Real-Time Transmission with Time-Sensitive Networking

Topics

Time-Sensitive Networking (TSN) deals with the implementation and use of mechanisms for deterministic transmission in Ethernet networks. This means that fixed guarantees can be met by reserving network resources or precisely scheduling the transmissions. This includes 0% packet loss from congestion, as well as fixed upper bounds for maximum latency and latency variation (jitter).

These technologies are intended to replace bus systems (CAN, PROFIBUS, ...) with a low bandwidth, and to unify the many existing technologies (AFDX, PROFINET, EtherCAT, ...). Prominent use cases for TSN include industrial automation (Industry 4.0) and in-vehicle networks, which link sensors, actuators and logic controllers. The standardization of the shapers and schedulers has largely been completed, but their application and configuration is still subject of current research.

This seminar should provide an overview of the current state of the art, describe the most important challenges and possible solutions, and also explain some connections. At the beginning of the seminar there will be an introductory event with a brief overview and initial contexts for orientation.

The website of the TSN working group  and this overview article  provide general information for most topics. Some more specific sources are provided for some topics. Participants are expected to conduct further research on their own.

0. Historic Overview: Was was used before TSN? FlexRay, CAN, PROFINET, and their Problems (Betreuer: Alexej Grigorjew)
   • Which mechanisms (Ethernet, Bus-Systeme) were used before TSN? What were their core principles?
   • What are the general requirements of real-time applications, for example in the automotive and industrial manufacturing sectors?
   • How does real-time transmission work in these systems? What are the environments and their conditions? What assumptions can be made?
1. Latency Guarantees in Ethernet - Overview and Challenges (Adviser: Alexej Grigorjew)
   • What are the goals of TSN switches compared to standard Ethernet? Which scenarios are the requirements based on?
   • Computation of latency bounds: Central and distributed approaches, challenges involved
   • Differences between global and local information
2. Latency Computation in Static Networks (Adviser: Alexej Grigorjew)
   • General: Classification, requirements, limits, possibilities for optimization
   • Basic strategies to compute latency bounds in static networks
   • Examples
3. Synchronous Mechanisms: TAS and CQF (Adviser: Nicholas Gray)
   • Explanation of the mechanisms Time-Aware Shaping and Cyclic Queuing and Forwarding
   • Differences in the mechanisms and their application, examples
   • Discussion: optimization potential, complexity, dependencies
4. Asynchronous Mechanisms: CBSA and ATS (Adviser: Florian Metzger)
   • Explanation of the Credit-Based Shaping and Asynchronous Traffic Shaping mechanisms
   • Differences in the mechanisms and in the application, examples
   • Discussion: flexibility, hardware implementation effort, jitter
5. TSN Control Plane and Complementary Mechanisms (Adviser: Stefan Geißler)
   • TSN configuration and reservation process: protocol, information, requirements
   • Other mechanisms: filters and frame preemption
   • Discussion: reservation of resources vs. granularity of hardware
6. Redundancy and Frame Replication and Elimination for Reliablity (Adviser: Nicholas Gray)
   • Purpose of additional redundancy despite reservation
   • Different types of redundancy
   • FRER standard: Frame Replication and Elimination for Reliability
7. SDN and TSN (Adviser: Stefan Geißler)
   • Explanation of the concept of Software-Defined Networking
   • Common architecture and synergies of the concepts
   • Opportunity for rapid prototyping
8. Looking beyond Layer 2 (Adviser: Florian Metzger)
   • Real-time guarantees across multiple networks, interoperability
   • IETF DetNet group for the standardization of layer 3
   • Outlook, further mechanisms and possible use cases: 5G, carrier networks, virtual controllers in the cloud

Notes

Despite the current limitations of the chair's daily work, the seminar will take place in the summer semester. You will find the schedule and current dates in the corresponding WueCampus course, to which you will be invited by your supervisor. 

The seminar will be held as a block seminar. The actual dates for presentation will be announced during the semester in WueCampus and via e-mail.

The seminar participants must contact their supervisor in a timely and independent manner to discuss the topic with him or her. We attach great importance to a professional talk. This includes both the content and the way of presenting.

The participants have to familiarize themselves with their topic and prepare a didactically well-prepared talk of 25 to 30 minutes. The material is provided by the supervisor. He*she is available for questions. Further study of literature is often expected, depending on the topic. The written elaboration of the topic is compulsory according to the examination regulations.

Two weeks before the official presentation date, we request a rehearsal presentation to practice and improve the talk. This one is mandatory. Some supervisors have business trips and other appointments, so they are time-bound. Therefore, the appointment must be coordinated with the supervisor in advance. We expect the participants to prepare for the rehearsal presentation and practice their talk several times with fellow students beforehand. Any feedback from fellow students should be incorporated.

Participants who have not prepared an appropriate final presentation in time are moved to the next semester. Unfortunately, it is not possible to postpone the presentation dates at short notice. Being present at all presentations of the seminar is compulsory for all participants. The presence is monitored.