MIT 6.875/6.5620/18.

425

Foundations of Cryptography
Lecture 1

Course website: https://mit6875.github.io/

 Course Staff
TAs

Instructor

Surya Mathialagan(smathi@mit)

Vinod Vaikuntanathan
(vinodv@mit)

Tina Zhang (tinaz@mit)

Course Website
https://mit6875.github.io/
 Crypto Cryptocurrencies
Blockchains/
Cryptocurrencies
“Trustworthy”
6.875 is not about machine learning

6.875 is about foundations:

Zero-knowledge
Proofs Public-key Encryption
Digital Signatures

Threshold
Homomorphic Cryptography Pseudorandomness
Encryption
 The Intellectual Origins
“Communication Theory of Secrecy Systems” (1945)
preceded
“A Mathematical Theory of Communication” (1948)
which founded Information Theory

Claude E. Shannon

Cryptanalysis of the Enigma Machine (~1938-39)

“On Computable Numbers, with an Application to

the Entscheidungsproblem” (1936)

which gave birth to Computer Science

Alan M. Turing
 Modern Cryptography:
Practice to Theory and Back
De
fini
tion
s

Problems Tools
Security, Math and
Privacy, CRYPTO Theoretical CS
Integrity

Solutions Ideas

Encryption Interactive Proofs

Digital Signatures Probabilistically checkable Proofs
Pseudorandom Functions Locally decodable Codes
… …
 6.875/6.5620 Themes

1. The Omnipresent, Worst-case, Adversary.

Central idea. model the adversary: what they know, what

they can do, and what their goals are.

Definitions will be our friend.

If you cannot define something, you cannot achieve it.

A key takeaway from 6.875:

Cryptographic (or, adversarial) thinking.
 6.875/6.5620 Themes
2. Computational Hardness will be our enabler.
(starting lecture 2)
Central theme: the cryptographic leash. Use
computational hardness to “tame” the adversary.

A classical source of hard problems: number theory.

“Both Gauss and lesser mathematicians may be justified in rejoicing that
there is one such science [number theory] at any rate, whose very
remoteness from ordinary human activities should keep it gentle and clean”
[G. H. Hardy, “A Mathematician’s Apology”]

More recently: geometry, coding theory, combinatorics.

Cryptography is the science of useful hardness.

 6.875 Themes
3. Security Proofs via Reductions.

“If there is an (efficient) adversary that breaks scheme A

w.r.t. definition D, then there is an (efficient) adversary
that factors large numbers.”

“Science wins either way”

Our reductions will be probabilistic and significantly more

involved than the NP-hardness reductions in, say, 6.045.
 6.875 Topics
 Pseudorandomness
 Secret-key Encryption and Authentication
 Public-key Encryption and Digital Signatures
 Cryptographic Hashing
 Zero-knowledge Proofs
 Secure Multiparty Computation
 Private Information Retrieval
 Homomorphic Encryption
 Advanced topics:
Threshold Cryptography, Differential Privacy,…
 Administrivia
o Course website, the central point of reference.
https://mit6875.github.io

Piazza for questions, Gradescope for psets.

Piazza code: publickey
Gradescope code: V5DDBG
 Administrivia
o Course website, the central point of reference.
https://mit6875.github.io

Piazza for questions, Gradescope for psets.

o Homework (95%): 6 psets, we will count your best 5.

6.875 is on https://psetpartners.mit.edu
 Administrivia
o Course website, the central point of reference.
https://mit6875.github.io

Piazza for questions, Gradescope for psets.

o Homework (95%): 6 psets, we will count your best 5.

o Class Participation (5%): Lecture, Recitations, Piazza.

o Prerequisites: Algorithms, Probability & Discrete Math, but

most of all, “mathematical maturity”.
o (Optional) special recitations: 1. probability (this Friday), 2.
basic complexity theory, 3. number theory.
 Secure Communication
m

Bob
Alice

Eavesdropper “Eve”

Alice wants to send a message m to Bob without revealing it to Eve.

 Secure Communication
m

Key k Key k

Eavesdropper “Eve”

SETUP: Alice and Bob meet beforehand to agree on a secret key k.

Key Notion: Secret-key Encryption
(or Symmetric-key Encryption)

Ciphertext
𝑚←De c(𝑘 ,𝑐 )

Key k Key k

Three (possibly probabilistic) polynomial-time algorithms:

o Key Generation Algorithm Gen:
Has to be probabilistic
o Encryption Algorithm Enc:

o Decryption Algorithm Dec:

 The Worst-case Adversary
 An arbitrary computationally unbounded algorithm EVE.*

 Knows Alice and Bob’s algorithms and but does not

know the key nor their internal randomness.
(Kerckhoff’s principle or Shannon’s maxim)

 Can see the ciphertexts going through the channel

(but cannot modify them… we will come to that later)

Security Definition: What is she trying to learn?

Shannon’s Perfect Secrecy Definition
Message space (probability distribution)

𝑐 ← Enc(𝑘 , 𝑚)
Ciphertext space
Key k Key k

Key space

IDEA: A-posteriori = A-priori

, ,

Pr [ ℳ=𝑚∨𝐸𝑛𝑐 ( 𝒦 , ℳ ) =𝑐 ] =Pr [ ℳ=𝑚 ]

A-posteriori A-priori
Perfect Indistinguishability Definition
Perfect indistinguishability: a Turing test

,
Pr ⁡[ 𝐸 ( K , 𝑚 ) =𝑐]=Pr ⁡[𝐸 ( K ,𝑚′ )=𝑐 ]

World 0: World 1:
kK k
𝑐= 𝐸 ( 𝑘 , 𝑚 ) 𝑐 ′ = 𝐸 ( 𝑘, 𝑚 ′ )

is a distinguisher (that gets c and tries to guess

which world she’s in)
The Two Definitions are Equivalent

THEOREM: An encryption scheme satisfies

perfect secrecy IFF it satisfies perfect
indistinguishability.

PROOF: Simple use of Bayes’ Theorem.

 1. Indistinguishability Secrecy
WE KNOW (IND): , ,

Pr [ 𝐸𝑛𝑐 ( 𝒦 , 𝑚 )=𝑐 ] =Pr [ 𝐸𝑛𝑐 𝒦 ,𝑚 )¿=𝑐

( 𝜶]
′

WE WANT (SEC): , ,

Pr [ ℳ=𝑚∨𝐸𝑛𝑐 ( 𝒦 , ℳ ) =𝑐 ] =Pr [ ℳ=𝑚 ]

Key Observation: .
Proof: definition of conditional probability.

.
 1. Indistinguishability Secrecy
WE KNOW (IND): , ,

Pr [ 𝐸𝑛𝑐 ( 𝒦 , 𝑚 )=𝑐 ] =Pr [ 𝐸𝑛𝑐 𝒦 ,𝑚 )¿=𝑐

( 𝜶]
′

WE WANT (SEC): , ,

Pr [ ℳ=𝑚∨𝐸𝑛𝑐 ( 𝒦 , ℳ ) =𝑐 ] =Pr [ ℳ=𝑚 ]

Proof:
= (Bayes)

= (key obs.)
 2. Secrecy Indistinguishability
WE KNOW (SEC): , ,

Pr [ ℳ=𝑚∨𝐸𝑛𝑐 ( 𝒦 , ℳ ) =𝑐 ] =Pr [ ℳ=𝑚 ]

WE WANT (IND): , ,

Pr [ 𝐸𝑛𝑐 ( 𝒦 , 𝑚 )=𝑐 ] =Pr [ 𝐸𝑛𝑐 𝒦 ,𝑚 ) =𝑐 ]

( ′

Proof: Pr [ 𝐸𝑛𝑐 ( 𝒦 , 𝑚 )=𝑐 ]=Pr ⁡[𝐸𝑛𝑐 ( 𝒦 , ℳ ) =𝑐∨ℳ=𝑚 ]

= (Bayes)

] (because of SEC)

¿ Pr [ 𝐸𝑛𝑐 ( 𝒦 ,𝑚 ′ )=𝑐 ] (symmetry)

 Perfect Secrecy is Achievable
The One-time Pad Construction:
Choose an -bit string k at random, i.e.
, where M is an n-bit message: Output
Output

bitwise exclusive OR (or XOR)

0 0= 1=0
0 1= 0=1
a b = a + b (mod 2)
 Perfect Secrecy is Achievable
The One-time Pad Construction:
Choose an -bit string k at random, i.e.
, where M is an n-bit message: Output
Output

Correctness:
 Perfect Secrecy is Achievable
The One-time Pad Construction:
Choose an -bit string k at random, i.e.
, where M is an n-bit message: Output
Output

Claim: One-time Pad achieves Perfect Indistinguishability (and

therefore perfect secrecy).
Proof: For any
 Perfect Secrecy is Achievable
The One-time Pad Construction:
Choose an -bit string k at random, i.e.
, where M is an n-bit message: Output
Output

Claim: One-time Pad achieves Perfect Indistinguishability (and

therefore perfect secrecy).
Proof: For any

¿Pr [ K =𝑐⨁𝑚′ ]= Pr [ 𝑚′⨁K=𝑐 ]= Pr [ Enc ( K,𝑚′ ) =𝑐 ]

 Perfect Secrecy is Achievable
The One-time Pad Construction:
Choose an -bit string k at random, i.e.
, where M is an n-bit message: Output
Output

Claim: One-time Pad achieves Perfect Indistinguishability (and

therefore perfect secrecy).
Proof: For any m
So, .
QED.
 Reusing a One-time Pad?
m0

𝑐 0=𝑚0 ⊕ 𝑘
Key k Key k

A week later:

m1
c

Is this still perfectly secret?

Super-secure Whisper room
 Reusing a One-time Pad?
Claim: Two-time Pad does not achieve Perfect
Indistinguishability (and therefore not perfect secrecy).

Proof: Perfect indistinguishability requires that for all pairs

P r ¿
¿ P r ¿
 Reusing a One-time Pad?
Claim: One-time Pad does not achieve Perfect
Indistinguishability (and therefore not perfect secrecy).

Proof: We want to pick s.t.

P r ¿
≠ P r ¿
′ ′
𝐏𝐢𝐜𝐤𝒎𝟎=𝒎𝟏=𝒎,𝒎𝟎 ≠𝒎𝟏 𝐚𝐧𝐝𝒄𝟎=𝒄𝟏=𝒄.
P r ¿
¿ P r [ En 𝑐 ( 𝑘 , 𝑚 ) = 𝑐 ] =1 / 2𝑛
 Reusing a One-time Pad?
Claim: One-time Pad does not achieve Perfect
Indistinguishability (and therefore not perfect secrecy).

Proof: We want to pick s.t.

P r ¿
≠ P r ¿
′ ′
𝐏𝐢𝐜𝐤𝒎𝟎=𝒎𝟏=𝒎,𝒎𝟎 ≠𝒎𝟏 𝐚𝐧𝐝𝒄𝟎=𝒄𝟏=𝒄.
P r ¿
P r ¿
 Perfect Secrecy has its Price
THEOREM: For any perfectly secure encryption scheme,

PROOF (by picture): Assume for contradiction that

ℳ 𝒞
𝑘 Pick any
𝑚 𝑐
𝑘′
𝑚′ Look at the set of
𝑘′ ′ possible msgs
𝑚′ ′ 𝑘′ ′ ′ ( etc.)
≤|𝓚 |<¿ 𝓜∨¿
~
𝒎 Distinct keys!
 Perfect Secrecy has its Price
THEOREM: For any perfectly secure encryption scheme,

PROOF (by picture): Assume for contradiction that

ℳ 𝒞
𝑘
𝑚 𝑐
𝑘′ Pr [ 𝐸𝑛𝑐 ( 𝒦 , 𝑚 )=𝑐 ] >0
𝑚′ 𝑘′ ′ Pr [ 𝐸𝑛𝑐 ( 𝒦 ,~
𝑚 )=𝑐 ]=0
𝑚′ ′ 𝑸𝑬𝑫.
≤|𝓚 |<¿ 𝓜∨¿
~
𝒎
 So, what are we to do?
RELAX the definition:
EVE is an arbitrary computationally bounded algorithm.

+ number theory/geometry/combinatorics

the promised crypto land

Fully homomorphic encryption

ZK proofs
Public-key encryption

Pseudorandomness
 To Summarize…
• Secure Communication: a quintessential problem in
cryptography.

• We saw two equivalent definitions of security:

Shannon’s perfect indistinguishability and perfect secrecy

• One-time pad achieves perfect secrecy.

• A Serious Limitation: Any perfectly secure encryption scheme

needs keys that are at least as long as the messages.

• Next Lecture: Overcoming the limitation with Computationally

Bounded Adversaries.