Description

Background:
Time-sensitive networks are used for safety-critical cyber-physical systems (CPSs) in
vehicles, planes, satellites or power plants. Their significance has been increasing over
the years and they are now used in many more applications, ranging from autonomous
cars, automated manufactures (industry 4.0) to 5G and beyond backbone networks. While
traditional public networks aim at improving the mean service performances (mean
round trip time, mean throughput), time-sensitive networks provide guarantees for the
worst case (e.g. guarantee of a maximal latency, guarantee of no loss, …). Time-sensitive
networks use specific layer-2 technologies from IEEE TSN [TSN] for providing
deterministic latency.
Among these technologies are traffic regulators such as Aynchronous Traffic Shaping.
Traffic regulators force the traffic to conform to a given specification, delaying the packets
if required. They are particularly useful as they remove the burst-cascade effect (the
burstiness of the flows tend to increase along their paths).
Time-sensitive networks control increasingly large and dynamic systems (smart-grid
systems, unmanned air traffic management, public transportation systems). In these
large-scale and often multi-actors networks, traffic regulators are required at each point
that crosses the frontier between two network domains. Yet, the traffic regulators
currently available do not scale to large-scale networks with a large number of different
flows.

Project description:
The goal of the PhD is to investigate, design and evaluate new traffic regulation
mechanisms that scale to a large number of flows and a large traffic rate and to be
deployed at the frontier of several network domains in a large-scale time-sensitive
network.
Among the possible research directions are:
– Investigate approaches for aggregating multiple flows into a compound flow,
where only the compound flow is regulated.
– Investigate new regulation techniques that do not require storing a per-flow state
in memory, possibly by having regulation data written in the packet’s header,
implement them on an embedded system to check their scalability.
– Investigate the efficient and scalable configuration of traffic regulators in largescale
multi-domains networks.

Compétences requises

· Very good programming skills. · Knowledge of computer networks (layered approach, etc). · Interest in time-sensitive systems is a plus. · Knowledge on embedded systems is a plus. · Knowledge on FGPA, VHDL, DPDK or P4 is a plus.

Bibliographie

[TSN] “Time-Sensitive Networking (TSN) Task Group |.” Accessed: Nov. 06, 2022.
Available: https://1.ieee802.org/tsn/
[LeBoudec01] J.-Y. Le Boudec and P. Thiran, Network Calculus, vol. 2050. in Lecture Notes in Computer Science, vol. 2050. Berlin, Heidelberg: Springer, 2001. doi: 10.1007/3-540-45318-0.
[Geyer22] F. Geyer and S. Bondorf, “Network Synthesis under Delay Constraints: The Power of Network Calculus Differentiability,” in IEEE INFOCOM 2022 - IEEE Conference on Computer Communications, May 2022, pp. 1539–1548. doi: 10.1109/INFOCOM48880.2022.9796777.
[Tsai24] C.-T. Tsai, S. M. Tabatabaee, S. Plassart, and J.-Y. Le Boudec, “Saihu: A common interface of worst-case delay analysis tools for time-sensitive networks,” SoftwareX, vol. 27, p. 101882, Sept. 2024, doi: 10.1016/j.softx.2024.101882.

Mots clés

Time-Sensitive Networking (TSN), Deterministic Networking (DetNet), Traffic Shaping, Traffic Policing, Network Calculus