Section: New Results

Private Information Retrieval

Imagine the following scenario, in which a researcher wants to access many sustrings a DNA sequences, while maintaining the privacy of the request. The privacy or the secrecy of the database is not a concern here: for instance, this researcher wants to access many DNA subsequences of drosophila melanogaster, hosted on a remote data broker, and clearly the concern is not to protect the private life of flies. But the information leaked about the queries may endanger the novel aspect of the discovery the researcher is about to make, by revealing which DNA sequences he is studying.

Private Information Retrieval (PIR) schemes are designed to achieve this goal: a user queries a database $T$ hosted on a remote server, and wants the $i$-th entry, i.e. $T\left[i\right]$. A cryptographic protocol is then run, and at the end of the protocol, the server must not know $i$, neither the $T\left[i\right]$ he answered, yet the user gets $T\left[i\right]$.

These PIR schemes can be achieved in an unconditionally secure way using the above Multiplicity codes, which N. Coxon made practical. In September, we explained this scenario and demoed our software at Nokia Bell Lab's Future X days a use case of Multiplicity codes for private access to DNA sequences.