Evaluating arbitrary functions on encrypted data is one of theholy grails of cryptography, with Fully Homomorphic Encryption (FHE) beingprobably the most prominent and powerful example. FHE, in its current stateis, however, not ecient enough for practical applications. On the other hand,simple homomorphic and somewhat homomorphic approaches are not powerfulenough, to support arbitrary computations.We propose a new approach towards a practicable system for evaluating func-tions on encrypted data. Our approach allows to chain an arbitrary number ofcomputations, which makes it more powerful than existing ecient schemes. Togain an eciency advantage over FHE we do not encrypt or in any way hide thefunction, that is evaluated on the encrypted data. It is, however, sucient thatthe function description is known only to the evaluator. This situation arises inpractice for Software as a Service (SaaS)-scenarios, where an evaluator providesa function only known to him and the user wants to protect his data. Anotherapplication might be the analysis of sensitive data, such as medical records.In this paper we restrict ourselves to functions with only one input parameter,which allow arbitrary transformations on encrypted data.
@article{392,
title = {Oblivious lookup-tables},
journal = {Tatra Mountains Mathematical Publications},
volume = {65},
year = {2016},
doi = {10.2478/tatra.v67i0.392},
language = {EN},
url = {http://dml.mathdoc.fr/item/392}
}
Wamser, Markus Stefan; Rass, Stefan; Schartner, Peter. Oblivious lookup-tables. Tatra Mountains Mathematical Publications, Tome 65 (2016) . doi : 10.2478/tatra.v67i0.392. http://gdmltest.u-ga.fr/item/392/