Dynamic Trust Models for Ubiquitous Computing Environments

1
Dynamic Trust Models for Ubiquitous Computing
Environments

• Colin English, Paddy Nixon, Sotirios Terzis,
  Andrew McGettrick, Helen Lowe
• Department of Computer and Information Sciences
• University of Strathclyde

2
Vision of Ubiquitous Computing

• Massively networked world with a diverse
  population of mobile entities
• Benefit from cooperation and interaction
• No specific security infrastructure
• New security challenges not addressed in existing
  security models
• Inherent risk of interaction where only partial
  information is available

3
Trust-Based Security

• Humans use trust everyday as a means of reasoning
  about and accepting risk
• Lacking a precise definition
• Certain aspects are widely acknowledged
• Trust is a subjective notion based on the
  evidence available
• Trust is also situation specific
• Trust is dynamic, evolving as new information
  becomes available

4
Trust Management Systems

• Current trend is to view trust implicitly through
  the delegation of privileges
• The lack of explicit trust evaluation precludes
  many of the aspects deemed necessary to reason
  about trust
• Problems for decentralised ubiquitous systems
• Reliance on complete information
• Reliance on a specific security infrastructure
• Dynamic aspects of trust are largely neglected

5
Position Objectives

• The ability to form and evolve explicit values
  for trust in other principals in an interaction
  allows autonomous computational entities to make
  better decisions in situations where only partial
  information is available
• The aim is to help create a user-intuitive
  Information Society
• Objectives
• Define a trust model to allow entities to reason
  about and compare the trustworthiness of other
  entities for security decisions
• Capture the dynamic aspects of trust with fine
  granularity
• Capture human intuitions about trust to ensure
  understanding

6
Characteristics of the Trust Model

• Range of explicit values representing trust
  provides a finer granularity of representation
• More information for security decisions
• Values stored in memory
• Three main sources of trust
• Personal Observations
• Recommendations
• Reputation

7
Dynamic Aspects of the Model

• Trust Formation
• evidence relevant to the current context carrying
  the most weight
• Trust Evolution
• Evaluation of the experience
• Certificate revocation insufficient
• Trust Exploitation
• Behaviour based on trust, risk and utility
• Risk is the probability of an outcome and the
  associated costs/benefits

8
Status

• Formal trust model
• Lattice of trust values
• Risk model
• Capture the interaction between trust and risk
• Entity recognition mechanisms
• Jean-Marc Seigneur
• Simulation
• trust based file sharing system
• trust based dynamic routing in ad-hoc networks

9
Open Issues

• The nature of context
• Recording of evidence
• Combining the three forms of evidence
• Issues of Byzantine behaviour
• Non-cooperative scenarios

10
Acknowledgements

• The work is this paper is supported by the EU
  project SECURE (IST-2001-32486) funded by the FET
  Programme under the GCI.
• http//www.cs.tcd.ie/Jean-Marc.Seigneur/secure/ind
  ex.htm