CybersecurityHacking cybersecurity in order to anticipate attacks

Imagine two groups at war. One defends every attack as it comes. The other anticipates threats before they happen. Which is more likely to win?

In cybersecurity, understanding the potential for attacks is critical. This is especially true for mobile and wireless devices, since they are constantly connected and continuously streaming and collecting data.

“We have these devices with us all the time. We trust them with many things—with the microphone they can hear us, with the camera they can see everything, we put all our pictures there. Our lives are on these devices,” said Guevara Noubir, professor in the College of Computer and Information Science.

Noubir recently organized the 2017 Conference on Security and Privacy in Wireless and Mobile Networks, hosted at Northeastern in July. The conference included some of the foremost experts in mobile security. One of the keynote speakers, Silvio Micali, is a winner of the Turing award, which is widely recognized as the “Nobel Prize” of computer science.

Researchers and students from all over the world convened at the conference to discuss mobile and wireless security, an area of cybersecurity Noubir says is “booming.” They shared new research, held tutorials, and listened to some of the brightest minds in the field debate one of society’s most pressing challenges.

The importance of reproducibility
News@Northeastern notes that as the organizer of the conference, Noubir implemented a new feature this year—the reproducibility label. Reproducibility—the premise that all studies should be replicable—is at the core of all good science. But Noubir explains that in wireless research, studies can be especially difficult to replicate, even if researchers provide every detail of how they conducted an experiment. That’s because unbeknownst to them, there are often other factors at play. For example, if a car drives by as one device is measuring the activity of another, signals from the car’s computer system might subtly interfere with the measurement.

To check reproducibility, Noubir provided conference participants with a software program that employs a “virtual machine” that analyzes all the data, graphs, charts, and tables in a study and tries to reproduce them. Of the 26 papers submitted to the virtual machine, only six of them were granted the reproducibility label, which underscores how challenging wireless studies can be to replicate.