[+1] is a random oracle. * RSATS-2 and FBTS-2 RSATS-2:Straightforward application of this idea to RSATS-1 Theorem:RSATS-2 is transitively secure against forgery under adaptive chosen-message attack if • the one-more RSA-inversion problem is hard • HN: {0,1}*→ZN is a random oracle. *
Z* N MRTS Discrete logarithmsStandard signatures Yes 2 stand. sigs2 points in G2 points in Zq RSATS-1 One-wayness of RSAStandard signatures No 2 stand. sigs3 points in Previously known schemes Scheme Security assumption Ad.? Signature size Trivial Standard signatures Yes O(path length)
Z* Z* Z* N N N RO? No No No No FBTS-1 FactoringStandard signatures Yes No 2 stand sigs3 points in RSATS-2 One-more RSA Yes Yes 1 point in FBTS-2 Factoring Yes Yes 1 point in Scheme contributions Scheme Security assumption Ad.? Signature size Trivial Standard signatures Yes O(path length) MRTS Discrete logarithmsStandard signatures Yes 2 stand. sigs2 points in G2 points in Zq RSATS-1 One-wayness of RSAStandard sigs No 2 stand. sigs3 points in Z* N RSATS-1 One-more RSAStandard signatures Yes 2 stand sigs3 points in Z* N