Authors: Anil K. Jain , Arun Ross and Sharath Pankanti

utterlypanoramicΑσφάλεια

30 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

81 εμφανίσεις

Authors:

Anil K. Jain , Arun Ross and Sharath Pankanti

Presented By: Payas Gupta

Outline


Today’s security related concerns


Commonly used Biometrics


Variance in Biometrics


Operation of Biometrics Systems


Attacks on Biometric System


Multi Biometric Systems


Level of fusion in Biometric Systems




Today’s Security related Questions


Is

she

really

who

she

claims

to

be?



Is

this

person

authorized

to

use

this

facility?



Is

he

in

the

watch

list

posted

by

the

government?

Biometrics,

Science of recognizing an individual based on his or
her physical or behavioral traits

Paper is all about


Examining

applications

where

biometrics

can

solve

issues

pertaining

to

information

security



Enumerating

the

fundamental

challenges

encountered

by

biometric

systems

in

real
-
world

applications




Discussing

solutions

to

address

the

problems

of

scalability

and

security

in

large
-
scale

authentication

systems

Commonly Used Biometrics

In this paper

Others, widely used


Research Going on…


Heart

Beat




Galvanic

Skin

Conductivity



Pulse

Rate



Brain

tissues



Respiration



DNA


Face


Location

and

shape

of

facial

attributes


Eyes,

eyebrows,

nose,

lips

and

chin


Global

Analysis

of

face

as

weighted

combination

of

no
.

of

canonical

faces
.



Disadvantages
:


Face

could

be

an

image

(photograph)


Difficult

to

locate

the

face

if

there

is

one


Difficult

to

match

from

any

pose


Template

Update

Problem



Fingerprint


Pattern

of

ridges

and

valleys

on

fingertip


Fingerprint

scanner,

cheap

approx
.

$
30


Fingerprint

of

Identical

twins

are

different



Disadvantage
:


Large

computation

resource

when

in

identification

mode


Hand Geometry


Human

Hand


Shape

and

size

of

palm


Length

and

widths

of

fingers


Relatively

easy

to

use


Inexpensive


Environmental

conditions

do

not

affect


Disadvantage


Not

very

distinctive


Individual

jewelry



Limitation

in

dexterity

(e
.
g
.

Arthritis)


Iris


Iris,

is

bounded

by

pupil

and

the

white

portion

of

eye


High

accuracy

and

speed


Each

Iris

is

believed

to

be

distinctive


Ability

to

detect

artificial

Irises

(contact

lenses)
.


Low

FAR

but

could

be

high

FRR



Disadvantages
:


Considerable

user

participation


expensive


Keystroke


Hypothesis


Behavioral

biometric


each

person

types

on

keyboard

in

a

characteristic

way
.


Not

unique,

but

sufficiently

different

that

permits

identity

verification



Disadvantage
:


typing

patterns

change

substantially

between

consecutive

instances

of

typing

the

password


different

style

of

keyboard

Signature


The

way

a

person

sign


Excepted

all

throughout

the

world


Behavioral

Biometric


Possibility

of

changing

in

emotional

and

physical

conditions
.



Disadvantage
:


Professional

forgers

may

be

able

to

reproduce

signatures

that

fool

the

system
.

Voice


Physical

+

Behavioral

Biometric


Depends

on
:


Shape

and

size

of

vocal

tracts,

mouth,

nasal

cavities

and

lips



Disadvantages
:


Physical

part

are

invariant

from

each

other


But,

behavioral

part

changes

from

age,

emotional

state

and

medical

conditions

(cold)
.


Background

noise



Major Problem

Inconsistent Presentation


Irreproducible Presentation

Imperfect Representation Acquisition

Operation of Biometric System

Validation

Compare these features against template

Extract Feature

Processed this signal

Raw Biometric Signal

E.g. Fingerprint

Feature set

Preprocess

Extract Features

Raw Biometric

Processed

Compare and
Validate

Functionalities of a Biometric system


Verification



“Is

this

person

truly

John

Doe?”



Identification



“Is

this

person

in

the

database?”



Screening



“Is

this

a

wanted

person?”


Matcher Accuracy


FMR

(False

Match

Rate)


Incorrectly

declares

a

successful

match


FNMR

(False

non

Match

Rate)


Incorrectly

declares

failure

of

match



FTE

(Failure

to

Enroll)


%

of

times

users

are

not

able

to

enroll

to

the

system


FTC

(Failure

to

Capture)


%

of

times,

device

fails

to

capture

the

sample

Zero effort attacks


Intruder
:

sufficiently

similar

biometric

traits

to

a

legitimate

user


Problem

under

examination


Determine

the

probability

that

any

two

(or

more)

individuals

may

have

sufficiently

similar

fingerprints

in

a

given

target

population


Given

a

sample

fingerprint,

determine

the

probability

of

finding

a

sufficiently

similar

fingerprint

in

a

target

population


Given

two

fingerprints

from

two

different

fingers,

determine

the

probability

that

they

are

sufficiently

similar
.



Adversary Attacks


Physical

traits

can

be

obtained

from

face,

fingerprints
.


Other Attacks…


Circumvention


Fraudulent

access

to

the

system,

tamper

the

sensitive

data


Repudiation


Bank

clerk

Modify

the

customer

record,

and

claim

that

intruder

spoof

his

biometric

trait


Collusion


Administrator

may

deliberately

modify

biometric


Coercion


at

gunpoint,

grant

him

access

to

the

system
.


Denial

of

Service

(
DoS
)


A

server

can

be

bombarded

with

a

large

number

of

bogus

requests

Vulnerabilities in Biometric System


Sensor

Application
Device

Matcher

Feature

Extractor

Stored
Templates

Fake Biometric

Override Feature Extractor

Override final Decision

Override Matcher

Synthesized

Feature Vector

Replay Old Data

Intercept the channel

Modify Template

Y/N

MultiBiometric systems


One can have the fusion of multiple biometrics, to get
a high amount of accuracy.


Feature Level

Score Level

Decision Level

Feature Level

FM

Feature
Extraction
Module

Feature
Extraction
Module

Templates

Matching
Module

Decision
Module

Accept/Reject

Score Level fusion

FM

Score
Normalization

Score
Normalization

Accept/Reject

Feature
Extraction
Module

Decision
Module

Matching
Module

Matching
Module

Templates

Templates

Feature
Extraction
Module

Feature Set

Feature Set

Decision Level fusion

FM

Decision
Module

Decision
Module

Accept/Reject

Feature
Extraction
Module

Matching
Module

Matching
Module

Templates

Templates

Feature
Extraction
Module

Feature Set

Feature Set

Summary and Conclusion


More

reliable

than

current

passwords


Cannot

be

easily

shared


Misplace

and

forged


A

well

implemented

biometric

system

with

sufficient

privacy

safeguards

may

be

a

further

basis

of

resistance
.


Multibiometric

systems

have

received

much

attention

in

recent

researches
.


Speech

+

face


Face

+finger


Multiple

fingers

of

users


Cont…



The complexity of biometric system depends on


Accuracy


Scale


Size of the database


Usability

Thank you


Questions / Comments?