Latency- And Resilience-aware Networking (LARN)

This project is part of the DFG (Deutsche Forschungsgemeinschaft) priority programme SPP 1914 "Cyber-Physical Networking". It started in October 2016 and is carried out in cooperation with Department of Computer Science 4 at Friedrich-Alexander-Universität Erlangen-Nürnberg.

Vision

We propose to develop transport channels for cyber-physical networks. Such channels need to be latency- and resilience-aware; i.e. the latency as seen by the application must be predictable and in certain limits, e.g. by balancing latency and resilience, be guaranteed. This is only possible by an innovative transport protocol stack and an appropriate fundament of operating system and low level networking support. Thereto the project unites the disciplines Operating Systems / Real-Time Processing and Telecommunications / Information-Theory.

Reliable Networking Atom (RNA) Architecture

Project target is the evolution of the PRRT (predictably reliable real-time transport) transport protocol stack towards a highly efficient multi-hop-protocol with loss domain separation. This is enabled by an interdisciplinary co-development with a latency-aware operating system kernel incl. wait-free synchronisation and the corresponding low level networking components (POSE, "predictable operating system executive"). The statistical properties of the entire system (RNA, "reliable networking atom") shall be optimised and documented.

A software-defined networking testbed for validation of the system in a real-world wide area network scenario is available. The developed components will be introduced during the workshops organised by the priority programme Cyber-physical Networking and will be made available to other projects during the entire run-time of the priority programme.

Objectives

  • Develop a latency- and resilience-aware atomic unit for cyber-physical networks.
  • Optimise the cross-layer communication while maintaining separability.
  • Validate the approach in theory and experiments.
  • Document the statistics of the predictable quality of communication (QoC).

Activities

Publications

2018

  • Reif, Stefan; Schmidt, Andreas; Hönig, Timo; Herfet, Thorsten; Schröder-Preikschat, Wolfgang: "∆elta: Differential Energy-Efficiency, Latency, and Timing Analysis for Real-Time Networks", 16th International Workshop on Real-Time Networks (ECRTS RTN), Barcelona, Spain, July 2018
  • Gil Pereira, Pablo; Schmidt, Andreas; Herfet, Thorsten: "Cross-Layer Effects on Training Neural Algorithms for Video Streaming" (arxiv), 28th ACM SIGMM Workshop on Network and Operating Systems Support for Digital Audio and Video (NOSSDAV), Amsterdam, Netherlands, June 2018
  • Reif, Stefan; Schröder-Preikschat, Wolfgang: "A Predictable Synchronisation Algorithm (Poster)", 23rd Annual Symposium on Principles and Practice of Parallel Programming (PPoPP), Vienna, Austria, February 2018

2017

2016

Student Projects & Theses

  • Lukas Bard, "Network Protocol for Video Transport in Cyber-Physical Systems", Master Thesis, November 2017
  • Daniel Birtel, "Transparent Transmission Segmentation for Multimedia Applications", Master Thesis, January 2017
  • Philipp Tennigkeit, "NEAT - Network Experiment Automation Tool", Project
  • Kai Vogelgesang, "High-Performance Network Functions with DPDK", Project
  • Marlene Böhmer, "Demonstration and Evaluation Platforms for Cyber-Physical Networking", Project
  • Ashkan Taslimi, "Hybrid Error Coding for Cyber-Physical Networking", Project

People and Partners

Timo Hönig (FAU), Adwait Datar (TUHH), Andreas Schmidt (SIC), Florian Rosenthal (KIT), Stefan Reif (FAU), Thorsten Herfet (SIC) at the "Hands-On Cyber-Physical Networks" BarCamp of the DFG SPP1914 Winter School, January 2018

 

BarCamp "Hands-On Cyber-Physical Networking" Partners

  • Adwait Datar, MSc. (TUHH)
  • Christine Kloock, MSc. (TUHH)
  • Florian Rosenthal, MSc. (KIT)