System, apparatus and method for privacy preserving contextual authentication
Methods and systems for authenticating users based on contextual data in a privacy preserving way are disclosed.
1. A method to secure a user's interaction with a remotely accessible computer-based application, the method comprising the steps of:
collecting one or more sets of contextual data;
anonymizing the collected one or more sets of contextual data;
transferring the anonymized contextual data sets to an authentication server;
at the authentication server analyzing the received anonymized contextual data sets for authenticating the user;
wherein the one or more sets of contextual data comprise values of one or more computable contextual data elements;
wherein the step of anonymizing the collected one or more sets of contextual data comprises transforming a first value of at least one of said computable contextual data elements using a similarity preserving anonymization transformation.
2. The method of claim 1 , wherein said analyzing the received anonymized contextual data for authenticating the user comprises exploiting a similarity between various received anonymized contextual data sets that is preserved by said similarity preserving anonymization transformation.
3. The method of claim 1 , wherein said similarity preserving anonymization transformation preserves said similarity in a probabilistic sense.
4. The method of claim 1 , wherein said similarity preserving anonymization transformation preserves said similarity in a deterministic sense.
5. The method of claim 1 , further comprising the step of learning one or more user models for said user using at least some of said anonymized collected one or more sets of contextual data.
6. The method of claim 5 , wherein the step of analyzing the received anonymized contextual data sets for authenticating the user comprises comparing at least some of said received anonymized contextual data sets with said learnt one or more user models.
7. The method of claim 6 , further comprising generating a score that reflects a degree that said at least some of said received anonymized contextual data sets match said learnt one or more user models.
8. The method of claim 5 , further comprising updating said learnt one or more user models.
9. The method of claim 8 , wherein said updating of said learnt one or more user models is performed at regular intervals.
10. A system to secure a user's interaction with a remotely accessible computer-based application, the system comprising:
a collection component adapted to collect one or more sets of contextual data;
an anonymization component adapted to anonymize the one or more sets of contextual data collected by the collection component;
an authentication server adapted to analyze the anonymized one or more sets of contextual data;
wherein the one or more sets of contextual data collected by the collection component comprise values of one or more computable contextual data elements; and
wherein the anonymization component is adapted to anonymize the collected one or more sets of contextual data by transforming a first value of at least one of said computable contextual data elements using a similarity preserving anonymization transformation.
11. The system of claim 10 , wherein the collection component is comprised in a personal client computing device associated with the user.
12. The system of claim 11 , wherein the collection component is adapted to receive values of at least some contextual data elements of said one or more sets of contextual data from one or more hardware sensors comprised in said personal client computing device.
13. The system of claim 11 , wherein the collection component is adapted to receive values of at least some contextual data elements of said one or more sets of contextual data from an operating system of said personal client computing device.
14. The system of claim 11 , wherein the anonymization component is also comprised in the personal client computing device.
15. The system of claim 11 , wherein at least the collection component is comprised in a software client application running on said personal client computing device.
16. The system of claim 10 , wherein said authentication server is adapted to analyze the anonymized contextual data for authenticating the user by exploiting a similarity between various received anonymized contextual data sets that is preserved by said similarity preserving anonymization transformation.
17. The system of claim 10 , wherein said similarity comprises a measure of distance.
18. The system of claim 10 , further comprising a user model learning component adapted to learn one or more user models for said user using at least some of said anonymized collected one or more sets of contextual data.
19. The system of claim 18 , wherein the authentication server is adapted to analyze the received anonymized contextual data sets for authenticating the user by comparing at least some of said received anonymized contextual data sets with said learnt one or more user models.
20. The system of claim 18 , wherein said user model learning component is further adapted to update said learnt one or more user models.
21. The system of claim 20 , wherein said user model learning component is further adapted to update said learnt one or more user models using at least some anonymized contextual data sets that have been received since the previous occurrence of said updating of said learnt one or more user models.