• Aucun résultat trouvé

Preventing denial of service and selfishness in ad hoc networks

N/A
N/A
Protected

Academic year: 2022

Partager "Preventing denial of service and selfishness in ad hoc networks"

Copied!
9
0
0

Texte intégral

(1)

Prevention of Denial of Service Prevention of Denial of Service

Attacks in MANET Attacks in MANET

Pietro

Pietro Michiardi Michiardi - - Refik Refik Molva Molva Institut EURECOM Institut EURECOM

Security in MANET Security in MANET

„„

A priori trust (military, corporate) A priori trust (military, corporate)

„„ Entity authentication ⇒Entity authentication ⇒correct operationcorrect operation

„

„ But: But:

requirement for

requirement for tampertamper--proof hardware and strong proof hardware and strong authentication infrastructure

authentication infrastructure

„„

No a priori trust (metropolitan) No a priori trust (metropolitan)

„

„ authentication does not guarantee correct operationauthentication does not guarantee correct operation

„

„ cooperative security schemescooperative security schemes

(2)

Node Misbehavior Node Misbehavior

Selfish Nodes Selfish Nodes

„

„ Do not cooperateDo not cooperate

„

„ Priority: battery saving Priority: battery saving

„„ No intentional damage to No intentional damage to other nodes.

other nodes.

„

„ Attacks:Attacks:

„„passive denial of passive denial of service

service

„„black holeblack hole

„„idle statusidle status

Malicious Nodes Malicious Nodes

„

„ Goal: damage to other Goal: damage to other nodes

nodes

„„ Battery saving is not a Battery saving is not a priority

priority

„

„ Attacks:Attacks:

„„active denial of active denial of service

service

„

„traffic subversiontraffic subversion

„

„attacks exploiting attacks exploiting the security the security mechanism mechanism

20 Nodes , Speed: 2m/s

0 0.2 0.4 0.6 0.8 1 1.2

0 5 10 15 20 25 30 35 40 45 50

% of Se lfish Node s

Throughput Type 1

Type 2

60 Nodes , Speed: 2m/s

0 0.2 0.4 0.6 0.8 1 1.2

0 5 10 15 20 25 30 35 40 45 50

% of Selfish Nodes

Throughput

Type 1

Type 2

20 Nodes , Speed: 2m/s

1.5 2 2.5

]

Type 2

60 Nodes , Speed: 2m/s

0.6 0.7 0.8 0.9 1

Example: passive denial of service attacks

(3)

Cooperation enforcement in Cooperation enforcement in

MANET MANET

„

„

Prevent network/service utilization by Prevent network/service utilization by misbehaving nodes.

misbehaving nodes.

„„

Approaches: Approaches:

„„

metering (currency) metering (currency)

„„

monitoring monitoring

CORE: design principles CORE: design principles

Local Reputation

Local Reputation as a measure of a node’s behavior.as a measure of a node’s behavior.

Basic idea Basic idea:: good reputation

good reputation ⇒⇒node can use the networknode can use the network bad reputation

bad reputation ⇒⇒network utilization gradually deniednetwork utilization gradually denied

⇓⇓

Isolation of misbehaving nodes Isolation of misbehaving nodes Utilization Contribution

(4)

requestor provider

req(f)

Start

monitoring f execute(f)

Validate(f) ack(f)

promiscuous listening

CORE: the protocol CORE: the protocol

Provider’s Direct Reputation Ê Indirect Reputation

for all cooperating nodes Ê

No Attack No Attack

a

b c i

d

g l

e

h m

f n

b: Ê

d: Ê Ê

e: Ê

h: Ê

m: Ê

Reputation table

Direct reputation

Indirect reputation

S

D

(5)

Black Hole (Passive DoS) Black Hole (Passive DoS)

a

b c i

d

g l

e

h m

f n

d: —

l: —

h: Ì

b: —

c: —

Node l has a valid route to node e:

<l, g, h, e>

Node h does not perform the PF function

No Distribution of Negative Ratings

DoS using CORE? (active DoS) DoS using CORE? (active DoS)

a

b c i

d

g k

e

h m

f n

b: —

e:

g:

...

c: good —

d: —

g: Ì Ì ...

c : good —

m: —

g : Ì Ì ...

Since nodes b and k know that node c has a good reputation, the peer validation mechanism detects the bogus explicit DoS and decreases g’s reputation.

(6)

Traffic subversion (active DoS) Traffic subversion (active DoS)

a

b c i

d

g l

e h

m

f n

l: —

e: —

g: Ê Ê

...

b: —

i : —

g: Ì Ì ...

source destination

DSR route request function

DSR route reply Packet forwarding function

c: —

a: —

g: Ì Ì ...

CORE: Properties CORE: Properties

•Passive DoS attacks are detected and cooperation is enforced

•Active DoS attacks are prevented

• Decaying of reputation

• Reputation is hard to build

DoS attacks exploiting CORE are prevented

Peer validation No rating distribution +

(7)

Game Theory Example Game Theory Example

MANET MANET

a

b c d

• m-dimensional prisoner’s dilemma

• sequential game

• players = {a, b, c}

• Set of actions {cooperate, defect}

• Players’ choice based on utility function

BEST STRATEGY: defect f = Packet forwarding

Game Theory Example Game Theory Example

MANET with CORE MANET with CORE

a

b c d

• Find a utility function that reflects the reputation mechanism

• Show that best strategy is to cooperate

f = Packet forwarding

(8)

Utility Function Utility Function

„

„

Equity, reciprocity and competition (ERC) Equity, reciprocity and competition (ERC)

„„

large group of players large group of players

= ∑

+

j j i i

i i i

i

y y

r y

u σ

σ β α

: share relative

: function

utility ( ) ( )

One One - - shot PD Game shot PD Game

Cooperation payoff

Cooperation payoff y y

ii

=B(k = B(k) ) – – C(k C(k) ) Defection payoff

Defection payoff y y

ii

=B(k = B(k) ) N nodes, k cooperate N nodes, k cooperate



 

+ +

− +

+

− + +

+

+ . ( 1) ( 1) ( 1)

) 1 ( ) 1 )] (

1 ( ) 1 (

[ NB k k C k

k C k

r B k

C k

B

u i

i β

α

: fect) Utility(de

: operate) Utility(co

(9)

Incentive Structure based on CORE Incentive Structure based on CORE

A1. Cooperation is socially desirable:

A1. Cooperation is socially desirable:

A2. Cooperation is individually desirable:

A2. Cooperation is individually desirable:

Analysis (based on A1 and A2):

Analysis (based on A1 and A2):

„„ Utility(cooperateUtility(cooperate) ) ≥≥Utility(defectUtility(defect))

„„ Solutions (Nash equilibrium): k = 0 and k Solutions (Nash equilibrium): k = 0 and k ≥≥ N/2N/2

) ( )

( )

1 ( ) 1 ( ) 1

( k k C k N B k k C k

B

N ⋅ + − + + ≥ ⋅ − ⋅

) ( )

( )

1 (

) 1

( k C k B k C k

B + − + ≥ −

Future work Future work

„„

Further investigation of ERC types ( Further investigation of ERC types ( α, α , β) β )

„„

Incentive structures (Shapley Incentive structures ( Shapley value) value)

„

„

Further validation by simulation NS Further validation by simulation NS → → QualNet QualNet

„

„

Performance evaluation. Performance evaluation.

Références

Documents relatifs

A node willing to reach a destination rst checks if the destination is in its own zone, if not it uses the Interzone Routing Protocol (IERP) and bordercasts the Route Request

The chap- ter presents the default remote invocation solution, detailing how a client can issue a content-based invocation request to all providers of a business service, and how

In this paper, we analyzed the runtime resulting from the authentication of a joining node by a distributed AAA framework within a mobile ad-hoc network. The built model

When the distance and position information of more than three nodes is collected, an estimated position is computed for each possible combination of three or more nodes,

Simulation results have shown that the proposed scheme performs generally better than the other network topology schemes in terms of: average number of CHs, average number of

In this study, we have proposed a new message dissemination protocol for ve- hicular ad-hoc networks based on cloud computing infrastructure, called Cloud computing-based

In the case of a closed wormhole attack, the malicious nodes will not appear in the information obtained in the multipath route discovery protocol, and no suspicious link with

Then, in the context of topology control for ad hoc network, Madhoc could successfully be extended in order to model connections from mobile nodes to a simulated infrastructure,