MAB_full

19
1 Most Active Band (MAB) Attack and Countermeasures in a Cognitive Radio Network Nansai Hu Stevens Institute of Technology

Transcript of MAB_full

1

Most Active Band (MAB) Attack

and Countermeasures in a

Cognitive Radio Network

Nansai Hu

Stevens Institute of Technology

2

Outline

• CR network and its vulnerability

• Most active band (MAB) attack

• Coordinated concealment strategy

(CCS)

• Power control on CCS

• Results

3

What’s in the air

4

Cognitive radio

• Cognitive radio / Dynamic spectrum access

• Cognitive radio (CR) technology is

considered to be a promising technique

to improve spectrum utilization by

seeking and opportunistically utilizing

resources in time, frequency, and space

domains without causing harmful

interference to legacy systems.

• Cognitive users (unlicensed, secondary

users) first sense radio spectrum, detect

licensed users (primary users) and then

take use of the spectral holes left by the

primary users to achieve opportunistic

access.

5

CR’s vulnerability

• Vulnerabilities and related security issues

• Attacks in A CR node/network:

• 1) In physical layer, a CR should sense the

environment before it use the spectrum hole for

communication, which gives the attackers more

chance to manipulate a target CR network

(Interference, denial of service)

• 2) In access behavior or MAC layer, misuse

(misbehavior, selfish or cheat user) could occur

since CRs use a flexible access paradigm.

6

MAB attack

• Most active band (MAB) attack:

A malicious CR node or agent senses and

monitors the signal activities over each band

(e.g., spectrum sensing through energy

detection) and then, attacks (with intentional

interference) the band which has the most

signal activities (e.g., the highest energy level)

to achieve its maximum attack outcome.

A type of cognitive

interference which has the

spectrum sensing (energy

detection) and cognitive

engine capabilities to

determine the band with

the most signal activities.

B1 B2 B3 B4 BM ...

Ch1 Ch2 Ch3 Ch4 Chc ...

Bands

Channels

7

MAB attack description

• We consider NP primary nodes and NS secondary nodes (CR nodes) operating in a

M-band CR network.

• The number of bands with primary nodes (primary bands) is assumed to be MP and

the number of vacant bands (secondary bands) is assumed to be MS (MP+MS=M).

• In each band, C is specified as the maximum user or node capacity.

The following equation describes the band (band i*) a MAB attacker (a

malicious CR node) selects to target,

where

8

MAB effects in a CR network

• Follower interference

• A MAB attacker could potentially target either a secondary band or a primary band.

• When a MAB attacker targets one primary band, the primary nodes under attack are unable to avoid the attacker since they have no spectrum sensing and reconfiguration capabilities.

• When a MAB attacker targets one secondary band, the secondary nodes under attack could hop to other bands to avoid the attacker.

– However, the MAB attacker could follow the secondary nodes due to its energy detection (spectrum sensing) capabilities. Therefore, this CR’s inherent signal/interference avoidance capability is no longer effective in countering a MAB attack.

– Notice that the conventional frequency hopping based methods (e.g., band hopping) are also no longer effective, since the MAB attacker can follow the CR to its new operating band.

– During the process of signal/interference avoidance (band change), significant amount of control signaling occurs (e.g., request, acknowledgement and channel setup, etc.), which reduces communication efficiency and introduces extra synchronization complexity.

9

Performance evaluations

• For performance evaluations under MAB attacks, we calculate the number of

surviving nodes (e.g., nodes which are not in a targeted band) over the total

number of nodes. We use ASi,i*(j) and AP

i,i*(k) to denote that whether a

secondary/primary node is under attack, respectively.

Further, the percentage of the total surviving nodes in the network, V , can be

obtained by

10

Considering a uniform distribution of primary nodes among primary bands

and a uniform distribution of secondary nodes among secondary bands,

we can easily calculate the total number of surviving nodes in a typical CR

network (with CR’s inherent signal/interference avoidance capability).

• In the scenario of a primary band is under attack, the total number of

surviving nodes in the network is

• In the scenario of a secondary band is under attack, the total number of

surviving nodes in the network is

CR’s inherent

signal/interference avoidance

11

Solutions

• Coordinated concealment strategy (CCS)

• Power control on CCS

12

CCS

• Coordinated concealment strategy (CCS)

• Where rj, rk follow

• In implementing CCS,

the distances between

nodes and the

attacker (rj and rk) can

be estimated based

on signal strength

information.

13

Solutions

• Coordinated concealment strategy (CCS)

• Power control on CCS

14

Power control

• Power control is widely used in wireless

communications

– Interference managements

Here, we introduce power control capability to CCS

15

Power control in CCS

The CCS algorithm as defined in (8) through (14) can be further improved by

incorporating power control in the secondary nodes.

The CCS algorithm with power control can be defined using (8) through (14),

substituting (10) with

In addition, we have the following constrains in implementing the power control.

Pj presents the transmission

power of secondary node j.

16

Results

The geographical locations of primary and second nodes are determined by R=1000m

and R0=10m. We place an attacker in the center of a simulated network, considering a

six-band CR network (M=6) where 50 primary nodes (NP=50) are operating within one

band (MP=1). The capacity of each band, C, is assumed to be 50. The number of

secondary nodes (NS) varies from 20 to 250.

• When NS=100, the

primary node survival

percentage increases from

approximately 60% (signal

avoidance), to 75% (CCS)

and 95% (CCS with power

control).

• The secondary node

survival percentage

increases from

approximately 85% (signal

avoidance), to 97% (CCS)

and 99% (CCS with power

control).

17

Results (Cont.)

When NS = 250, the total

primary node and

secondary node survival

percentage improves

from approximately 80%

(signal avoidance), to

95% (CCS) and 98%

(CCS with power control).

• Total survival percentage of PU & SU

18

Fairness issue

Due to the

– Nature of random distributions of SUs

– Movements

– Effect of channel fading

a SU node can be ”randomly” selected as a sacrificing node (following a network optimization process). This, to a certain extend, inherently addresses the issue of fairness among all SUs.

19

Conclusions

• We introduce a MAB attack and investigate its impacts on a CR network.

• We further proposed a countermeasure strategy (CCS) towards the MAB attack. The results show that the CCS outperforms CR inherited signal/interference avoidance and frequency hopping method.

• Meanwhile, we develop the power control capability of CR nodes with the CCS method and obtain a better countermeasure performance.