The use case describes the outsourcing of major updates on biometric databases. This use case is typical of biometric systems. A major accuracy update requires reprocessing the raw images to enable the new algorithms. This process usually takes time (e.g., several months) and requires in-house hardware. Thus, the perspective delegation of such computations to the cloud seems appealing. However, for privacy reasons, outsourced biometric data should be encrypted. Therefore, the use-case raises the issue of applying update algorithms over encrypted biometric data.
The storage of biometric data is at the core of biometric systems. Managing big identity databases composed of millions of records is by itself cumbersome, but a lot of technical and privacy concerns are added when databases store biometric data. In particular, privacy concerns often preclude outsourcing computations on biometric data. The current practice is to not outsource the computation over biometric data at all. On the other side, encrypting biometric data supplies data privacy, but precludes the ability to compute over the data. As a result, efficient solutions for delegating processing over encrypted biometric data would supply clear advantages over current solutions.
Algorithms using biometric data regularly evolve, as well as the formats under which the biometric data are stored. As a result, biometric systems sometimes need major upgrades, meaning that the stored biometric data must be processed in order to be compatible with the new formats and algorithms. Biometric data cannot be processed by an external cloud system for privacy reasons. As a result, the current solution prohibits the outsourcing of system upgrades. The use-case introduces a model that allows outsourcing the processing of biometrics data. According to this model, a pre/post-processing entity is deployed in a private cloud environment and a cloud update server is deployed with a public cloud provider. The pre/post-processing entity lies between the biometric database and the cloud, ensuring first the outsourcing of the system upgrades, and then the integration of the result supplied by the cloud.
Expected Outcomes and Contribution of TREDISEC
In order to outsource the computation over sensitive biometric data, MORPHO expects that the outcome of TREDISEC will provide privacy-preserving primitives for processing biometric data. More specifically, the encryption primitives should be compatible (and efficient) with the signal processing operations to be carried out on the raw biometric images. Additionally, if the cloud server were able to prove that the process over encrypted biometric data has been correctly performed, it would ensure that the biometric data are correctly updated once integrated in the biometric system. From a business perspective, such solutions provided, built upon the technologies brought by TREDISEC, would significantly decrease the overall time and cost of biometric systems upgrades.