Abstract:

It is well known that if you want to prevent aliens (or just other people) from reading your thoughts, you can simply wear a tinfoil hat. However, as you walk down Strøget in downtown Aarhus proudly sporting your tinfoil hat, what and whom you are looking at can still reveal lot about what is in your head. You can try sunglasses, but now and then you will want to turn your head. Is there a way to fully “encrypt” where you are looking? 

Modern computers have similar problems: 

• Today data is often stored in the cloud. You can encrypt the individual files such that the cloud provider cannot read what is in the file. But if there is a particular file that you access often, the cloud provider still learns that fact. And the access pattern to data can reveal a lot about the data itself. 
• When a program is running on a computer it stores its data in memory, a memory which could be shared with other malicious programs like viruses. The program can encrypt the individual data items, but the access pattern to the data can still leak sensitive information to rough programs running on the same machine. 

Is there some way for a computer or program to “encrypt” its access pattern to the data it is working on? Yes there is! 

The problem of encrypting access patterns was proposed more than 30 years ago. Over the years a lot of increasingly efficient solutions have been proposed. Recently, however, the progress seems to have come up against a brick wall. In this talk I will present you some solutions to the Encrypted Glance problem. I will also dive into a recent result I discovered together with Kasper Green Larsen (Aarhus University), which proves that the reason why progress has grind to a halt is that there is no way to do better than the currently best known solutions. This is an example of what is called a lower bound theorem in theoretical computer science. 

This research won the best paper award at CRYPTO 2018, the most important international conference for cryptographic research.

Inaugural lecture:

1 April 2018 Jesper Buus Nielsen was appointed professor in cryptography and security at the Department of Computer Science. Jesper is one of the top cited and publishing researchers in secure multiparty computation. His primary research areas are implementing secure multiparty computation in practice, rational cryptography, leakage resilient cryptography, tamper resilient cryptography, and universal composability.

The lecture will be followed by an informal reception.

See the official invitation