Establishing Trust In Pure Adhoc Networks

1
Establishing Trust In Pure Ad-hoc Networks

• Asad Amir Pirzada and Chris McDonald
• The 27th Australasian Computer Science Conference
• Koo, Seungmo
• (Dept of Computer Science, KAIST)

2
Contents

• Introduction
• Trust Model
• Trust Mechanism on DSR
• Trust Derivation on DSR
• Trust Quantification on DSR
• Trust Computation on DSR
• Conclusion

3
Introduction (1/2)

• Security issues of MANET
• A MANET can only exist and operate if nodes
  cooperate
• There may exist malicious nodes which eavesdrop
  and disrupt
• Passive attack eavesdrops and extract valuable
  information
• Active attack modification, fabrication,
  impersonation
• Managed ad-hoc networks
• Previous secure routing protocol in MANETs
• Dependent on a central trust authority (TA or CA)
• Requirement of pre-shared keys or digital
  certificates
• Initially configured before the network was
  established
• Contrast to the actual aim of ad-hoc networks

4
Introduction (2/2)

• Pure ad-hoc networks
• No assumed infrastructure
• Establishes an improvised network
• Computes trust levels from network knowledge
• Trustworthiness of routes can be computed
• Routes computed through this mechanism may not be
  secure
• But have an accurate measure of reliability
• ? Focused on a trust model suitable for
  application to MANETs

5
Agenda

• Introduction
• Trust Model
• Trust Mechanism on DSR
• Trust Derivation on DSR
• Trust Quantification on DSR
• Trust Computation on DSR
• Conclusion

6
Trust Model (1/2)

• General trust
• Based on all previous transactions in all
  situations
• Utility and Importance of a situation into a
  variable (Weight)
• Trust agent
• Each node has a trust agent
• Gathers data from events, filters it, assigns
  weights to each event
• Computes different trust levels based on weights
• Basically performs 3 functions
• Trust Derivation
• Trust Quantification
• Trust Computation

7
Trust Model (2/2)

• Trust Derivation
• Events gathered in passive mode
• (e.g.) frames received, data packets forwarded,
  control packets forwarded
• The information from these events is classified
  into trust categories
• Trust Quantification
• Represents trust from -1 to 1 (continuous range)
• -1 completely distrust, 1 completely trust
• Trust Computation
• Tx(y) trust of node y by node x
• Wx(i) weight of ith trust category to x
• Tx(i) situational trust of x in ith trust
  category
• n number of trust categories

8
Agenda

• Introduction
• Trust Model
• Trust Mechanism on DSR
• Trust Derivation on DSR
• Trust Quantification on DSR
• Trust Computation on DSR
• Conclusion

9
Trust Derivation on DSR (1/3)

• Acknowledgements
• Passive acknowledgement method provides
  information about next hop
• (e.g.) it is acting like a black hole if the
  packet is not retransmitted
• For every packet transmitted, the counter is
  incremented depending if the neighbor node has
  correctly forwarded it or not

Trust table based on Passive Acknowledgement
10
Trust Derivation on DSR (2/3)

• Packet Precision
• Accuracy of received data and routing packets
  offers a measure to compute trust levels
• (e.g.) if routing packets received are correct,
  the originator can be allotted a higher trust
  value along with the set of nodes provided in
  that packet

Trust table based on Packet Precision
11
Trust Derivation on DSR (3/3)

• Gratuitous Route Replies
• Route shortening to avoid unnecessary
  intermediate nodes by other overhearing nodes
• Blacklists
• (e.g.) selfish nodes
• Salvaging
• When intermediate nodes rescue a sources routing
  errors

12
Agenda

• Introduction
• Trust Model
• Trust Mechanism on DSR
• Trust Derivation on DSR
• Trust Quantification on DSR
• Trust Computation on DSR
• Conclusion

13
Trust Quantification on DSR (1/3)

• Trust category PA (derived from Passive
  Acknowledgement)

Tn(PA) situational trust in node n for trust
category PA W weight assigned to the event
that took place with node n
14
Trust Quantification on DSR (2/3)

• Trust category PP (derived from Packet Precision)

Tn(Pp) situational trust in node n for trust
category Pp W weight assigned to the event that
took place with node n
15
Trust Quantification on DSR (3/3)

• Trust category BL (derived from Blacklists)
• Trust category GR (derived from Gratuitous Route
  Replies)
• Trust category SG (derived from Salvaging)

B boolean reflecting the presence or absence of
a node in the blacklists
Tn(X) situational trust in node n for trust
category X W weight assigned to the event that
took place with node n
16
Agenda

• Introduction
• Trust Model
• Trust Mechanism on DSR
• Trust Derivation on DSR
• Trust Quantification on DSR
• Trust Computation on DSR
• Conclusion

17
Trust Computation on DSR

• Trust T in node y by node x is represented as
  Tx(y)
• In order to determine an aggregate trust level
  for a particular node, situational trust values
  are combined with assigned weights

Tx(C) situational trust of x in that trust
category C Wx(C) weight assigned to a trust
category C by x
Aggregate trust table
18
Agenda

• Introduction
• Trust Model
• Trust Mechanism on DSR
• Trust Derivation on DSR
• Trust Quantification on DSR
• Trust Computation on DSR
• Conclusion

19
Conclusion

• Feature
• A framework for trust establishment in MANET
  without CA
• The route found using this may not be safe in
  terms of security
• However, the most trustworthy path to the
  destination can be found
• Aims to build confidence measures regarding route
  trustworthiness
• Drawbacks
• Problems about the passive mode
• Ambiguous collision cannot hear due to local
  collision in passive node
• Receiver collision cannot hear due to remote
  collision at next hop receiver node
• Passive ACK might not work due to using varying
  transmission power ranges
• Future works
• Implement this model on DSR
• Extending this model to AODV, DSDV, and TORA