Skip to content

Latest commit

 

History

History
228 lines (173 loc) · 15.3 KB

File metadata and controls

228 lines (173 loc) · 15.3 KB

Cryptographic Protocols Modeling & Models of Communication (and Intruder)

[TOC]

Res

Related Topics

Mathematical Modeling & AbstractionModels of Computation & Abstract Machines

Cryptographic Protocols Modeling & Verification

AnB (Alice and Bob) Notation & AnBx Languages

Other Resources

Intro

🔗 https://en.wikipedia.org/wiki/Models_of_communication

Models of communication simplify or represent the process of communication. Most communication models try to describe both verbal and non-verbal communication and often understand it as an exchange of messages. Their function is to give a compact overview of the complex process of communication. This helps researchers formulate hypotheses, apply communication-related concepts to real-world cases, and test predictions. Despite their usefulness, many models are criticized based on the claim that they are too simple because they leave out essential aspects. The components and their interactions are usually presented in the form of a diagram. Some basic components and interactions reappear in many of the models. They include the idea that a sender encodes information in the form of a message and sends it to a receiver through a channel. The receiver needs to decode the message to understand the initial idea and provides some form of feedback. In both cases, noise may interfere and distort the message.

Models of communication are classified depending on their intended applications and on how they conceptualize the process. General models apply to all forms of communication while specialized models restrict themselves to specific forms, like mass communication. Linear transmission models understand communication as a one-way process in which a sender transmits an idea to a receiver. Interaction models include a feedback loop through which the receiver responds after getting the message. Transaction models see sending and responding as simultaneous activities. They hold that meaning is created in this process and does not exist prior to it. Constitutive and constructionist models stress that communication is a basic phenomenon responsible for how people understand and experience realityInterpersonal models describe communicative exchanges with other people. They contrast with intrapersonal models, which discuss communication with oneself. Models of non-human communication describe communication among other species. Further types include encoding-decoding models, hypodermic models, and relational models.

The problem of communication was already discussed in Ancient Greece but the field of communication studies only developed into a separate research discipline in the middle of the 20th century. All early models were linear transmission models, like Lasswell's model, the Shannon–Weaver model, Gerbner's model, and Berlo's model. For many purposes, they were later replaced by interaction models, like Schramm's model. Beginning in the 1970s, transactional models of communication, like Barnlund's model, were proposed to overcome the limitations of interaction models. They constitute the origin of further developments in the form of constitutive models.

Why Models of Communication & Cryptographic Protocols

🤖 GPT-5.2 https://chatgpt.com/share/698a3655-f540-800f-939f-c8ce2c90027f

One-sentence summary: Cryptographic models are needed to precisely define security, reason about powerful attackers, find subtle protocol flaws, enable proofs and automation, and make assumptions explicit—something testing and intuition can never guarantee.


We need cryptographic models because security is not something we can “test” exhaustively—it must be reasoned about. Models give us a precise way to define attackers, assumptions, and guarantees, so we can prove (or disprove) security claims.

Here’s the intuition, then the formal reasons.

Intuition: why models are necessary -- Saying “this protocol is secure” without a model is meaningless.

  • Secure against whom?
  • Secure under what assumptions?
  • Secure in which environment? Cryptographic models answer these questions.

Formal reasons:

==To define what security means== Security is not absolute. Examples:

  • “The message is secret”
  • “The protocol authenticates the server”
  • “Keys cannot be stolen” These statements are vague unless we specify:
  • What the attacker can do
  • What it means to “break” security

Models provide formal definitions For example:

  • In the Dolev–Yao model, secrecy means:
    • The attacker cannot derive the message term.
  • In the computational model, secrecy means:
    • No PPT adversary can distinguish the message from random. Without a model, there is no precise notion of security.

==To reason about extremely powerful attackers== Real attackers:

  • Control networks
  • Replay messages
  • Run protocols concurrently
  • Adapt based on observations We cannot enumerate all possible attacks.

Models let us assume a maximal attacker

  • Dolev–Yao: attacker controls the entire network
  • Computational: attacker is any PPT algorithm

If a protocol is secure against these attackers, it is secure against weaker real-world attackers.

==To find subtle protocol flaws before deployment== Many famous protocol failures were logic errors, not broken crypto:

  • Needham–Schroeder (authentication flaw)
  • TLS renegotiation attack
  • Kerberos replay issues These flaws:
  • Passed testing
  • Used strong cryptography
  • Failed under formal analysis

Models expose:

  • Replay attacks
  • Man-in-the-middle attacks
  • Confusion of identities
  • Key reuse mistakes Testing alone cannot catch these.

==To enable formal proofs and automation== Models make security machine-checkable. Thanks to models, we can:

  • Automatically verify protocols
  • Get counterexamples (attacks)
  • Prove properties hold for all executions Tools rely on models:
  • ProVerif → symbolic (Dolev–Yao–style)
  • Tamarin → symbolic + algebraic
  • CryptoVerif → computational Without models, automated verification is impossible.

==To separate protocol logic from cryptographic strength== Models help answer different questions: Symbolic models (e.g., Dolev–Yao)

  • “Is the protocol logic correct assuming crypto works?” Computational models
  • “Does this cryptographic construction resist feasible attacks?”

This separation is crucial:

  • A protocol can be logically broken even with perfect crypto
  • A cryptosystem can be cryptographically broken even if protocol logic is sound

==To make assumptions explicit== Every security claim depends on assumptions. Models force you to state them clearly:

  • Is encryption probabilistic?
  • Are keys long-term or ephemeral?
  • Can sessions run concurrently?
  • Are hashes ideal? This prevents hidden assumptions, which are a major cause of insecurity.

==To reason about composition and real-world use== Protocols rarely run in isolation. Models like Universal Composability (UC) ensure:

  • Security still holds when protocols interact
  • Security survives concurrent executions
  • Components can be safely reused Without models:
  • A “secure” protocol can become insecure when combined with others

==Why multiple models exist (not just one)== No single model captures everything:

Model Strength Weakness
Dolev–Yao Finds logic flaws, automatable Ignores crypto weaknesses
Computational Realistic crypto Hard to automate
UC Strong composability Very complex
Info-theoretic Absolute guarantees Often impractical

We need different models for different questions.

Types of Models of Communication & Cryptographic Protocols

🤖 GPT-5.2 https://chatgpt.com/share/69a6af9d-125c-8010-bb42-788c68a525ca

Model Family Example / Framework Crypto Assumption Attacker Network Power Computation Power Corruption Model Concurrency Physical / Leakage Typical Use
Symbolic Dolev–Yao model Perfect (black-box) Full control Term algebra Usually none Multi-session (abstract) None Protocol logic flaws
Symbolic Strand Spaces Perfect Message interception Symbolic None Yes None Authentication proofs
Symbolic (Algebraic) DY + XOR / DH Idealized but algebraic Full control Symbolic w/ equations None Yes None Diffie–Hellman modeling
Computational Bellare–Rogaway (IND-CPA/CCA) Real crypto Oracle access PPT Static/adaptive Usually single-session None Encryption security
Computational Universal Composability (UC) Real crypto Fully adversarial scheduling PPT Static / adaptive Fully concurrent None Composable protocols
Hybrid Computationally Sound DY Real crypto Full network PPT Limited Multi-session None Bridging symbolic & real
Info-theoretic Shannon secrecy Perfect secrecy Full Unbounded None Usually single None One-time pad theory
Distributed Systems Byzantine model Crypto optional Message scheduling Unbounded / bounded Arbitrary faulty nodes Yes None Consensus
MPC Semi-honest Real crypto Honest protocol execution PPT Passive corruption Yes None Privacy computation
MPC Malicious adversary Real crypto Arbitrary deviation PPT Active corruption Yes None Secure computation
Leakage Model Side-channel model Real crypto + leakage Physical access Varies N/A N/A Timing/power/cache Hardware security
Fault Model Fault injection Real crypto Physical tampering Varies N/A N/A Faults allowed Embedded security
Post-Quantum Quantum adversary Quantum-resistant Oracle access Quantum Static/adaptive Yes None PQC security
Rational Game-theoretic adversary Crypto optional Strategic deviation Utility-maximizing Strategic Yes None Blockchain, mechanism design

This unified version includes:

  • Crypto abstraction level
    • Perfect symbolic
    • Real computational
    • Ideal functionality (UC)
    • Information-theoretic
  • Network power
    • Passive
    • Full Dolev–Yao
    • Oracle-based
  • Computation bounds
    • Symbolic term algebra
    • PPT
    • Unbounded
    • Quantum
  • Corruption type
    • None
    • Passive (semi-honest)
    • Malicious
    • Adaptive
    • Byzantine
  • Concurrency model
    • Single-session
    • Multi-session
    • Fully concurrent composability
  • Physical realism
    • No leakage
    • Side-channel
    • Fault injection

How to Read This Big Picture Think of adversary models as spanning three orthogonal axes:

  • Axis A — Cryptographic Realism
    • Symbolic → Computational → Info-theoretic
  • Axis B — System Interaction Power
    • Passive → Active → Fully adaptive + concurrent
  • Axis C — Physical Capabilities
    • Network-only → Side-channel → Fault → Quantum

Different research communities emphasize different axes.

Ref