CSC 379:Week 4, Group 4: Difference between revisions
No edit summary |
|||
| (6 intermediate revisions by 2 users not shown) | |||
| Line 23: | Line 23: | ||
==What are Biometrics?== | ==What are Biometrics?== | ||
A biometric is a measure of some physical or | A biometric is a measure of some physical or behavioral characteristic of an individual. These measurements can then be used to identify and individual or at least aid in confirming a claimed identity. Some examples of human characteristics that can be conveniently captured in a biometric include: Fingerprints, Retinal scan, Voice, Signature, DNA, Gait. After your individual information is recorded it can be used in a biometric system. | ||
===The Purpose of Biometrics=== | ===The Purpose of Biometrics=== | ||
| Line 42: | Line 42: | ||
===Eye scans=== | ===Eye scans=== | ||
There are two kinds of eye scans, one that scans the retina(blood vessels) and the other that scans the iris(colored part). Both of these share the advantage of scanning an internal organ which is less vulnerable to damage over time. Both suffer from a difficulty in building a database to compare to due to how they have to scan in addition to a lack of public approval for getting their eyes scanned. People associate the technology with something they would see in a movie and pair it will a loss of privacy as is the case in such movies. | There are two kinds of eye scans, one that scans the retina (blood vessels) and the other that scans the iris(colored part). Both of these share the advantage of scanning an internal organ which is less vulnerable to damage over time. Both suffer from a difficulty in building a database to compare to due to how they have to scan in addition to a lack of public approval for getting their eyes scanned. People associate the technology with something they would see in a movie and pair it will a loss of privacy as is the case in such movies. | ||
===DNA scans=== | ===DNA scans=== | ||
| Line 53: | Line 53: | ||
This method relies on the unique manner in which a user types on a keyboard to identify them. After an initial session of typing where the system learns to identify the user's typing style, the system can compare a typing pattern to the reference template it has created for the user. This form of user authentication can be inexpensive since it only requires software and an existing keyboard; however, it is not entirely reliable. Using different keyboards alters a user's typing pattern, and something as simple as being tired could affect a person's typing enough for the system to fail to recognize it. Keystroke dynamics has a false acceptance rate of 7% and a false reject rate of 0.1%. | This method relies on the unique manner in which a user types on a keyboard to identify them. After an initial session of typing where the system learns to identify the user's typing style, the system can compare a typing pattern to the reference template it has created for the user. This form of user authentication can be inexpensive since it only requires software and an existing keyboard; however, it is not entirely reliable. Using different keyboards alters a user's typing pattern, and something as simple as being tired could affect a person's typing enough for the system to fail to recognize it. Keystroke dynamics has a false acceptance rate of 7% and a false reject rate of 0.1%. | ||
[http://play.tm/wire/click/1240438 The public versus keystroke recognition security] | |||
===Signature Verification=== | ===Signature Verification=== | ||
Signature identification analyzes an individual's signature to determine his or her identity. This can involve analyzing the signature as it is being made, or a static image of the signature after completion. These two methods can be used to examine features of either the signature itself or the process by which it was signed, including qualities such as speed, pen pressure, directions, and stroke length. Because individuals' signatures change over time, signature identification systems take into any account slight variances over time. Notably, the Internal Revenue Service has used signature identification analysis for electronically filed tax returns. | Signature identification analyzes an individual's signature to determine his or her identity. This can involve analyzing the signature as it is being made, or a static image of the signature after completion. These two methods can be used to examine features of either the signature itself or the process by which it was signed, including qualities such as speed, pen pressure, directions, and stroke length. Because individuals' signatures change over time, signature identification systems take into any account slight variances over time. This method has a low total error rate of about 1.5% per session. Notably, the Internal Revenue Service has used signature identification analysis for electronically filed tax returns. | ||
[http://www.vision.caltech.edu/mariomu/research/sigverif/ Visual Signature Verification] | |||
[http://www.haifa.il.ibm.com/projects/image/sv/ Online Signature Verification] | |||
===Voice Verification=== | ===Voice Verification=== | ||
Voice authentication relies on unique features in an individual's voice to confirm his or her identity. A voice authentication system extracts these features from a sample of the individual's voice, and compares them to a sample presented for verification. Ambient noise can affect the accuracy of this method, as can illness or even the emotional state of the individual. Because of this, voice verification technology may have up to a 2 – 5% false accept rate and an up to 10 – 20% false reject rate. Voice verification is used for a variety of applications, including using automated phone calls to monitor compliance of persons under house arrest. | Voice authentication relies on unique features in an individual's voice to confirm his or her identity. A voice authentication system extracts these features from a sample of the individual's voice, and compares them to a sample presented for verification. Ambient noise can affect the accuracy of this method, as can illness or even the emotional state of the individual. Because of this, voice verification technology may have up to a 2 – 5% false accept rate and an up to 10 – 20% false reject rate. Voice verification is used for a variety of applications, including using automated phone calls to monitor compliance of persons under house arrest. | ||
[http://www.biometricgroup.com/reports/public/reports_voice-scan.html International Biometrics Group - Voice Verification Technology] | |||
[http://www.businessweek.com/technology/content/apr2005/tc20050420_1036_tc024.htm?campaign_id=rss_techn May I See Your Voice, Please? BusinessWeek] | |||
==Ethical Issues Surrounding Biometrics== | ==Ethical Issues Surrounding Biometrics== | ||
| Line 66: | Line 76: | ||
Many forms of biometric data do not expire. While a person's face or voice may change some as they age, for the most part your physical and behavioral characteristics are going to stay the same throughout your life. | Many forms of biometric data do not expire. While a person's face or voice may change some as they age, for the most part your physical and behavioral characteristics are going to stay the same throughout your life. | ||
With current forms of identification your can, for example, change your password or get a new credit card or driver's | With current forms of identification your can, for example, change your password or get a new credit card or driver's license or passport. For the most part with biometric identification, you can't simply change your information (short of surgery). This poses a large privacy issue. Not only will anonymity be almost impossible in a world were biometric identification is the norm, but identity theft will be every more devastating because one can't simply adopt new identifying data to defeat the thief. | ||
===Biometrics as a Means of Discrimination=== | ===Biometrics as a Means of Discrimination=== | ||
Latest revision as of 20:48, 2 August 2007
Biometrics and Privacy of Genetic Data
A concern with ID cards that contain biometric information is that once one is stolen, it would be more difficult for someone to reclaim their identity as the nature of the representation of biometric data that would be used would be something that would not change over a person’s lifetime (e.g. fingerprints or eye-scan)[1]. Thus for technologies that rely solely on the biometric data contained within the IDs to establish identity, once a card is stolen, ones “identity” may never be able to be reclaimed.
Ethical issues related to privacy of genetic data follow closely with this topic since it suffers from similar issues as biometrics, although genetic data is much more invasive to ones privacy as society becomes more able to interpret it. There have been discussions of a constitutional amendment to prohibit genetic discrimination. If ones genetic information is made available, or information derived from, it opens the person to an increased threat of discrimination (jobs, insurance, social), as well as other threats to privacy not yet realized, but that will likely be realized within ones lifetime as there becomes a greater capability to interpret genetic data.
What types of protections should be afforded to biometric data compared to other types of data? For genetic data? Should biometric/genetic data be incorporated into various technology from ID cards to diagnostic equipment? Examine ethical issues related storage and use.
Resources
Relevant External Links:
The National Science and Technology Council some good resources related to biometrics and privacy issues.
Wikipedia's article on biometrics: http://en.wikipedia.org/wiki/Biometrics
EFF has an introduction to some concerns voiced about biometrics: http://www.eff.org/Privacy/Surveillance/biometrics/
Relevant Class Website Links:
What are Biometrics?
A biometric is a measure of some physical or behavioral characteristic of an individual. These measurements can then be used to identify and individual or at least aid in confirming a claimed identity. Some examples of human characteristics that can be conveniently captured in a biometric include: Fingerprints, Retinal scan, Voice, Signature, DNA, Gait. After your individual information is recorded it can be used in a biometric system.
The Purpose of Biometrics
Biometric systems have several purposes. They can verify that a person is who they claim to be by checking your information against the data the system has on you. A biometric system can identify you without you telling the system who you are by your physiological or behavioral characteristics. And biometric systems can also be used to screen individuals. The system may raise a flag or alert the police if it identifies you as being on its watch list.
Flaws in Biometric Systems
These systems are not flawless though. Depending on what characteristic is being measured, there is often a significant chance of error. Each biometric has its own false accept rate and false reject rate. A false accept rate is the chance that the biometric verifies that you are who you claim to be, when you actually aren't. The higher the percent chance for a false acceptance, the easier the system is to fool. A false reject rate is the chance that a system says you are lying about who you claim to be when you are telling the truth. The false accept rate and false reject rate of a particular biometric should be taken into account when designing an identification or verification system.
Physiological Biometric Systems
Facial Identification
Facial identification can either identify people based on the image of their face or a thermal scan of their face. At first glance this seems to be a sound method, after all it's how people identify each other without technology. Unfortunately, there are several issues with implementing it as a technology. For cosmetic scans, the image can look very different based on lighting or could be fooled with makeup. In addition, both facial and thermal scans will change as people age. While one of the easiest and least obtrusive means of identification, it is also one of the least accurate and most easily fooled.
Fingerprint
Fingerprint identification is one of the most widely used biometric means of identification in the world today. The ridges of the skin on the finger are taken and compared to known records taken in the past. It is well suited to its task for several reasons. A person's fingerprints do not change as they age nor are they easily copied. In addition they are completely unique. Due to its widespread use there are fewer personal issues with the scanning of fingerprints than some other biometric identifiers.
Hand veins
Similar to fingerprints, the lines of the hand may also be used for identification. It is harder to fool than a fingerprint scan, but also slightly less accurate. It is also more vulnerable to changes over time than fingerprints and sometimes harder to tell the difference between the scans of two individuals.
Eye scans
There are two kinds of eye scans, one that scans the retina (blood vessels) and the other that scans the iris(colored part). Both of these share the advantage of scanning an internal organ which is less vulnerable to damage over time. Both suffer from a difficulty in building a database to compare to due to how they have to scan in addition to a lack of public approval for getting their eyes scanned. People associate the technology with something they would see in a movie and pair it will a loss of privacy as is the case in such movies.
DNA scans
DNA is the most accurate way of identifying a person. Nobody, with the exception of identical twins, will have the same DNA. Unfortunately, a sample from someone else will allow a fraudulent person to trick the system. In addition, collecting information for a database to compare scans to would be a major issue, as most people do not want to go in to a government office to have their DNA sampled. This brings up the next issue, because it is so accurate there is would be a massive loss of anonymity which would receive large public backlash.
Behavioral Biometric Systems
Keystroke Dynamics
This method relies on the unique manner in which a user types on a keyboard to identify them. After an initial session of typing where the system learns to identify the user's typing style, the system can compare a typing pattern to the reference template it has created for the user. This form of user authentication can be inexpensive since it only requires software and an existing keyboard; however, it is not entirely reliable. Using different keyboards alters a user's typing pattern, and something as simple as being tired could affect a person's typing enough for the system to fail to recognize it. Keystroke dynamics has a false acceptance rate of 7% and a false reject rate of 0.1%.
The public versus keystroke recognition security
Signature Verification
Signature identification analyzes an individual's signature to determine his or her identity. This can involve analyzing the signature as it is being made, or a static image of the signature after completion. These two methods can be used to examine features of either the signature itself or the process by which it was signed, including qualities such as speed, pen pressure, directions, and stroke length. Because individuals' signatures change over time, signature identification systems take into any account slight variances over time. This method has a low total error rate of about 1.5% per session. Notably, the Internal Revenue Service has used signature identification analysis for electronically filed tax returns.
Voice Verification
Voice authentication relies on unique features in an individual's voice to confirm his or her identity. A voice authentication system extracts these features from a sample of the individual's voice, and compares them to a sample presented for verification. Ambient noise can affect the accuracy of this method, as can illness or even the emotional state of the individual. Because of this, voice verification technology may have up to a 2 – 5% false accept rate and an up to 10 – 20% false reject rate. Voice verification is used for a variety of applications, including using automated phone calls to monitor compliance of persons under house arrest.
International Biometrics Group - Voice Verification Technology
May I See Your Voice, Please? BusinessWeek
Ethical Issues Surrounding Biometrics
The Permanence of Biometric Identification
Many forms of biometric data do not expire. While a person's face or voice may change some as they age, for the most part your physical and behavioral characteristics are going to stay the same throughout your life.
With current forms of identification your can, for example, change your password or get a new credit card or driver's license or passport. For the most part with biometric identification, you can't simply change your information (short of surgery). This poses a large privacy issue. Not only will anonymity be almost impossible in a world were biometric identification is the norm, but identity theft will be every more devastating because one can't simply adopt new identifying data to defeat the thief.
Biometrics as a Means of Discrimination
If a particular biometric can not be measured on a person-- they are missing a finger, for example-- then inherently that person is going to seem more suspicious. Administrators of the system might think: "Sure maybe they have a legitimate reason for not doing the fingerprint ID, but maybe they are trying to bypass the system."
Biometrics systems could be used in small scale discriminatory screening processes such as stores using a biometric ID system to deny business to those found guilty of shoplifting in the past.
Steps to Take to Mitigate Invasion of Privacy
Paraphrased from the IBG BioPrivacy Initiative
- Biometric systems should only be expanded when absolutely necessary.
- Systems should not be expanded without public knowledge.
- Biometric information should not be used as a universal unique identifier.
"Universal unique identifiers facilitate the gathering and collection of personal information from various databases, and can represent a significant threat to privacy if misused." - These systems should not store more information about an individual than is absolutely necessary to verify the individual's identity.
Links and Resources
- Biometrics and Privacy by Roger Clarke
- Who's watching you? by William Abernathy and Lee Tien
- IBG BioPrivacy Initiative
- Biometric News Portal
- Biometrics: Security Boon or Busting Privacy? by Saumya Roy
- Good ol' Wikipedia