Futureproofing Computer Security

Living our lives online is not without risk – issues like identity theft and banking fraud have affected many of us as we share more and more of our personal data over the internet. Computer systems need advanced mathematical tools to disguise or encrypt our information and keep it safe from prying eyes. We’re fortunate that expert cryptographers, like Dr Lisa Kohl at Centrum Wiskunde & Informatica (CWI) – the national research institute for mathematics and computer science in the Netherlands – are searching for innovative ways to keep us secure online. She says, “Cryptography is the perfect way to combine my passion for mathematics with my desire to contribute to a bigger purpose and do something useful. There are many questions that are yet to be solved, and it’s a very exciting research area.” 

For Lisa, CWI is the perfect environment to carry out her research. She completed a master’s degree there in 2015 and loved the experience so much, she was happy to return in 2020 to take up a research position in the Cryptology Group. The support and encouragement offered to researchers for their career development is something she really values. Lisa’s work focuses on secure computation which is how to compute data securely while preserving individual privacy. She’s interested in methods that give long-term security and will still be useful when quantum computers, the advanced supercomputers of the future, become a reality. And it’s not just enough to design computer algorithms to communicate data securely, in our modern world, efficiency is of prime importance too.

Secure multiparty computation is central to Lisa’s research. This is where you have several parties who want to share data without revealing any private information. For example, a hospital needs to protect patient data, but also wants to share information with researchers for a study, or a bank wants to use its data to identify money-laundering activities. If the data was shared in a nonsecure way, each party would analyse the data locally on their own computer or network. The complexity of the computation would depend on the size of the data input. “But with secure computation, there’s a huge overhead [extra computing resources needed] – a huge increase in communication,” Lisa says. “The input and the complexity of the programme you want to compute can also affect this, and this is what really kills secure computation for a lot of applications. A particular focus of my work is how we can get secure computation that’s still very communication efficient.”

Recently Lisa and her colleagues have had some exciting results. They’ve been working on a ‘preprocessing phase’ where some of the complex analysis needed for secure computation is carried out offline ahead of time. Then, when data is exchanged in the online phase, the computation can be carried out very quickly. In the past, a major drawback of this preprocessing phase was that it required a lot of computational time and storage, as part of the process involves generating many random multiplications. These random multiplications prevent adversaries from learning the other parties’ private data. Now, Lisa and her colleagues have developed ‘pseudorandom correlation generators’, which have made this process much more efficient and improved this bottleneck. She explains, “Depending on the complexity of your programme, you might need several gigabytes of storage. But with pseudorandom correlation generators, you can really compress this by a factor of 1,000 or more, so you need much less communication and less storage to set it up.” Since network speed is often a bottleneck, this leads to significant speed-ups in practice.

CWI’s excellent reputation was a big draw for Lisa when she was considering her career options. Its location in the Science Park in Amsterdam was an added bonus as it’s easy to attract great candidates when you’re building a research group to come and work there. There are also a lot of fun events organised for PhD students and opportunities to network with other researchers from the institute. She appreciates that interdisciplinary research is encouraged so you can get different perspectives on your research problems. “You have a lot of communication at CWI - much more than I have seen elsewhere,” she says. “And it’s a very friendly atmosphere to work in, very welcoming and helpful. At CWI, there’s a willingness to explore areas that could lead to something maybe not tomorrow, but in 10 or 20 years’ time, and I like this kind of openness. I really could not imagine a better place to be.”