1. 1

[..] In this talk, we present several methods that make the large scale security analyses of embedded devices a feasible task. We implemented those techniques in a scalable framework that we tested on real world data. First, we collected a large number of firmware images from Internet repositories and then performed simple static analysis. Second, since embedded devices often expose web interfaces for remote administration, therefore we developed techniques for large scale static and dynamic analysis of such interfaces. Finally, identifying and classifying the firmware files, as well as fingerprinting and identifying embedded devices is difficult, especially at large scale. Using these techniques, we were able to discover a large number of new vulnerabilities in dozens of firmware packages, affecting a great variety of vendors and device classes. We were also able to achieve high accuracy in fingerprinting and classification of both firmware images and live devices.