Objective and Secure Reputation-Based
Incentive Scheme for Ad-Hoc Networks
Dapeng Oliver Wu
Electrical and Computer Engineering
University of Florida
(Joint work with Qi He and Pradeep Khosla at Carnegie Mellon University)
1

• Mobile ad hoc network (MANET) has no fixed infrastructure
• Communications rely on intermediate nodes
But why should intermediate nodes relay?
• Need incentive mechanism for packet forwarding in non-cooperative MANET
2

Problem and motivation
Previous work
 Reputation-based schemes
 Pricing-based schemes
Our scheme
 Design objective
 Basic scheme
 Security enhancement
Conclusion
3

(S. Marti et al , Stanford University, 2000)
 Watchdog: identifies selfish nodes
S A B
X C D
 Pathrater: gets around identified selfish nodes
X Y
S A B C D
4

(S. Marti et al , Stanford University, 2000)
Pros:
 Improve throughput
Cons:
 Unfairly makes well behaving nodes busier
 Indirectly encourages misbehavior
5

(S. Buchegger and J-Y Le Boudec, IBM and EPFL, 2002)
 Detect misbehavior of neighbors
 Share reputation information with friends
 Punish selfish nodes based on the shared information
6

(S. Buchegger and J-Y Le Boudec, IBM and EPFL, 2002)
Pros
 Use keys to authenticate nodes
 Identify and punish misbehavior
Cons
 How to build a network of friends is not clear
 Key distribution is not addressed
 Globally shared reputation makes it not scalable
7

Problem and motivation
Previous work
 Reputation-based schemes
 Pricing-based schemes
Our scheme
 Design objective
 Basic scheme
 Security enhancement
Conclusion
8

(L. Buttyan and Hubaux, Swiss Federal Institute of Technology -- EPFL, 2000)
Scheme
 Virtual currency (nuglet)
 Centralized authority issuing nuglets
 Same amount of packets to forward
 Tamper-resistant hardware
Problem:
 Require balanced traffic
9

Micro-payment Scheme Encouraging Collaboration
M. Jakobsson, J-P Hubaux, and L. Buttyan
RSA Lab, Swiss Federal Institute of Technology, 2003
Multi-hop Cellular Networks (hybrid network)
 Mobile nodes form ad-hoc networks
 Base stations are connected to a backbone network backbone
10

M. Jakobsson, J-P Hubaux, and L. Buttyan
RSA Lab, Swiss Federal Institute of Technology, 2003 backbone
Registers to home network which shares a secret key move
Accounting Center
(Clearing house)
MAC
$
1.
Select a reward
2.
Generate an MAC
3.
Send out the packet
Forward the packet
Keep the MAC for reward
1.
Check MAC
2.
Send service record to clearing house
11

M. Jakobsson, J-P Hubaux, and L. Buttyan,
RSA Lab, Swiss Federal Institute of Technology 2003
Pros
 Symmetric key crypto: reduce computational cost
 Payment aggregation: lower communication cost
Cons
 Substantial communication overhead
 Requirement of infrastructure
 Centralized trust authority
12

Problem and Motivation
Previous work
 Reputation-based schemes
 Pricing-based schemes
Our scheme
 Design objective
 Basic scheme
 Security enhancement
Conclusion
13

 Practicality
 Available technologies
 Realistic context of ad-hoc networks
 Efficiency
 Affordable computational cost
 Moderate communication overhead
14

 Nodes are non-cooperative
 No collusion among nodes
 Broadcast transmission
 All participating nodes desire to communicate
 Invariant identity
 Selfish but not malicious
 Promiscuous mode (listening mode)
15

Problem and motivation
Previous work
 Reputation-based schemes
 Pricing-based schemes
Our scheme
 Design objectives
 Basic scheme
 Security enhancement
Conclusion
16

 Each node N maintains a Neighbor Node List (NNL
N
)
 RFP
N
(X): (Requested to Forward Packets)
The number of packets N requests X to forward
 HFP
N
(X): (Has Forwarded Packets)
The number of packets that have been forwarded by X and noticed by N
 LER
N
(X): Local Evaluation Record {G
N
(X), C
N
(X)}
Generosity
G
N
( X )

HFP
N
( X )
RFP
N
( X )
Confidence
C
N
( X )

RFP
N
( X )
17

 Every neighbor has its local evaluation record about X.
 Everyone periodically broadcasts its LER(X).
OER 
N
( X )

 k

NNL
Credibility N

{ N },  k

1
X
( i

)
N 
( k )

C
( ( i i k
( X ) i

NNL
N


{ N },
 i

N
X
( i
(X)

0
)

C i
( X )

G i
( X ) node i earned from N.
  

N
N
( N
( A )

N
)
( B )
*
*
*
C
N
(X), G
N
(X)
C
A
(X), G
A
(X)
C
B
(X), G
B
(X)
N
A
C
A
(X), G
A
(X)
X
B
G
B
(X), C
B
(X)
18

OER
N
( X )

 k

NNL
N

{ N }, k

X
1

N
( k )

C k
( X ) i

NNL
N


{ N

N
}, i

( i
X
)

C i
( X )

G i
( X )
 Quantified by objective observations
 Weighted by confidence for accuracy
 Weighted by credibility to limit impact of selfish nodes e.g., fake a non-existing node to broadcast information
19

Punishment Action
Drop packets from X with a probability p : p
 q
  if q
 
0 otherwise
Selfishness q = 1 - OER
N
(X)
20

 Network Simulator (NS-2)
 Total number of nodes: 50 (5 selfish nodes)
 Area: 670X670m 2
 IEEE 802.11 for medium access control
 DSR for routing
 CBR traffic: 1 packet/s
 No. of connections: 10
 Connection duration: 10s
 Random waypoint mobility model
 Max speed of movement: 20m/s
21

0.25
Well-behaving node
Selfish node
0.2
0.15
0.1
0.05
0
10 20 30 40 50 60 70
Dropping probability of selfish nodes (%)
80 90 100
22

Problem and motivation
Previous work
 Reputation-based schemes
 Pricing-based schemes
Our scheme
 Design objectives
 Basic scheme
 Security enhancement
Conclusion
23

Impersonate a node with a good reputation to propagate fake observation information
C
A
(X), G
A
(X)
C
A
(X), G
A
(X)
N
C
B
(X), G
B
(X)
C
A
(X), G
A
(X)
A
C
A
(X), G
A
(X)
X
B
C
B
(X), G
B
(X)
24

Identification and Authentication r h ( r ) … h n
 i
 d
( r ) … h n
 i
( r )
… h n

1
( r ) h n
( r ) f f f
K n
K n

1
K i
 d
K i
K
1
… … …
{ M y
| MAC ( K i
 d
, M y
) | h n
 i
( r )} { M x
| MAC ( K i
, M x
) | h n
 i
 d
( r )} f ( h n
 i ( r ))

K i
25

 Incentive scheme with punishment mechanism
• Reputation objectively quantified by observations
• Punishment action quantitatively suggested by reputation
• Effectively identify and punish selfish nodes
 Security enhancement
• Identification and authentication constructed collectively
• Protection from impersonation
26

27