Bochum 2018 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

HK: Fachverband Physik der Hadronen und Kerne

HK 52: Poster

HK 52.67: Poster

Donnerstag, 1. März 2018, 16:30–18:45, Audimax Foyer

Deep Learning and Isolation Based Security for Grid Computing — •Andres Gomez Ramirez for the ALICE collaboration — Infrastruktur und Rechnersysteme in der Informationsverarbeitung (IRI), Goethe Universität Frankfurt

The ALICE Grid, powered by AliEn, is a remarkable example of High-Troughput Computing (HTC) distributed infrastructure used in High Energy Physics. Due to their large size, complexity, reputation and good network connectivity, including access from general Internet, Grid systems are continuously exposed to attackers. Authenticated users have the freedom to execute arbitrary code and to transfer arbitrary data that is required for their work. External or insider attackers may take advantage of the Grid infrastructure to carry out unauthorized activities, for instance to run malware. The Grid is a heterogeneous and dynamic environment where it is difficult to adapt traditional rule based Intrusion Detection Systems. Distributed HTC systems require innovative methods and tools to identify cybersecurity incidents and perform autonomous actions. They also require methods to isolate and trace job payload activity in order to protect users and find evidence of malicious behavior. We introduce an integrated approach of Security by Isolation with Linux Containers and Deep Learning methods for the analysis of real time monitoring data in Grid jobs running on a virtualized HTC infrastructure in order to detect intrusions. A dataset for malware detection in Grid computing is collected from a controlled environment. We show in addition the utilization of generative methods with Recurrent Neural Networks to improve the collected dataset. We present a prototype implementation of the proposed methods. We empirically study the performance and accuracy of our proposed methods.