UT researcher develops algorithm for checking software running in parallel

Tri Minh Ngo, researcher at the University of Twente, has developed an algorithm with which software running in parallel can be checked for possible leaks of sensitive data. This makes it easier to detect leaks in programs.

With parallel programmed software, where simultaneous calculations are performed, it can be a challenge to find out whether the various parallel tasks are not leaking information. For example, an individual user should not be able to deduce information about private data by looking at the public data. Marieke Huisman, associate professor at the University of Twente, provides payroll software as an example. Normally the salary data of individual employees is private, but based on other information, such as the number of employees and the total amount of salaries, something could be said about the salaries of individual employees.

Researcher Ngo has developed an algorithm that would make it easier to check for leaks. For example, his software could indicate how information is leaked and an estimate could be made of how much information is leaked. Based on this, two different software implementations can be compared. Ultimately, Ngo’s algorithm could make parallel software more secure. On April 17, Ngo will receive his PhD within the Formal Methods and Tools department of the CTIT research institute at the University of Twente for his dissertation ‘Quantitative and qualitative information Flow Analysis for multi-threaded programs’.