Acamedia - Research

AIDOS

Artificial Intelligence-based DDoS Mitigation at Internet Exchange Points

2021 - 2023

Securing Internet Exchanges with AI · Joint project with

Funded by BMBF

Abstract

Internetdienste sind immer häufiger Ziel von Angriffen zur Störung der Verfügbarkeit – sogenannten Distributed Denial of Service (DDoS)-Angriffen. Diese können für betroffene Unternehmen zu existenzbedrohenden monetären- und Imageschäden führen. Der Schutz gegen Angriffe solcher Art ist daher von steigender gesellschaftlicher und wirtschaftlicher Bedeutung. Die bisher zum Einsatz kommenden Schutzmaßnahmen sind jedoch mit hohen Kosten für die Infrastrukturbetreiber und Dienstanbieter verbunden oder können Angriffsverkehr nur unzureichend erkennen, was zu inakzeptablen Kollateralschäden führen kann.

Hier setzt das Projekt AIDOS an und entwickelt neue Verfahren zur effizienten Erkennung und Unterbindung von DDoS-Angriffen direkt im Kern des Internets: an Internetverkehrsknoten (Internet Exchange Points – IXP), die als Verteiler für einen Großteil des globalen Internetverkehrs dienen. Zur Erkennung von Angriffsmustern verfolgt AIDOS einen innovativen Ansatz, bei dem zur Analyse sämtlicher Internetverkehrsdaten der IXP auf jüngste und vielversprechende Fortschritte aus dem Feld der künstlichen Intelligenz zurückgegriffen wird. Auf diese Weise können Angriffe frühzeitig unterbunden und herkömmliche Gegenmaßnahmen unterstützt oder ersetzt werden.

Die in AIDOS entwickelten Verfahren können Angriffsverkehr schnell und zielsicher filtern und durch die Anwendung von maschinellen Lernverfahren eine kostengünstige und effiziente teilautomatische Erkennung von Angriffsmustern in großen Datenmengen durchführen. Die Anwendbarkeit in IXP wird durch die Erprobung am DE-CIX in Frankfurt am Main, der als größter IXP weltweit gilt, sichergestellt. Die Ergebnisse sollen es bestehenden Anbietern von Diensten zur DDoS-Bekämpfung und Internet Service Providern ermöglichen, einen Teil ihrer Dienstleistungen zum DDoS-Schutz an die IXP auszulagern und somit Kosten einzusparen und Angriffe besser identifizieren zu können. Gerade auch für kleinere Unternehmen ist der AIDOS-Ansatz interessant, da er günstige Schutzmaßnahmen für sonst schutzlose Angriffsziele bietet.

Chair of Computer Networks

LaBeCo

Language Behaviour in Context · A Transdisciplinary Approach

2018 - 2023

Research Initiative · Language Behaviour in Context · Consortium

Abstract
LaBeCo is a transdisciplinary research initiative based on the observation that language use - and more generally human behaviour - forms patterns through types of situations that reflect social, cultural and material factors. The object of study of all LaBeCo members means that the context of these types of situations must be included in their research in order to adequately explain variations in their data.
However, existing research approaches neglect the situation context, or analyse it exclusively within their own discipline, thus missing more comprehensive explanatory approaches by including findings from other disciplines. LaBeCo wants to close this gap with a multidisciplinary research approach.
Consortium
The LaBeCo research initiative extends across four faculties of RWTH Aachen University - in addition to the Faculty of Philosophy, the Faculty of Mathematics, Informatics and Natural Sciences, the Faculty of Electrical Engineering and Information Technology and the Faculty of Medicine - and also includes eight external partners from six other universities: the University of Cologne, the Brandenburg University of Technology, the University of Wuppertal, the Friedrich-Alexander University Erlangen-Nuremberg, the University of Salzburg and the University of Hildesheim.
Our Contribution
Making context computationally tangible (Transformer/Embeddings).

Chair of Computer Networks

PAM'21

Passive and Active Measurement Conference

March 2021

(Virtual) Conference Organization · 400 Registrations

  • Event Workflow · Registration / Planning / Platform / Operations

Abstract
The Passive and Active Measurement (PAM) conference brings together researchers and operators to discuss novel and emerging work in the area of network measurement and analysis. PAM is inclusive of all areas of network measurement, but focuses on systems-based research and real-world data. Indeed, measurement technology is needed at all layers of the network stack, ranging from power profiling of hardware components to virtualization in data centers to application profiling and even user experience. Work with operational impact or relevance to the broader network research community is especially welcome, as is early and promising measurement technique.

Chair of Computer Networks

COMTEX

Context in Anonymous Short Texts

2018 - 2019

Social Network Analysis · Joint project with

Funded by Excellence Initiative

Abstract
Today’s digital day-to-day communication is dominated by short text exchanges. Recently, anonymous communication platforms are emerging and have found widespread adoption. Anonymous communication largely lacks metadata and opens questions concerning efficient methods. This project brings together linguistic, sociological and computational expertise to explore a methodology for investigating social patterns of online communities as they manifest in linguistic behaviour with the long-term goal of developing computational mechanisms safeguarding against anti-social behaviour. The main outcome of the project will be a first understanding of the problems and possible solutions for determining context in anonymous short message exchanges. Ultimately, this knowledge of the social and linguistic patterns in online communities could be of use not just for retrospectively explaining (anti-)social behaviour in such communities but also proactively for designing platforms.

COMSYS

IoTRUST

Security of Consumer Electronic Devices

2018 - 2019

Security · Risk Assessment / Analysis · Joint project with

Funded by RWTÜV Foundation

COMSYS

SSICLOPS

Scalable and Secure Infrastructures for Cloud Operations

2015 - 2018

Federated Cloud Infrastructures
· Joint project across Europe

Funded by Horizon 2020

Abstract
The SSICLOPS project will focus on techniques for the management of federated private cloud infrastructures, in particular cloud networking techniques (within software-defined data centres and across wide-area networks). Key deliverables from the project will include a meta data description language for workloads, resources and policies, a flexible scheduling system using meta data, workload-specific adaptations to TCP/IP stacks, and data center performance analysis tools.

Addressing topics, such as dynamic configuration, automated provisioning and orchestration of cloud resources the SSICLOPS projects will investigate high-performance, vertically integrated network stacks for intra/inter-cloud communication and efficient, scalable, and secure intra/inter-DC and client-facing transport mechanisms.

The project will design, implement, demonstrate, and evaluate three specific use cases, namely a cloud-based in-memory database, the analysis of physics experiment data, and the prototypical extension of network stacks for a telecom provider in the SSICLOPS testbed.

COMSYS

Infrastructure

Enabling · Measurements / Evaluation / Teaching

2015 - 2023

Onprem Secure Infrastructure Management

  • Multi- User / Server Large Scale Computing Platform
  • Requirements · Processes · Sourcing · Setup
    • Network / CPU / GPU / RAM / Local Storage
    • Large Scale Storage / Snapshotting / Backup
    • -empowerd Server · Configuration / Security
  • DevOps · Pushing the limits
    • Relational Data Warehousing
    • Generic Parallelized Compute
    • Tensor-based GPU Compute
    • Students Playground
    • Teaching Services · Mini-Internet / Latency Measurement

Chair of Computer Networks · COMSYS

Earlier

Ticketing System

Migration · Ticketing System

2014 - 2015

Change Management

  • Multiclient Large-Scale Pilot Test
  • Tailored International Process · Design / Implementation
  • Monitoring · KPIs

ALDI SÜD

Master Data Management

Buying

2013 - 2015

Master Data Management Suite

  • Requirements Management · KPIs
  • Requirements Engineering · Solution Design
  • Data · Migration / Interdependencies
  • Advanced Reporting
  • Technical Consulting
  • Joint Taskforce Expansion

ALDI SÜD

... more

Jens Helge Reelfs Photo

Address
Jens Helge Reelfs
Email
spam AT helge DOT one
Networking
LinkedIn
Xing
ResearchGate
Twitter
Publications
DBLP
Google Scholar