Dynamically Optimizing End-To-End Latency For Time-Triggered Networks

Zonghui Li,Hai Wan,Boxu Zhao,Yangdong Deng,Ming Gu
DOI: https://doi.org/10.1145/3341558.3342203
2019-01-01
Abstract:In time-triggered (TT) networks, frames for industrial control, so-called TT frames, are statically scheduled to be transmitted at precise sending points. Meanwhile, the remaining frames, so-called BE frames, are forwarded with best-effort (BE) strategies. TT frames meet application-specific end-to-end latency requirements by scheduling the worst cases. Furthermore, the planned sending points tend to prevent them from delivering as soon as possible. However, BE frames are transmitted as soon as possible but not assure bounded latency even lead to loss of frames due to queuing, congestion, etc.In this paper, we propose a BE assisted TT communication protocol (BEST TT protocol) to dynamically optimize end-to-end latency of TT frames by opportunistically taking advantage of BE transmission. Given a TT frame, the BEST TT protocol generates a cloned copy that is transmitted with the BE strategies. Whichever the TT frame or its copy, the first frame arriving at the end device is finally delivered and the other frame is ignored. So, the protocol ensures that the latency of the final delivered frame is not worse than that of the TT frame. We implement the protocol in an industry-strength TT switch and present three test scenarios to demonstrate its optimization compared with the previous state-of-art TT transmission.
What problem does this paper attempt to address?