Unit 4 Biometrics
Surveillance: Technologies,
Techniques and Ramifications
Surveillance & the Internet
Digital Footprint
direct user activity
IoT-based input
Other sources of input
CCTV
Facial recognition
ANPR
IP-based CCTV
Thermal Imaging
Drones
Body cameras
Analytics
Server-based
Edge-based
Movement/object detection/tracking
Bios (life) metrikos (measure)
Metadata
data dervived from data
Blundell p29 - 4.4. Biometrics
Fingerprints
widespread use in banking/schools
Could change over time
accidents/manual labour
Facial Recognition
requires cooperation
non-intrusive
covert
Meyer's experiments at 'dazzling' camouflage in Washington DC
Sharif et al's dodging or impersonation research
Even small changes to appearance can be magnified in Deep Neural Network systems
Black box
algorythms unknown
White box
algorythms known
Iris & Retina
requires cooperation due to device requirements
can be affected by eye conditions
Even in identical twins, Iris pattern differs
DNA (deoxyribonucleic acid ) #
Human Rights Watch
40m citizens DNA captured in China
Future use of stored samples unknown
Contamination of samples can easily occur
Voice
Can be fooled by hidden voice commands or recordings
Carlini et al's research into attack approaches
Gait
characteristic walking motion
Can be captured covertly
can change over time - age/injury/modification
click to edit
General Problems
Can't be exchanged if 'stolen'
Biometric matching
False Match Rate
False Non-Match Rate
FNMR
FMR
Denoting the rate at which a system mistakenly identifies
two different people as the same person.
Denoting the rate at which a system mistakenly identifies two biometric measurements from the same person as representing different people
Ethics of requiring Biometric data from people
Thresholds for matches may be arbritray
Blundell 4.5 Augmented (mixed) Reality
Education
Hands-Free Info. Overlay
Benefitting from advice
Exams made easy
Invasion of Privacy
Medical students remote (surgery) lectures
Open Loop
Closed Loop
Information simply transmitted from the glasses
Complementing real-world view with information
e.g. Engineer's view of equipment is overlaid with diagnostic data
E.g, Transmitting the view from a lecturer's standpoint to remote students
My Biometric Data
Passport
Fingerprinted at US border
Work
Laptop had fingerprint reader
My choice to use instead of password
no choice
Eye tracking HID
My choice, but no alternative to be able to use the functionality
Retina
Opticians
Routine examination of eyes (no health monitoring without this)
DNA
Ancestry
my choice
Health Research
Cannot take part without this:
fingerprint
personal laptop
my choice
Tom Keenan 'Techno Creep' Black Hat USA 2015 - Hidden Risks Of Biometric Identifiers And How To Avoid Them
Trading Technical risk for Human risk
Neutrality
Body odour as Biometric?
ECG/EEG as biomentric
Heartbeat Pattern
Brainwave Pattern
Girls around me app
Shut down by 4Square
Facebook live feed when individuals are near
Thermal Imaging - Heat signatures?
DNA capture
From PIN # pads?
DNA truck - who's the daddy?
once Biometrics captured, can't be reversed
Face recognition
Nametag app
Sex offender database
Gait
Aberdeen research lab USA
Intellistreet cameras
Fitness trackers
Apply a 'creepy' lens
Multi-factorial
In higher risk settings
BioHacking
unregulated activities at risk of attack
Attributes of biometric measurement 
Generality: this indicates the extent to which a physiological or behavioural characteristic is exhibited by everybody. If generality is deemed high, then this suggests that the biometric measurement can be obtained from everyone. In contrast, if generality is indicated to be likely, then this suggests that some people may not be able to provide the biometric measurement. For example, DNA measurements exhibit high generality, whereas the measurement of gait would not be possible in the case of paraplegics.
Invariance: this concerns changes to a biometric characteristic that may occur over time. For example, gait gradually changes with age or infirmity, and voice-based biometric measurements may not only change with age but also with, for example, somebody who has laryngitis.
Uniqueness: this concerns the extent to which more than one person may exhibit (or appear to exhibit) identical biometric characteristics.
Collectability: here, there are three categories. Firstly, biometric collection without needing a person's cooperation (indicated as 'high' in Table 4.1). Secondly, biometric collection that requires a degree of cooperation (indicated as 'medium' in Table 4.1), and, thirdly, biometric collection that requires more-complex collection techniques (indicated as 'low' in Table 4.1). Biometrics exhibiting high collectability can often be achieved covertly (i.e. without a person's knowledge or consent).
Acceptability: this relates to public acceptance across cultures. For example, in certain cultures, facial recognition would be deemed highly intrusive. Also, techniques based on DNA, for example, may raise concerns in view of the potential for secondary use of the data.
Circumvention: this relates to the possibilities of fraud. Even in the case that a biometric measurement demonstrates a high degree of uniqueness, this does not necessarily mean that it does not lend itself to fraudulent activity.
Facial Recognition Systems
click to edit
Eye Recognition Systems
Gait #
Voice
Text dependent
Text independent
While voice interfaces allow far increased accessibility and potentially easier human-computer interaction, they are at the same time susceptible to attacks. Voice is a broadcast channel open to any attacker that is able to create sound in the vicinity of a device.
Carlini et al. (2016)
DNA