Methods to protect bio-metric technologies without violating an individual’s right to privacy.
Biometrics can identify a person with a certain likelihood. Multiple biometrics methods can be combined.
This is sensitive, since one may have, through biometrics, access to many data concerning an individual.
The technology in itself is not dangerous, but which organisation is using it, for what purpose, and with which control? That is always the question: granting access to people, but giving people control on the organisations accessing their data
Biometric credentials cannot be 100% secured in the companies' databases, as well as knowledge-based, and what user have credentials. However, stolen or lost biometric credentials cannot be reinstated, replaced, or renewed, whereas other ones can. Hence, biometric credentials should not be permitted by privacy laws as providing high danger to individuals, enterprises, and government organizations.
What control method do you think can protect the template database of the bio-metric technologies?
Whatever control methods will be applied won't be 100% efficient. If you can 100% automate still malfunctioning of the systems may arise, meanwhile if it depends on people to protect them depends on what moral perspective one has with the technology and information. Where the moral thing is neither static nor universal.
However, can think of hybridisation of control methods where on method controls another in the mutual trust way. This may raise the efficiency of the control.
Dear Gao,
Thank you for the message, Could you please, send the link for the papers mentioned.
Regards
There are different ways to protect biometric data which is a very sensitive personal data, as for example :
- Secure computing: matching in the encrypted domain [Bringer, et al. 2012], [Chabanne, et al. 2013].
- Crypto-biometrics: Fuzzy vault, Secure Sketches [Rathgeb and Uhl 2011]
- Transformation: BioHashing [Teoh, et al. 2004]
Is nor an exhaustive list, nor a survey. Just to give some ideas.
I think using Multi-biometric system can overcome the limitations of unimodal biometrics, improve the performance of the overall system and enhance the security. Also, using vital biometrics such as (ECG and EEG) offers the advantage of liveness detection to the system that makes the system robust to spoof attacks. For template protection I have published a paper "
Cancelable biometric authentication system based on ECG" you can see it I discuss some techniques of template protection.
Kind regards,
Mohamed
Dear Mohamed,
Thank you for the message, could you please provide the link for the paper mentioned.
Best Regards
The proposed approach above only shifts the problem to another representation space, as changing coordinates from (x, y) to (r, theta).
Any form of re-identifiable data needs protection, including biometrics data of people. That includes voice recordings, pictures, etc...
@Arpita Sarkar
If an attacker has a new transformation table and at least one person's row bio-data, or the previous template - perhaps, it could be enough to brute force the new one way function altogether.
@Arpita Sarkar
The transformed template has a genetic link to the user fictitious names list, as well as any other reincarnation of such a template with a new one way function. Also, the new one way function is in the code to preprocess any incoming data for authentication. It creates options for external cyber intrusions as well as it could be a danger for an internal attack by a rogue employee. It makes sense only under condition that a network security is stronger than the security of the authentication protocol. Those are just my brief comments from the first glance.
@Arpita Sarkar: "Moreover the transformation function is one-way …"
Some of them have a quite simple structure, yet a brute force attack would use a direct function operation of the one way function that is not much computer intensive.
@Arpita Sarkar
Please, see my comments preceded with >>
_____
With internal attack in place ,it is not only difficult but impossible to defend any attacks whatsoever.
_____
>> It's a very generic allegation, and it isn't true. It's possible to make it extremely difficult for an internal intruder. Moreover, it should always be considered as an important task for any practical security system.
_____
If the one-way function is something like this y=( ax+b) mod 20 say , then for each x if we change the value of a and b every time we will be able to generate different value of y every time.
_____
>> It will swell the memory occupied by user information, increase program complexity and reduce maintainability, reduce operational speed, and create undesirable complexity in implementing multithreaded solutions important for any mass security system. Product managers are unlikely to approve this.
_____
Secondly it is impossible to derive value of x from given value of y, a,b . And it is impossible because of making use of modulus operation.
_____
>> x is your original biometric code, which security depends on many factors, not necessarily algorithmic ones.
I agree with Len Leonid Mizrahi, the method proposed by Arpita Sarkar has only a chance to work if the transformation itself is encrypted. But then biometric data encryption happens in the process.
Thank you all for the comments discussed from Arpita, Len and Renaud
it has been a very interesting dialog
Regards
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS