In time-sensitive networks, bounds on worst-case delays are typically obtained by using network calculus and assuming that flows are constrained by bit-level arrival curves. However, in IEEE TSN or IETF DetNet, source flows are constrained on the number of packets rather than bits. A common approach to obtain a delay bound is to derive a bit-level arrival curve from a packet-level arrival curve. However, such a method is not tight: we show that better bounds can be obtained by directly exploiting the arrival curves expressed at the packet level. Our analysis method also obtains better bounds when flows are constrained with g-regulation, such as the recently proposed Length-Rate Quotient rule. It can also be used to generalize some recently proposed network-calculus delay-bounds for a service curve element with known transmission rate.

Time-sensitive networks (IEEE TSN or IETF DetNet) may tolerate some packet reordering. Re-sequencing buffers are then used to provide in-order delivery, the parameters of which (timeout, buffer size) may affect worst-case delay and delay jitter. There is so far no precise understanding of per-flow reordering metrics nor of the dimensioning of re-sequencing buffers in order to provide worst-case guarantees, as required in such networks. First, we show that a previously proposed per-flow metric, reordering late time offset (RTO), determines the timeout value. If the network is lossless, another previously defined metric, the reordering byte offset (RBO), determines the required buffer. If packet losses cannot be ignored, the required buffer may be larger than RBO, and depends on jitter, an arrival curve of the flow at its source, and the timeout. Then we develop a calculus to compute the RTO for a flow path; the method uses a novel relation with jitter and arrival curve, together with a decomposition of the path into non order-preserving and order-preserving elements. We also analyse the effect of re-sequencing buffers on worst-case delay, jitter and propagation of arrival curves. We show in particular that, in a lossless (but non order-preserving) network, re-sequencing is “for free”, namely, it does not increase worst-case delay nor jitter, whereas in a lossy network, re-sequencing increases the worst-case delay and jitter. We apply the analysis to evaluate the performance impact of placing re-sequencing buffers at intermediate points and illustrate the results on two industrial test cases.

Dampers are devices that reduce delay jitter in the context of time-sensitive networks, by delaying packets for the amount written in packet headers. Jitter reduction is required by some real-time applications; beyond this, dampers have the potential to solve the burstiness cascade problem of deterministic networks in a scalable way, as they can be stateless. Dampers exist in several variants: some apply only to earliest-deadline-first schedulers, whereas others can be associated with any packet schedulers; some enforce FIFO ordering whereas some others do not. Existing analyses of dampers are specific to some implementations and some network configurations; also, they assume ideal, non-realistic clocks. In this paper, we provide a taxonomy of all existing dampers in general network settings and analyze their timing properties in presence of non-ideal clocks. In particular, we give formulas for computing residual jitter bounds of networks with dampers of any kind. We show that non-FIFO dampers may cause reordering due to clock non-idealities and that the combination of FIFO dampers with non-FIFO network elements may very negatively affect the performance bounds. Our results can be used to analyze timing properties and burstiness increase in any time-sensitive network, as we illustrate on an industrial case-study.