Computer Networks and Distributed Systems

Image
slide_expl_vcampus3_ohne.jpg
Group leader
Foto Jürgen Schönwälder
Computer Networks and Distributed Systems
Specific themes and goals
  • Standards: The Computer Networks and Distributed Systems (CNDS) group has been involved in the development of international standards for more than 15 years. These standards enable robust automation of computer network operations. This includes standards that support the configuration of large, decentralized collections of networking devices as well as standards that enable the efficient monitoring of networking infrastructure. 
  • Measurement systems: Recently, we have focused on the development of standards for large-scale distributed measurement systems that can provide a global view on Internet performance. Since the Internet is essentially a large collection of networks that are all under different administrative control (also known as autonomous systems), regulators are interested in unbiased independent measurements that describe network performance from a large collection of independent and decentralized vantage points. Such large-scale measurement activities are important to understand how technical changes, that are deployed incrementally, impact network performance and ultimately the experience delivered to users and businesses. 
  • Resilient, secure technologies: Our overall goal is to develop technologies that help to provide resilient and secure services over the Internet even in the face of failures and attacks on the infrastructure. Furthermore, we are interested in novel technologies that can bolster the security of computing elements within complex distributed systems. Increasingly, our research focuses on device, network, and system security aspects, with a special focus on secure embedded systems.
Highlights and impact
  • We worked on an architectural framework, which introduced an extensible set of configuration datastores for network protocols that allow for the automatic changing of device configuration. These datastores can, for example, differentiate between the intended configuration of a device and the currently applied configuration of a device. Exposing these differences is necessary to monitor device behavior and to implement robust automated control loops. In 2019, we completed this work, which we performed in close collaboration with industrial partners. We published two more proposed standards that support the new architecture by defining extensions of the NETCONF and RESTCONF protocols.
  • The OpenWRT operating system is a Linux-based operating system for embedded devices with a very small memory footprint. It is popular on networking devices such as access points or smallscale home routers. We designed a RESTCONF implementation that specifically targets the OpenWRT system. A paper describing the design of our implementation received the best experience paper award at IEEE/IFIP NOMS 2020. 
  • We measured how the deployment of the Internet Protocol version 6 (IPv6) affects Internet services delivered to end users. In 2019, we published a summary of our findings on how the transition affects web sites accessible by both the old and the new version of the Internet protocol. 
  • We investigated how metamorphic testing can be applied to homomorphic cryptography. Homomorphic encryption schemes allow us to perform computations on encrypted data without revealing the data. Metamorphic testing is a property-based software testing strategy that does not require oracles to predict expected test outcomes. By applying metamorphic testing to an open-source homomorphic cryptography library, we were able to reveal several shortcomings. 
  • With an increased interest in security, we started to investigate active malware analysis techniques. We executed malware samples in a sandbox environment to detect malicious behavior. To be effective, we had to stimulate the malware samples since specific user action or network input is often required to trigger malicious behavior. A challenge in active malware analysis techniques is to generate suitable inputs that are likely to trigger malicious behavior. We are investigating learning techniques to generate these inputs.
Group composition & projects/funding

The research of the CNDS group has been largely funded by EU grants. The EU project CONCORDIA (2019-2023) provides funding for a PhD student as well as support for external PhD students and collaborators of associated research groups.

Selected publications
  • V. Bajpai, J. Schönwälder: A Longitudinal View of Dual-stacked Websites: Failures, Latency and Happy Eyeballs. IEEE Transactions on Networking 27(2), April 2019 
  • M. Björklund, J. Schönwälder, P. Shafer, K. Watsen, R. Wilton: RESTCONF Extensions to Support the Network Management Datastore Architecture. RFC 8527, March 2019 
  • M. Granderath, J. Schönwälder: A Resource Efficient Implementation of the RESTCONF Protocol for OpenWrt Systems. 17th IEEE/IFIP Network Operations and Management Symposium (NOMS 2020), April 2020 
  • M. Wolf, J. Schönwälder: Applying Metamorphic Testing to Homomorphic Cryptography. 6th IEEE/ACM International Workshop on Metamorphic Testing (MET 2021), June 2021 
  • A. Hota, J. Schönwälder: A Bayesian Model Combination based approach to Active Malware Analysis. IEEE Conference on Cyber Security and Resilience Workshop on Data Science for Cyber Security, 2022.