Performance of Adaptive Beam Nulling in - PowerPoint Presentation

Slide1

Performance of Adaptive Beam Nulling in Multihop Ad Hoc Networks Under Jamming

Suman

Bhunia

,

Vahid

Behzadan

, Paulo Alexandre Regis,

Shamik

SenguptaSlide2

Outline

Introduction

Motivation

Some Related WorkProposed ModelAlgorithmsPerformance EvaluationConclusion and Future Work

2Slide3

Multi hop ad hoc networksAd hoc

: Collection of nodes communicating with each other independent of a central infrastructure.

Multihop

: Data traverses through multiple nodes Applications include sensor networks, vehicular networks, emergency radio networks in disaster zones, tactical mobile networks, and UAV communications3Slide4

Jamming based DoS attack

Wireless

medium is vulnerable to jamming based denial of service attack.

Attacker emits jamming signal to create high interferenceJamming a subset of nodes in multihop networks is sufficient for maximal disruptionDisruption of omnidirectional radios completely disables the node4Slide5

Adaptive Beamforming

Spatial filtering of

Tx

/RxAdjust the influence of signals received by different array element via controlling the weights of signal streamsAdaptive NullingWeights chosen to suppress signals arriving from certain directionsFiltering sources of interferenceDirection of Arrival (DoA): Signal arrives at elements in different timesEstimation of DoA based on time(phase) difference between elements5Slide6

Outline

Introduction

Motivation

Some Related WorkProposed ModelAlgorithmsPerformance EvaluationConclusion and Future Work

6Slide7

Advantage of ANA against jammingBefore jamming

Shortest path routing

A

− B − C − DAfter jammingOmnidirectionalE, B, C deactivatedAvoid entire jammed regionA − F − G − H − I − DANA Null jammer’s directionA − B − E − C − DNodes retain connectivity7Slide8

Aims and ObjectivesStudy adaptive

beam nulling

as a mitigation technique against

jammingMobile multihop ad hoc network Mobile jammerDevelop distributed frameworkNodes determine beamnull individuallyDynamic control of beamnull direction and width based on jammer’s mobilityInvestigate survivability of links and connectivity of network8Slide9

Outline

Introduction

Motivation

Some Related WorkProposed ModelAlgorithmsPerformance EvaluationConclusion and Future Work

9Slide10

Defense Against JammingChannel Surfing

Migrate to a channel upon detection of jamming

Proactive frequency hopping

Spatial RetreatMobile nodes relocate themselves physically Mapping Jammed RegionMulti-hop and intensely populated networkAvoid jammed links Spread Spectrumlow bandwidth data stream uses higher bandwidth channelHoneypotsingle channel honeypot based channel surfing has been proposed upon detection of attack, the network switches its channel10Slide11

Outline

Introduction

Motivation

Some Related WorkProposed ModelAlgorithmsPerformance EvaluationConclusion and Future Work

11Slide12

System assumptionsMobile jammer

Jamming signal is distinctly recognizable

Nodes monitor

DoA of jammerNodes equipped with antenna arrays and beamforming controllersIdeal beamformers – 0 gain for nulled regionsOperation time for beamnulling is negligible DoA estimation and communication occur asynchronously Periodic sensing between communicationsLink failure between 2 nodes occurs when:2 nodes fall into the jammed regionOne node falls within the beamnull of another Mac and upper layers not affectedJammed nodes assumed to be out of range

12Slide13

Methodology Nodes monitor DoA

of jamming signal (

θ

) at every τ seconds according to their local coordinate systemHistory of jammer’s position is updatedNull width is computed based on history of jammer’s mobilityPrediction of jammer’s movement in the next τ secondsNull angle adjusted to include predicted trajectory of jammer during interval between sensing phasesA buffer width takes the possibility of jammer changing direction into account13Slide14

Wide vs. Narrow Nulling

14Slide15

Null angle calculation

Borders computed based on

DoA

and predicted movement of attacker is attacker’s DoA estimation in sensing phase

is the history of jammer’s velocity

α is

an adaptive weight to incorporate the randomness in jammer’s movement

15Slide16

Outline

Introduction

Motivation

Some Related WorkProposed ModelAlgorithmsPerformance EvaluationConclusion and Future Work

16Slide17

17Slide18

Outline

Introduction

Motivation

Some Related WorkProposed ModelAlgorithmsPerformance EvaluationConclusion and Future Work

18Slide19

Simulation Parameters19Slide20

Simulation snapshot20Slide21

Performance MetricsConnectivity

is

defined as the total number of

connected pairs Average number of active links A link is the one hop communication between two neighborsAverage number of islandsNumber of isolated groups of nodesFor completely connected network, the number of island is 121Slide22

Simulation with fixed α

22Slide23

Jammer’s trajectory models

23Slide24

Results for different trajectories

24Slide25

Simulation with varying number of Nodes

25Slide26

Simulation with varying error

26Slide27

Outline

Introduction

Motivation

Some Related WorkProposed ModelAlgorithmsPerformance EvaluationConclusion and Future Work

27Slide28

Conclusion and Future WorkInvestigated

the performance of adaptive

beam nulling

in multihop ad hoc networks under attack from a moving jammer.Connectivity of various network topologies with different mobility patterns of the jammer are studied.Effects of varying inherent errors are observed.Results demonstrate a significant improvement in survivability of connectivity.Future work:Beam nulling in 3D spaceSophisticated tracking mechanismCross-layer optimization28Slide29

AcknowledgementThis research was supported by NSF CAREER grant

CNS

#1346600 and CAPES Brazil #13184/13-0.

29Slide30

30

Thank You!