Leakage Aware Design Automation (LADA): Tools & Techniques for Software Crypto Implementations

点击翻译按钮获取中文摘要

Information leakage via side channels is a widely recognised threat to cyber security: in particular small devices are known to leak information via physical channels (power consumption, electromagnetic radiation, and timing behaviour). Side channel leakage provides skilled adversaries with information about otherwise secret internal variables, which can ultimately lead to complete security breaches in the form of secret key (or data) recovery. For small embedded devices, which feature architectures of limited complexity (i.e. a small number of pipeline stages, few data and/or address buses), the nature of the leakages can be appropriately modelled using statistical tools such as regression analysis or by estimating (multivariate) normal distributions. Our research hypothesis is that one can make meaningful statements about the leakage behaviour of new implementations on such devices by utilising a priori derived (instruction level) leakage models. Aiming to allow engineers with limited domain-specific knowledge to do just this, and hence improve the quality of software they develop, our overarching goal is: based on the development of a suitable instruction-level leakage model (for a device) plus specification of cryptographic primitives, we will explore techniques that allow sound assessment of leakage-related attacks on associated implementations without the need for a fully equipped side channel lab.