Analysis of Security Protocols (III) John C. Mitchell Stanford University
Analyzing Security Protocols l Non-formal approaches (can be useful, but no tools…) Some crypto-based proofs [Bellare, Rogaway] l BAN and related logics Axiomatic semantics of protocol steps H Methods based on operational semantics Intruder model derived from Dolev-Yao Protocol gives rise to set of traces Perfect encryption Possible to include known algebraic properties
Example projects and tools l Prove protocol correct Paulson’s “Inductive method”, others in HOL, PVS, etc. Bolignano -- Abstraction methods MITRE -- Strand spaces Process calculus approach: Abadi-Gordon spi-calculus l Search using symbolic representation of states Meadows: NRL Analyzer, Millen: Interrogator l Exhaustive finite-state analysis FDR, based on CSP [Lowe, Roscoe, Schneider, …] Mur -- specialized input language Clarke et al. -- state search with axiomatic intruder model
Explicit Intruder Method Intruder Model Analysis Tool Formal Protocol Informal Protocol Description Gee whiz. Looks OK to me.
A notation for inf-state systems l Define protocol, intruder in minimal framework l Disadvantage: need to introduce new notation Formal Logic ( ) Process Calculus Finite Automata Proof search (Horn clause) Multiset rewriting
Protocol Notation l Non-deterministic infinite-state systems l Facts F ::= P(t 1, …, t n ) t ::= x | c | f(t 1, …, t n ) l States { F 1, …, F n } Multiset of facts Includes network messages, private state Intruder will see messages, not private state Multi-sorted first-order atomic formulas
State Transitions l Transition F 1, …, F k x 1 … x m. G 1, …, G n l What this means If F 1, …, F k in state , then a next state ’ has Facts F 1, …, F k removed G 1, …, G n added, with x 1 … x m replaced by new symbols Other facts in state carry over to ’ Free variables in rule universally quantified Pattern matching in F 1, …, F k can invert functions
Finite-State Example Predicates: State, Input Function: Constants: q 0, q 1, q 2, q 3, a, b, nil Transitions: State(q 0 ), Input(a x) State(q 1 ), Input(x) State(q 0 ), Input(b x) State(q 2 ), Input(x)... q0q0 q1q1 q3q3 q2q2 b a a a b b b a b
Existential Quantification l Natural-deduction proof rule [y/x] ( elim) x. l Summary: for proof from x. , choose new symbol and proceed from [y/x] y not free in any other hypothesis
Infinite-State Example Predicates: State, Input, Color Function: Constants: q 0, a, b, nil, red, blue Transitions: State(q), Input(a x), Color(q,red) q’. State(q’), Input(x), Color(q’,blue), Color(q,red)... Input a: change color Input b: same color Need to preserve facts explicitly q0q0 a a a a a b b bb
Turing Machine l Predicates Current(state,cell) -- current state, tape pos. Contents(cell, symbol) -- contents of tape cell Adjacent(cell, cell) -- keep cells in order l Constants q 0, q 1, q 2, … -- finite set of states c 0, c eot -- initial tape cells “0”, “1”, “b” -- tape symbols
Turing Machine (II) l Transitions Adjacent(c 0, c eot ) Adjacent(c, c eot ) c’. Adjacent(c,c’), Adjacent(c’,c eot ) Current(q i,c), Contents(c,“0”), Adjacent(c,c’) Current(q k,c’), Contents(c,“1”), Adjacent(c,c’) Current(q i,c), Contents(c,“1”), Adjacent(c’,c) Current(q k,c’),Contents(c,“0”), Adjacent(c,c’) infinite linear tape sample move right sample move left...ceotc...c’eot
Simplified Needham-Schroeder l Predicates A i, B i, N i -- Alice, Bob, Network in state i l Transitions x. A 1 (x) A 1 (x) N 1 (x), A 2 (x) N 1 (x) y. B 1 (x,y) B 1 (x,y) N 2 (x,y), B 2 (x,y) A 2 (x), N 2 (x,y) A 3 (x,y) A 3 (x,y) N 3 (y), A 4 (x,y) B 2 (x,y), N 3 (y) B 3 (x,y) picture next slide A B: {n a, A} Kb B A: {n a, n b } Ka A B: {n b } Kb l Authentication A 4 (x,y) B 3 (x,y’) y=y’
Sample Trace A B: {n a, A} Kb B A: {n a, n b } Ka A B: {n b } Kb A 2 (n a ) A 1 (n a ) A 2 (n a ) A 3 (n a, n b ) A 4 (n a, n b ) B 2 (n a, n b ) B 1 (n a, n b ) B 2 (n a, n b ) B 3 (n a, n b ) B 2 (n a, n b ) N 1 (n a ) N 2 (n a, n b ) N3( nb)N3( nb) x. A 1 (x) A 1 (x) A 2 (x), N 1 (x) N 1 (x) y. B 1 (x,y) B 1 (x,y) N 2 (x,y), B 2 (x,y) A 2 (x), N 2 (x,y) A 3 (x,y) A 3 (x,y) N 3 (y), A 4 (x,y) B 2 (x,y), N 3 (y) B 3 (x,y)
Common Intruder Model l Derived from Dolev-Yao model [1989] Adversary is nondeterministic process Adversary can Block network traffic Read any message, decompose into parts Decrypt if key is known to adversary Insert new message from data it has observed Adversary cannot Gain partial knowledge Guess part of a key Perform statistical tests, …
Formalize Intruder Model l Intercept and remember messages N 1 (x) M(x) N 2 (x,y) M(x), M(y) N 3 (x) M(x) l Send messages from “known” data M(x) N 1 (x), M(x) M(x), M(y) N 2 (x,y), M(x), M(y) M(x) N 3 (x), M(x) l Generate new data as needed x. M(x) Highly nondeterministic, same for any protocol
Attack on Simplified Protocol A 2 (n a ) A 1 (n a ) A 2 (n a ) B 1 (n a ’, n b ) N 1 (n a ) x. A 1 (x) A 1 (x) A 2 (x), N 1 (x) N 1 (x) M(x) x. M(x) M(x) N 1 (x), M(x) N 1 (x) y. B 1 (x,y) M(n a ) M(n a ), M(n a ’) N 1 (n a ’ ) A 2 (n a ) M(n a ), M(n a ’) A 2 (n a ) M(n a ), M(n a ’) Continue “man-in-the-middle” to violate specification
Modeling Perfect Encryption l Encryption functions and keys For public-key encryption two key sorts: e_key, d_key predicate Key_pair(e_key, d_key) Functions enc : e_key msg -> msg dec : d_key msg -> msg (implicit in pattern-matching) l Properties of this model Encrypt, decrypt only with appropriate keys Only produce enc(key, msg) from key and msg (!!!) This is not true for some encryption functions
Steps in public-key protocol l Bob generates key pair and publishes e_key u. d_key v. Bob 1 (u,v) Bob 1 (u,v) N Announce (u ), Bob 2 (u,v) l Alice sends encrypted message to Bob Alice 1 (e,d,x), N Announce (e’) Alice 2 (e,d,x,e’) Alice 2 (e,d,x,e’) N 1 (enc(e’, x,e )), Alice 3 (u,v,x,w) l Bob decrypts Bob 1 (u,v), N 1 (enc(u, x,y )) z. Bob 1 (u,v,x,y,z)
Intruder Encryption Capabilities l Intruder can encrypt with encryption key M e (k), M data (x) N i (enc(k,x)), M e (k), M data (x) l Intruder can decrypt with decryption key N j (enc(k,x)),Key_pair(k,k’), M d (k’), M data (x),... l Add to previous intruder model Assumes sorts data, e_key, d_key with typed predicates M data (data), M e (e_key), M d (d_key)
Intruder: power and limitations l Can find some attacks Needham-Schroeder by exhaustive search l Other attacks are outside model Interaction between protocol and encryption l Some protocols cannot be modeled Probabilistic protocols Steps that require specific property of encryption l Possible to prove erroneous protocol correct Requires property that crypto does not provide
Optimize Protocol + Intruder l Adversary receives all messages; no net Replace Alice i (x,y) N j (x), Alice k (x,y) N j (x) M(x) M(z) N j (z), M(z) N j (x), Bob i (w) Bob j (w,y) By Alice i (x,y) M(x), Alice k (x,y) M(z), Bob i (w) Bob j (w,y) Alice’s message can go to Bob or M. M can replay or send different msg All messages go directly to M. M can forward or send different msg
Additional Optimizations l Intruder can simulate honest participants If additional independent sessions are useful for attack, then intruder can simulate these sessions Therefore -- suffices to consider single initiator, single responder, and intruder (for this protocol). l For decidability, bound on intruder [Lowe] Suffices to bound the number of new nonces Analyze...
Analysis of Protocol+Intruder l Prove properties of protocols Unbounded # of participants, message space Prove that system satisfies specification Paulson, etc: prove invariant holds at all reachable states Spi-calculus: prove protocol equivalent ideal protocol l Symbolic search with pruning Search backward from error Prune search by proving forward invariants l Exhaustive finite-state methods Approximate infinite-state system by finite one Search all states, perhaps with optimizations
Example description languages l First- or Higher-order Logic Define set of traces, prove protocol correct l Horn-clause Logic x… (A 1 A 2 … B) Symbolic search methods l Process calculus FDR model checker based on CSP Spi-calculus proof methods based on pi-calculus l Additional formalisms CAPSL protocol description language [Millen] Mur language for finite-state systems
Paulson’s Inductive Method l Define set TR of traces of protocol+intruder Similar to traces in unifying formalism Transition F 1, …, F k x 1 … x m. G 1, …, G n gives one way of extending trace l Auxiliary functions mapping traces to sets Analz(trace) = data visible to intruder Synth(trace) = messages intruder can synthesize l Definitions and proofs use induction Similar inductive arguments for many protocols
Symbolic Search Methods l Examples: NRL Protocol Analyzer, Interrogator l Main idea Write protocol as set of Horn clauses Transition F 1, …, F k x 1 … x m. G 1, …, G n can be Skolemized and translated to Prolog clauses Search back from possible error for contradiction This is usual Prolog refutation procedure l Important pruning technique Prove invariants by forward reasoning Use these to avoid searching unreachable states
Process Calculus Description l Protocol defined by set of processes Each process gives one step of one principal Can derive by translation from unifying notation F 1, …, F k x 1 … x m. G 1, …, G n is one process Replace predicates by port names Replace pattern-matching by explicit destructuring In pi-calculus, use in place of Example B 1 (x,y) N 2 (x,y), B 2 (x,y) b 1 (p). let x=fst(p) and y=snd(p) in n 2 x,y | b 2 x,y end
Spi-Calculus [AG97,...] l Write protocol in process calculus l Express security using observ. equivalence Standard relation from programming language theory P Q iff for all contexts C[ ], same observations about C[P] and C[Q] Context (environment) represents adversary l Use proof rules for to prove security Protocol is secure if no adversary can distinguish it from an idealized version of the protocol
Finite-state methods l Two sources of infinite behavior Many instances of participants, multiple runs Message space or data space may be infinite l Finite approximation Transitions: F 1, …, F k x 1 … x m. G 1, …, G n choose fixed number of Skolem constants Terms: restrict repeated functions f(f(f(f(x)))) l Can express finite-state protocol + intruder in CSP : FDR-based model checking projects Other notations: Mur project, Clarke et al.,...
Security Protocols in Mur l Standard “benchmark” protocols Needham-Schroeder, TMN, … Kerberos l Study of Secure Sockets Layer (SSL) Versions 2.0 and 3.0 of handshake protocol Include protocol resumption l Discovered all known or suspected attacks l Recent work on tool optimization [Shmatikov, Stern,...]
Malleability [Dolev,Dwork,Naor] l Idealized assumption If intruder produces Network(enc(k,x)) then either Network(enc(k,x)) M (enc(k,x)) (replay) M(k), M(x) M (enc(k,x)) (knows parts) l Not true for RSA encrypt(k,msg) = msg k mod N property encr(x*y) = encr(x) * encr(y) l Model Network(enc(k,x)) M (…)... Network (enc(k,c*x)) Can send encrypted message without “knowing” message Finite state ?
Authentication and Secrecy for Handshake Protocols l How many protocols are there to verify? Average length 7 steps Data fields per message 5 fields Distinct ways to fill a field 50 entries Number of possible combinations 1750 protocols l Research directions Get the monkeys and typewriters going Easier description and specification, faster tools Improved analysis of timestamps,... Interaction between protocol and crypto primitives