Aggregation of sensitive data

Introduction

When Diffie and Hellman introduced public-key cryptography in the mid-nineteen seventies, it was clear that beyond its mathematical interest, it would have a huge effect on real-world data processing. This intuition was largely proven to be true. As information flows in networks, the security of the data deeply affects the trust relationship between the communicating participants. For example, online payment would not exist if the buyer did not trust that its data are correctly secured. It is not exaggerating to state that cryptography is a key ingredient of the modern information society.

These issues became more and more important since we have realized the value of data. Collecting data securely requires careful application of cryptographic techniques. But data owners also want to be able to capitalize on them, and computing over data while maintaining privacy requires techniques that are at the forefront of modern cryptographic research.

Issues arise even from simple problems such as extracting statistics from distributed data. The data might be too valuable or regulated in a way that prevents sending them directly to a third party.

Classical cryptography offers solutions to perform these operations. In particular, secure multiparty computing allows mistrustful parties to compute over their inputs while maintaining privacy. Some secure computing solutions also involve third parties but the theory ensures that they will not get any information during the process.

Participatory trust and delegated trust

The two main security models considered for secure computing are participatory trust and delegated trust. In the participatory trust model, data owners perform a collective computation which is private by-design. The participants thus bear the responsibility for the privacy. In the delegated trust model, a third party aggregates the data from various participants and runs the computation. He is responsible for the privacy of the process. Various security issues arise from these situations. In the delegated trust model, the first step is to centralize the data. This requires to secure data in-transit as well as stored data. The second step is to compute over the securely stored data. In the participatory trust model, the protocol executed by the participants should be private by-design.

In both cases, quantum networks can increase security. In the example developed above, the aggregation is performed on healthcare data. These are very sensitive data that require carefully designed security. In particular, the long-term security that is inherent to quantum cryptography could strengthen the security of communication and storage in the case of delegated trust, and privacy by-design in the case of participatory trust.

Using anonymous transmission

In some cases, the only information that needs to remain private is the sender’s identity. For example, monitoring car traffic can lead to a better management. Drivers, however, might not be willing to share their speed to avoid being caught over the speed limit for example. Quantum anonymous transmission could be used to hide the drivers’ identities while collecting valuable data.

Cryptography can be used to make mistrustful parties collaborate to reach a common goal. While the amount of data is increasing, the responsibility of data’s owners is increasing as well. Quantum networks could help making better use of data, without sacrificing privacy and security.