Trustworthy Sensor Networks

1
Trustworthy Sensor Networks

• Daniel Aegerter, 41542053
• Supervisor Rajan Shankaran

2
Agenda

• Problem Statement
• Wireless Sensor Networks
• Notion of Trust in Wireless Sensor Networks
• Comparative Analysis
• Recommendations

3
Problem

• Security is critical in many applications of
  Wireless Sensor Networks (WSNs)
• Battlefield surveillance
• Patient monitoring
• Environment monitoring
• Security mechanisms assume trustworthiness of
  participating nodes
• What happens if nodes get compromised?
• Access key material
• Change content of messages
• Drop messages
• Lives and livelihoods might depend on the
  correctness of the data

4
Wireless Sensor Networks (WSNs)

• Sensor nodes sense data from the environment and
  detect specific events
• Sensor nodes are equipped with sensors to monitor
  a wide range of physical conditions
• Temperature, Humidity, Light, Pressure, Object
  motion, Noise, etc.
• Sensor nodes are constrained by limited resources

Memory
Communicationdevice
Processor Unit
Sensor Unit
Power Unit
5
Network Architecture

• Components of Wireless Sensor Networks
• Common Sensor Nodes
• Base Station
• Multihop communication

6
Notion of Trust

• Trust is the subjective probability by which an
  individual, A, expects that another individual,
  B, performs a given action on which its welfare
  depends (Gambetta, 1988)

In the context of WSNs

• Trustworthy sensor nodes dont
• Manipulate gathered information
• Alter information received from neighbouring
  nodes
• Flood the network with bogus routing information
• Drop messages received from other nodes

7
Challenges in Evaluating Trust for WSNs

• Limited processing, storage, and energy resources
• Existing protocols and mechanisms are not
  applicable
• Minimise communication overhead
• Trusted authority not present in WSNs
• Public key mechanisms and certificates not
  suitable
• Adoption of architectural network changes
• Dynamic nature of WSNs
• Nodes may become faulty or compromised
• ? Trust re-evaluation is essential

8
Concept of Reputation-Based Trust Systems

• Trust is based on a nodes behaviour
• Does the node behave in a correct manner?
• Network events and correctness of gathered
  information
• Trust evaluation through first and second-hand
  information
• Direct observations
• Recommendations from peers
• Building blocks
• Watchdog mechanism ? collects evidence
• Reputation system ? evaluates and maintains trust

9
Comparative Analysis

• Compares five reviewed reputation-based
  frameworks
• Reputation-based framework for high integrity
  sensor network (RFSN)
• Gaussian trust model and reputation system
  (GRSSN)
• Lightweight group based trust management scheme
  (GTMS)
• Trust-based cluster head election
• Certificate and behaviour-based approach
• Determines characteristics and constraints of
  each framework
• Proposes categories to conduct the comparative
  analysis
• Trust management
• Node and network requirement
• Trust evaluation

10
Comparative Analysis Some Criteria

• Evidence
• Network events vs. correctness of sensed data
• Pre-established trust relationships
• Do sensor nodes trust each other at time of
  deployment?
• Storage complexity
• Tables, keys, certificates
• Revocation
• What happens with non-cooperative sensor nodes?

11
Comparative Analysis Outcomes

• All frameworks have some design problems
• Difficult to take all characteristics and
  constraints of WSNs into account
• Frameworks are application specific
• Different assumptions
• Different requirements (e.g. supernodes, keys,
  certificates)
• Sophisticated frameworks are more complex
• Economical issues
• Management issues
• Evidence analysis
• Network events or also correctness of sensed data?

12
Recommendations

• Collaborative reputation-based approach for
  establishing trust
• Behaviour of nodes is observed by a subset of
  sensor nodes
• Other nodes do not have to monitor network events
• Guardian nodes
• Guardian nodes evaluate collaborative
  trustworthiness of nodes
• All sensor nodes are directly observable by
  guardian nodes
• Guardian node evaluate whether a node is
  cooperative
• Opinions are shared among guardian nodes
• Blacklist entry has to be confirmed by other
  guardians
• Sensor node keep list with non-cooperative nodes

13
Recommendations - Example
14
Future Research Directions

• Analyse correctness of sensed data not only
  network events
• Trust evaluation for different roles that sensor
  node can perform
• Forward messages
• Aggregate data
• Sense information

Thank you
15
Questions