RFC 9912
Reliable and Available Wireless (RAW) Architecture
Reliable and Available Wireless (RAW) Architecture 🎀
Okay bestie, this is RFC 9912, and it's giving wireless reliability glow-up 💖. P. Thubert, Ed. is the main author, and they're writing as Independent, so no company or group sponsored it. The category is Informational ✨, which means it's here to teach and explain, not to set a mandatory standard. It was published in April 2026 🌸, and its ISSN is 2070-1721. The header doesn't list any RFCs it obsoletes or updates, so it's standing on its own 🦋.
Abstract ✨
Reliable and Available Wireless (RAW) takes the reliability and availability goals of Deterministic Networking (DetNet) and brings them to networks that mix wired and wireless segments, in any combo 🔥. It builds on RFC 8655 ("Deterministic Networking Architecture") and adapts it for the wireless medium's main drama, which is intermittent transmission loss 😭. The document defines a network control loop that uses constrained bandwidth and energy wisely while still delivering the DetNet services everyone expects 💅. The loop includes a new Point of Local Repair (PLR) function in the DetNet Service sub-layer. This PLR picks the DetNet path(s) for packets so they can route around local connectivity degradation, like a main character taking the scenic detour 🦋🧚.
Status of This Memo ✨
Quick heads up: this document is not an Internet Standards Track spec. It's published purely for informational purposes 🌈.
It's still a big deal though! It's a product of the Internet Engineering Task Force (IETF) and represents the consensus of the IETF community 🫶. It went through public review and got approved for publication by the Internet Engineering Steering Group (IESG) 💜. Just know that not every IESG-approved document is a candidate for any level of Internet Standard. See Section 2 of RFC 7841 for the details 🙈.
If you want the latest status, any errata, or a way to send feedback, check https://www.rfc-editor.org/info/rfc9912 🍓.
Copyright Notice ✨
Copyright (c) 2026 IETF Trust and the persons identified as the document authors. All rights reserved 💖.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) as they're in effect on the date of publication. Please actually read these, because they spell out your rights and restrictions for this document 👀. Any Code Components pulled from this document must include the Revised BSD License text, as described in Section 4.e of the Trust Legal Provisions. They also come without warranty, as described in the Revised BSD License ⭐.
Table of Contents 🧚
1. Introduction
2. The RAW Problem
3. Terminology
3.1. Abbreviations
3.2. Link and Direction
3.2.1. Flapping
3.2.2. Uplink
3.2.3. Downlink
3.2.4. Downstream
3.2.5. Upstream
3.3. Path and Recovery Graphs
3.3.1. Path
3.3.2. Recovery Graph
3.3.3. Forward and Crossing
3.3.4. Protection Path
3.3.5. Segment
3.4. Deterministic Networking
3.4.1. The DetNet Planes
3.4.2. Flow
3.4.3. Residence Time
3.4.4. L3 Deterministic Flow Identifier
3.4.5. Time-Sensitive Networking
3.4.6. Lower-Layer API
3.5. Reliability and Availability
3.5.1. Service Level Agreement
3.5.2. Service Level Objective
3.5.3. Service Level Indicator
3.5.4. Precision Availability Metrics
3.5.5. Reliability
3.5.6. Availability
4. Reliable and Available Wireless
4.1. High Availability Engineering Principles
4.1.1. Elimination of Single Points of Failure
4.1.2. Reliable Crossover
4.1.3. Prompt Notification of Failures
4.2. Applying Reliability Concepts to Networking
4.3. Wireless Effects Affecting Reliability
5. The RAW Conceptual Model
5.1. The RAW Planes
5.2. RAW Versus Upper and Lower Layers
5.3. RAW and DetNet
6. The RAW Control Loop
6.1. Routing Timescale Versus Forwarding Timescale
6.2. OODA Loop
6.3. Observe: RAW OAM
6.4. Orient: The RAW-Extended DetNet Operational Plane
6.5. Decide: The Point of Local Repair
6.6. Act: DetNet Path Selection and Reliability Functions
7. Security Considerations
7.1. Collocated Denial-of-Service Attacks
7.2. Layer 2 Encryption
7.3. Forced Access
8. IANA Considerations
9. References
9.1. Normative References
9.2. Informative References
Acknowledgments
Contributors
Author's Address1. Introduction ✨
Deterministic Networking (DetNet) is all about bounded latency and killing congestion loss, even when it's sharing the network with best-effort traffic 💅. It can carry specified unicast or multicast data flows for real-time apps with super low packet loss rates, and it guarantees a maximum end-to-end delivery latency 🫶. If you want the general background and concepts behind DetNet, go check [DetNet-ARCH] 📚.
DetNet and the related IEEE 802.1 Time-Sensitive Networking (TSN) [TSN] started out focused on wired infrastructure 🔌. Wired gives you a much more stable communication channel than wireless does 🫠. Wireless networks run on a shared medium where uncontrolled interference, including self-induced multipath fading, can cause intermittent transmission losses 😭. Fixed and mobile obstacles and reflectors can block or change the signal, causing transient and unpredictable swings in the throughput and Packet Delivery Ratio (PDR) of a wireless link 💀. All of that adds new layers of statistical chaos to the quality and reliability of the link 🙈.
Still, deterministic capabilities are required in a bunch of wireless use cases too [RAW-USE-CASES] 🌸. Scheduled radios like Time-Slotted Channel Hopping (TSCH) and Orthogonal Frequency-Division Multiple Access (OFDMA) are being developed to give determinism over wireless links at the lower layers, and that's what made providing DetNet capabilities possible 🧚✨. For more on TSCH, OFDMA, and other technologies, see [RAW-TECHNOS] 🦋.
Reliable and Available Wireless (RAW) takes on the challenge of delivering highly available and reliable end-to-end performance in a DetNet network that might include wireless segments 👑. To pull it off, RAW uses every transmission diversity and redundancy option it can to make sure packets get delivered, while also using the shared spectrum efficiently to save bandwidth and energy 🔋💜. So RAW defines protection paths that can be turned on dynamically when failures happen, plus a control loop that switches the feasible protection paths on and off on the fly to react quickly to intermittent losses 🔥.
RAW aims to meet Service Level Objectives (SLOs) for DetNet flows in terms of PDR, maximum contiguous losses, or latency boundaries 🎯. That covers networks mixing wired and wireless parts, including wireless access and meshes (see Section 2 for more on the RAW problem) 🌈. This document introduces and/or leverages terminology (see Section 3), principles (see Section 4), and concepts like protection paths and recovery graphs 🍓. Together those build a conceptual model for RAW (see Section 5) ✨. From that model, the document then walks through an in-network optimization control loop (see Section 6) 💖.
2. The RAW Problem ✨
2.1. Why This Is a Whole Thing 🫶
okay so bestie, the generic "Deterministic Networking Problem Statement" [RFC8557] applies to both wired and wireless media 💖, but the "Deterministic Networking Architecture" [DetNet-ARCH] MUST be extended to deal with less consistent transmissions, energy conservation, and shared spectrum efficiency 🌸. it's giving "the basics work, but wireless needs its own glow-up" 💅
2.2. Layer 3 Stays in Its Lane 🦋
operating at Layer 3, RAW does not change the wireless technology at the lower layers 🎀. BUT it can totally increase diversity in the spatial, time, code, and frequency domains 🌈 by letting multiple link-layer wired and wireless technologies run in parallel or one after another 👑. that means higher resilience and a wider applicability, periodt ✨
RAW can also give homogeneous services to critical applications beyond the boundaries of a single subnetwork 🧚, e.g., by using diverse radio access technologies to optimize the end-to-end application experience 💜. main character energy fr 🔥
RAW extends the DetNet services by adding elements that are specialized for transporting IP flows over deterministic radio technologies such as those listed in [RAW-TECHNOS] 🦋. conceptually, RAW is agnostic to the lower layer 🫧, though the capability to control latency is assumed so it can ensure the DetNet services it extends ⏱️. how the lower layers do that (and whether a radio network is single hop or meshed, for example) is opaque to the IP layer and is not part of the RAW abstraction 🙈. still, cross-layer optimizations may take place to make sure there's proper link awareness (like link quality) and packet handling (like scheduling) 💅
2.3. The Architecture Glow-Up 🌟
the RAW architecture extends the DetNet Network Plane to handle one or multiple hops of homogeneous or heterogeneous wired and wireless technologies 🍓. RAW adds reactivity to the DetNet forwarding sub-layer so it can deal with the dynamics of radio links in terms of lossiness and bandwidth 🌊. this may apply to mesh networks like the ones in Figure 4 🕸️, or diverse radio access networks like Figure 10 📡
2.4. Wireless Is Unreliable, Lowkey Iconic 💀
as opposed to wired links, the availability and performance of an individual wireless link cannot be trusted over the long term 😭. it varies with transient service discontinuity, and any path with wireless hops is bound to face short periods of high loss 🌧️. BUT, being broadcast in nature, the wireless medium has capabilities that are atypical on modern wired links 🦄, and the RAW architecture can leverage them opportunistically to improve end-to-end reliability over a collection of paths 💖
Those capabilities include:
2.4.1. Promiscuous Overhearing 👂
some wired and wireless technologies let multiple lower-layer attached nodes receive the same packet sent by another node 🫣. this is different from a lower-layer network that is physically point-to-point, like a wire 🔌. with overhearing, more than one node in the forward direction of the packet may hear (or overhear) a transmission, and the reception by one may make up for the loss at another ✨. so the concept of path can be revisited in favor of multipoint-to-multipoint progress in the forward direction, plus the statistical chances of successful reception of any transmission by any of the receivers 🎯
2.4.2. L2-Aware Routing 🗺️
as the quality and speed of a link varies over time, the concept of metric has to be revisited too 😤. shortest-path cost loses its absolute value, and hop count turns into a bad idea as the link budget drops with physical distance 📉. routing over radio requires both:
- a new and more dynamic sense of link metrics, with new protocols such as the Dynamic Link Exchange Protocol (DLEP) and Layer 2 (L2) triggers to keep Layer 3 (L3) up to date with link quality and availability 📶, and
- an approach to multipath routing, where multiple link metrics are considered, since simple shortest-path cost loses its meaning with the instability of the metrics 🧠
2.4.3. Redundant Transmissions 🔁
though it's feasible on any technology, proactive (forward) and reactive (acknowledgment-based) error correction is typical for wireless media 🛡️. bounded latency can still be obtained on a wireless link while operating those technologies, provided that the link latency used in path selection allows for the extra transmission, or the introduced delay is compensated along the path ⏳. for coded fragments and retries, it makes sense to vary all the possible physical properties of the transmission 🎨, so the same effect is less likely to cause the loss of both the original and the redundant transmissions 🙌
2.4.4. Relay Coordination and Constructive Interference 🌈
though it can be difficult to achieve at high speed, a fine time synchronization and a precise sense of phase let the energy from multiple coordinated senders add up at the receiver 🤯. that can actually improve the signal quality, compensating for distance or for physical objects in the Fresnel zone that would reduce the link budget 🏗️. from a DetNet perspective, this may translate to taking into account how transmission from one node may interfere with the transmission of another node attached to the same wireless sub-layer network 🫶
2.5. Paths, Flows, and Who's Really in Charge 👑
RAW and DetNet enable application flows that need special treatment along paths that can provide that treatment 🎀. this may be seen as a form of Path Aware networking, and it may be subject to the impediments documented in [RFC9049] 🙈
the mechanism used to establish a path is not unique to RAW, and isn't necessarily impacted by it 🤷♀️. the mechanism is expected to be the product of the DetNet Controller Plane [DetNet-PLANE] 🧚. it may use a Path Computation Element (PCE) [RFC4655] or the DetNet YANG data model [RFC9633], or it may be computed in a distributed fashion as per the Resource ReSerVation Protocol (RSVP) [RFC2205] 🌸. either way, the assumption is that it is slow relative to local forwarding operations along the path 🐢. to react fast enough to transient changes in radio transmissions, RAW leverages DetNet Network Plane enhancements to optimize the use of the paths and match the quality of transmissions over time ⚡
as opposed to wired networks, installing a path over a set of wireless links may be very slow relative to how fast the radio conditions vary 😵💫. so in the wireless case, it makes sense to provide redundant forwarding solutions along alternate paths (see Section 3.3) and leave it to the Network Plane to select which of those forwarding solutions to use for a given packet based on current conditions 🧩. the RAW Network Plane operations happen within the scope of a recovery graph (see Section 3.3.2) that is pre-established and installed by means outside the scope of RAW 🏗️. a recovery graph may be strict or loose, depending on whether each hop or just a subset of the hops is observed and controlled by RAW 🔍
RAW distinguishes the longer timescale at which routes are computed from the shorter timescale where forwarding decisions are made (see Section 6.1) ⏰. the RAW Network Plane operations happen at a timescale that sits between the routing and forwarding timescales 🥪. within the resources delineated by a recovery graph, their goal is to dynamically select the protection path(s) that the upcoming packets of a DetNet flow shall follow 🎯. since they influence the path for the entirety of the flows or a portion of them, the RAW Network Plane operations may affect the metrics used in rerouting decisions, which could potentially lead to oscillations 🌀. such effects MUST be avoided or dampened 🛑
3. Terminology 📚
RAW reuses terminology defined for DetNet in "Deterministic Networking Architecture" [DetNet-ARCH], e.g., "PREOF" to stand for "Packet Replication, Elimination, and Ordering Functions" 🫧. RAW inherits and augments IETF recovery mechanisms such as the ones provided in DetNet [DetNet-ARCH] and in Traffic Engineering, e.g., [RFC4090] 🔄
RAW also reuses terminology defined for Operations, Administration, and Maintenance (OAM) protocols in Section 1.1 of "Framework of Operations, Administration, and Maintenance (OAM) for Deterministic Networking (DetNet)" [DetNet-OAM] and in "Active and Passive Metrics and Methods (with Hybrid Types In-Between)" [RFC7799] 📏
RAW also reuses terminology defined for MPLS in [RFC4427], such as the term "recovery" to cover both protection and restoration for a number of recovery types 🛠️. that document defines a number of concepts, such as the recovery domain, that RAW mechanisms use, and it defines the new term "recovery graph" 🗂️. a recovery graph associates a topological graph with usage metadata that represents how the paths are built and used within the recovery graph 🗺️. the recovery graph provides excess bandwidth for the intended traffic over alternate potential paths, and the use of that bandwidth is optimized dynamically 💫
the concept of a recovery graph is agnostic to the underlying technology and applies to, but isn't limited to, any full or partial wireless mesh 🕸️. RAW specifies strict and loose recovery graphs, depending on whether the path is fully controlled by RAW or traverses an opaque network where RAW cannot observe and control the individual hops 🫣
RAW also reuses terminology defined for RSVP-TE in [RFC4090], such as the "Point of Local Repair (PLR)" 🧑🔧. the concept of a backup path is generalized with the term protection path, which is the term mostly found in recent standards and is used in this document 🦋
RAW also reuses terminology defined for 6TiSCH in [6TiSCH-ARCH] and equates the 6TiSCH concept of a Track with that of a recovery graph 🎀
3.1. Abbreviations 💖
RAW uses the following abbreviations, and I'm going to walk you through every single one, bestie 🫶
ARQ means Automatic Repeat Request 🔁. It's a well-known mechanism that makes a transmission get acknowledged and retried so errors and loss get handled. ARQ can show up at different layers of a network. It's typically done per hop (not end to end) at Layer 2 in wireless networks, and it helps delivery on lossy wireless. ARQ retransmissions can also be capped by a bounded time so they still meet end-to-end packet latency constraints. You can find more details and considerations for ARQ in [RFC3366] 📖
FEC is Forward Error Correction 🛡️. It adds redundant data so a partial loss can be protected against without any retries. Less back-and-forth, more protection, very efficient vibes ✨
HARQ is Hybrid ARQ, which is a combo of FEC and ARQ. Two powerhouses together, it's giving teamwork 👯♀️
ETX is Expected Transmission Count 📊. It's a statistical metric that represents the expected total number of packet transmissions (retransmissions included) needed to successfully deliver a packet along a path. 6TiSCH uses it, as noted in [RFC6551].
ISM stands for Industrial, Scientific, and Medical 🏭🔬. It refers to a group of radio bands or parts of the radio spectrum (for example, 2.4 GHz and 5 GHz) that are reserved internationally for radio frequency (RF) energy meant for industrial, scientific, and medical purposes (for example, microwaves, depth radars, and medical diathermy machines). Cordless phones, Bluetooth and Low-Power Wireless Personal Area Network (LoWPAN) devices, near-field communication (NFC) devices, garage door openers, baby monitors, and Wi-Fi networks may all use the ISM frequencies. Still, these low-power transmitters are not considered to be ISM devices 📻 Generally, communications equipment operating in ISM bands must tolerate any interference generated by ISM applications, and users have no regulatory protection from ISM device operation in these bands. Lowkey a lot of responsibility for the equipment 😭
PER is Packet Error Rate 📉. It's the ratio of the number of packets received in error to the total number of transmitted packets. A packet counts as being in error if even a single bit within it is received incorrectly. One wrong bit and the whole packet is marked as errored, no mercy 💀
PDR is Packet Delivery Ratio (PDR) 📬. It's the ratio of the number of successfully delivered data packets to the total number of packets transmitted from the sender to the receiver. Basically the success rate, and I love a good success rate 🌸
RSSI is Received Signal Strength Indication. It's also known as "Energy Detection Level" 📡. It measures the incoherent (raw) RF power in a channel. That power can come from any source: other transmitters using the same technology, other radio technology using the same band, or background radiation. For a single hop, RSSI may be used for LQI 🧚
LQI is Link Quality Indicator ✨. It indicates the quality of the data packets received by the receiver. It's typically derived from packet error statistics, and the exact method depends on the network stack being used. LQI values may be exposed to the Controller Plane for each individual hop or cumulated along segments. Outgoing LQI values can be calculated from coherent (demodulated) PER, RSSI, and incoming LQI values. So many moving parts, iykyk 🧠
OAM stands for Operations, Administration, and Maintenance 🛠️. It covers the processes, activities, tools, and standards involved with operating, administering, managing, and maintaining any system. This document uses the term in line with "Guidelines for the Use of the 'OAM' Acronym in the IETF" [RFC6291], and the system observed by the RAW OAM is the recovery graph 🗺️
OODA means Observe, Orient, Decide, Act 👀. It's a generic formalism for representing the operational steps in a control loop. In the context of RAW, OODA is applied to network control and convergence. See Section 6.2 for more 🔄
SNR is Signal-to-Noise Ratio. It's also known as "S/N Ratio" 📶. It's a measure used in science and engineering that compares the level of a desired signal to the level of background noise. SNR is defined as the ratio of signal power to noise power, often expressed in decibels. Signal goes brrr, noise goes away 🎧
3.2. Link and Direction 🧭
This document uses the following terms relating to links and direction in the context of RAW. Let's get our directions straight, bestie 💅
3.2.1. Flapping
In the context of RAW, a link flaps when the reliability of the wireless connectivity drops abruptly for a short period of time, typically a duration of a subsecond to seconds. Think of it as the connection having a little glitch moment ⚡
3.2.2. Uplink
An uplink is the connection from end devices to data communication equipment. In the context of wireless, uplink refers to the connection between a station (STA) and a controller (AP), or between a User Equipment (UE) and a Base Station (BS) such as a 3GPP 5G gNodeB (gNB) 📱➡️🗼
3.2.3. Downlink
A downlink is the reverse direction from uplink. Opposite day, but make it networking 🙃
3.2.4. Downstream
Downstream refers to the direction of the flow data path along a recovery graph. Going with the flow, literally 🌊
3.2.5. Upstream
Upstream refers to going against the direction of the flow data path along a recovery graph. Swimming against the current, and honestly, respect 💪
3.3. Path and Recovery Graphs 🕸️
This document uses the following terms relating to paths and recovery graphs in the context of RAW. This is where it gets spicy, so pay attention 👑
3.3.1. Path
Section 1.3.3 of [INT-ARCH] gives us a definition of path, and it goes like this:
At a given moment, all the IP datagrams from a particular source
host to a particular destination host will typically traverse the
same sequence of gateways. We use the term "path" for this
sequence. Note that a path is uni-directional; it is not unusual
to have different paths in the two directions between a given host
pair.Section 2 of [RFC9473] points to a longer, more modern definition of path, which begins like this:
A sequence of adjacent path elements over which a packet can be
transmitted, starting and ending with a node.
Paths are unidirectional and time dependent, i.e., there can be a
variety of paths from one node to another, and the path over which
packets are transmitted may change. A path definition can be
strict (i.e., the exact sequence of path elements remains the
same) or loose (i.e., the start and end node remain the same, but
the path elements between them may vary over time).
The representation of a path and its properties may depend on the
entity considering the path. On the one hand, the representation
may differ due to entities having partial visibility of path
elements comprising a path or their visibility changing over time.So the general acceptance of a path is a linear sequence of links and nodes, as opposed to a multi-dimensional graph, and it's defined by the experience of the packet that went from a node A to a node B 🛤️ In the context of this document, a path is observed by following one copy or one fragment of a packet that keeps its uniqueness and integrity. For instance, if C replicates to E and F, and if D eliminates duplicates, a packet from A to B can experience two paths: A->C->E->D->B and A->C->F->D->B. Those paths are called protection paths 🛡️ Protection paths may be fully non-congruent. Alternatively, they may intersect at replication or elimination points. Two routes, one packet, zero drama 🎀
With DetNet and RAW, a packet may be duplicated, fragmented, and network coded, and the various byproducts may travel different paths that are not necessarily end to end between A and B. We refer to this complex scenario as a DetNet path 🌈 As such, the DetNet path extends the above description of a path, but it still matches the experience of a packet that traverses the network. Big brain energy, but still the same packet vibes ✨
With RAW, the path experienced by a packet is subject to change from one packet to the next, but all the possible experiences are contained within a finite set. Therefore, we introduce the term "recovery graph" (see the next section) that coalesces that set and covers the overall topology where the possible DetNet paths are all contained. As such, the recovery graph coalesces all the possible paths a flow may experience, each with its own statistical probability of being used 🦋 Every path has its moment, and the recovery graph is the group chat that keeps track of all of them 💜
3.3.2. Recovery Graph 🌸
Okay so a recovery graph is a networking graph you can follow to carry packets with equivalent treatment, and it comes with usage metadata attached 📎. Unlike a path (see above), a recovery graph is a potential path, not an actual one 🦋. It's also not necessarily a neat linear sequence like a simple path, and it's not necessarily fully traversed (flooded) by every packet of a flow like a DetNet path is. ngl, the casual reader can think of a recovery graph as basically a DetNet path that bundles together multiple paths which might overlap or fork and then rejoin, for example to enable a protection service through the PREOF operations 💅
_________
| |
| IoT G/W |
|_________|
EGRESS <<=== Elimination at Egress
| |
---+--------+--+--------+--------
| Backbone |
__|__ __|__
| | Backbone | | Backbone
|__ __| Router |__ __| Router
| # |
# \ # / <-- protection path
# # #-------#
\ # / # ( Low-power )
# # \ / # ( Lossy Network)
\ /
# INGRESS <<=== Replication at recovery graph ingress
|
# <-- source device
#: Low-power device
Figure 1: Example IoT Recovery Graph to an IoT Gateway with 1+1
RedundancyRefining further, a recovery graph is defined as the coalescence of all the feasible DetNet paths that a packet with an assigned flow may be forwarded along 🧚 A packet that is assigned to the recovery graph experiences one of the feasible DetNet paths, based on the current selection by the PLR at the time the packet traverses the network.
Refining even further, the feasible DetNet paths within the recovery graph may or may not be computed in advance. Instead, they may be decided upon the detection of a change from a clean slate. Also, the PLR decision may be distributed, which yields a large combination of possible and dependent decisions, with no node in the network capable of reporting which is the current DetNet path within the recovery graph 🙈
In DetNet [DetNet-ARCH] terms, a recovery graph has the following properties 👑
- A recovery graph is a Layer 3 abstraction built upon IP links between routers. A router may form multiple IP links over a single radio interface.
- A recovery graph has one ingress and one egress node, which operate as DetNet edge nodes.
- A recovery graph is reversible, meaning that packets can be routed against the flow of data packets, e.g., to carry OAM measurements or control messages back to the ingress.
- The vertices of a recovery graph are DetNet relay nodes that operate at the DetNet Service sub-layer and provide the PREOF functions.
- The topological edges of a recovery graph are strict sequences of DetNet transit nodes that operate at the DetNet forwarding sub-layer.
Figure 2 shows the generic concept of a recovery graph, between an ingress node and an egress node. The recovery graph is composed of forward protection paths, forward segments, and crossing segments (see the definitions of those terms in the next sections) 🦋 The recovery graph contains at least two protection paths: a main path and a backup path.
------------------- forward direction ---------------------->
a ==> b ==> C -=- F ==> G ==> h T1
/ \ / | \ /
I o n E -=- T2
\ / \ | / \
p ==> q ==> R -=- T ==> U ==> v T3
I: Ingress
E: Egress
T1, T2, T3: external targets
Uppercase: DetNet relay nodes
Lowercase: DetNet transit nodes
Figure 2: A Recovery Graph and Its ComponentsOf note:
I ==> a ==> b ==> C: a forward segment to targets F and oC ==> o ==> T: a forward segment to target T (and/or U)G | n | U: a crossing segment to targets G or UI -> F -> E: a forward protection path to targets T1, T2, and T3I, a, b, C, F, G, h, E: a path to T1, T2, and/or T3I, p, q, R, o, F, G, h, E: a segment-crossing protection path
3.3.3. Forward and Crossing ➡️
Forward refers to progress towards the egress of the recovery graph. Forward links are directional, so packets that are forwarded along the recovery graph can only be transmitted along the link direction 🏃♀️ Crossing links are bidirectional, meaning they can be used in both directions, though a given packet may use the link in one direction only. A segment can be forward, in which case it is composed of forward links only, or it can be crossing, in which case it is composed of crossing links only. A protection path is always forward, meaning that it is composed of forward links and segments 💖
3.3.4. Protection Path 🛡️
A protection path is an end-to-end forward path between the ingress and egress nodes of a recovery graph. A protection path in a recovery graph is expressed as a strict sequence of DetNet relay nodes, or as a loose sequence of DetNet relay nodes that are joined by segments in the recovery graph. Background information on the concepts related to protection paths can be found in [RFC4427] and [RFC6378] 📚
3.3.5. Segment 🧩
A segment is a strict sequence of DetNet transit nodes between two DetNet relay nodes. A segment of a recovery graph is composed topologically of two vertices of the recovery graph and one edge of the recovery graph between those vertices ✨
3.4. Deterministic Networking 🎀
This document reuses the terminology in Section 2 of [RFC8557] and Section 4.1.2 of [DetNet-ARCH] for deterministic networking and deterministic networks. This document also uses the following terms.
3.4.1. The DetNet Planes 🌈
[DetNet-ARCH] defines three planes: the Application (User) Plane, the Controller Plane, and the Network Plane. The DetNet Network Plane is composed of a Data Plane (packet forwarding) and an Operational Plane where OAM operations take place. In the Network Plane, the DetNet Service sub-layer focuses on flow protection (e.g., using redundancy) and can be fully operated at Layer 3. Meanwhile, the DetNet forwarding sub-layer establishes the paths, associates the flows to the paths, ensures the availability of the necessary resources, and leverages Layer 2 functionalities for timely delivery to the next DetNet system. For more information, see Section 2 🫶
3.4.2. Flow 🌊
A flow is a collection of consecutive IP packets defined by the upper layers and signaled by the same 5-tuple or 6-tuple (see Section 5.1 of [RFC8939]). Packets of the same flow must be placed on the same recovery graph to receive an equivalent treatment from ingress to egress within the recovery graph. Multiple flows may be transported along the same recovery graph. The DetNet path that is selected for the flow may change over time under the control of the PLR 🔥
3.4.3. Residence Time ⏱️
A residence time (RT) is defined as the time interval between when the reception of a packet starts and the transmission of the packet begins. In the context of RAW, RT is useful for transit nodes, not ingress or egress nodes 💅
3.4.4. L3 Deterministic Flow Identifier 🆔
The classic IP 5-tuple that identifies a flow comprises the source IP, destination IP, source port, destination port, and the Upper-Layer Protocol (ULP). DetNet uses a 6-tuple, where the extra field is the Differentiated Services Code Point (DSCP) field in the packet (see Section 3.3 of [DetNet-DP]). The IPv6 flow label is not used for that purpose 🙅♀️
3.4.5. Time-Sensitive Networking 🕰️
Time-Sensitive Networking (TSN) denotes the IEEE 802 efforts regarding deterministic networking, originally for use on Ethernet. See [TSN]. Wireless TSN (WTSN) denotes extensions of the TSN work on wireless media, e.g., the RAW technologies described in [RAW-TECHNOS] 📡
3.4.6. Lower-Layer API 🔌
RAW includes the concept of a lower-layer API (LL API) that provides an interface between the lower-layer (e.g., wireless) technology and the DetNet layers. The LL API is technology dependent, as what the lower layers expose towards the DetNet layers may vary. Furthermore, different RAW technologies are equipped with different reliability features (e.g., short-range broadcast, Multiple User - Multiple Input Multiple Output (MU-MIMO), physical layer (PHY) rate and other Modulation Coding Scheme (MCS) adaptation, coding and retransmissions methods, and constructive interference and overhearing; see [RAW-TECHNOS] for more details). The LL API enables interactions between the reliability functions provided by the lower layer and the reliability functions provided by DetNet. That is, the LL API makes cross-layer optimization possible for the reliability functions of different layers, depending on the actual exposure provided via the LL API by the given RAW technology. The Dynamic Link Exchange Protocol (DLEP) [DLEP] is an example of a protocol that can be used to implement the LL API 🧠
3.5. Reliability and Availability 💜
In the context of the RAW work, reliability and availability are defined as follows, along with the following other terms.
3.5.1. Service Level Agreement 📝
In the context of RAW, a Service Level Agreement (SLA) is a contract between a provider (the network) and a client (the application flow) that defines measurable metrics such as latency boundaries, consecutive losses, and Packet Delivery Ratio (PDR) 🤝
3.5.2. Service Level Objective 🎯
A Service Level Objective (SLO) is one term in the SLA, for which specific network settings and operations are implemented. For instance, a dynamic tuning of packet redundancy addresses an SLO of consecutive losses in a row by augmenting the chances of delivery of a packet that follows a loss 🍓
3.5.3. Service Level Indicator 📊
A Service Level Indicator (SLI) measures the compliance of an SLO to the terms of the contract. For instance, it can be the statistics of individual losses, or losses in a row, during a certain amount of time 👀
3.5.4. Precision Availability Metrics 🔍
Precision Availability Metrics (PAMs) [RFC9544] aim to capture service levels for a flow, specifically the degree to which the flow complies with the SLOs that are in effect 💫
3.5.5. Reliability 🛠️
Reliability is a measure of the probability that an item (e.g., system or network) will perform its intended function with no failure for a stated period of time (or for a stated number of demands or load) under stated environmental conditions. In other words, reliability is the probability that an item will be in an uptime state (i.e., fully operational or ready to perform) for a stated mission (e.g., to provide an SLA). See more in [NASA1] 🧁
3.5.6. Availability ⚡
Availability is the probability of an item's (e.g., a network's) mission readiness (e.g., to provide an SLA). Availability is expressed as (uptime)/(uptime+downtime). Note that it is availability that addresses downtime (including time for maintenance, repair, and replacement activities), and not reliability. See more in [NASA2] 🥹
4. Reliable and Available Wireless ✨
4.1. High Availability Engineering Principles ✨
Okay bestie, so the reliability rules of a critical system seep into every single piece of it 🫶. If that system has a data network inside, the network inherits the same reliability and availability rules 💅. So it's only natural to borrow the art of high availability engineering and apply it to wireless comms in the context of RAW 🦋.
There are three principles (the pillars) of high availability engineering 👑:
- elimination of each single point of failure
- reliable crossover
- prompt detection of failures as they occur
These principles are common to all high availability systems, not just the ones with Internet tech at the center 🌈. Both non-Internet and Internet examples are included, so this is not just a networking thing 🧚.
4.1.1. Elimination of Single Points of Failure
Physical and logical components happen to fail 💀. It can be wear and tear when they're used beyond acceptable limits, or a software bug 🐛. It's necessary to decouple component failure from system failure so the latter doesn't happen 🙈. This lets failed components get restored while the rest of the system keeps functioning ✨.
IP routers use routing protocols to reroute to alternate routes when something fails 🛣️. When links are cabled through the same conduit, they form a Shared Risk Link Group (SRLG) and share the same fate if the conduit gets cut, which makes the reroute useless 😭. The same thing can happen with virtual links that end up in the same physical transport because of the intricacies of nested encapsulation 🌀. In the same fashion, an interferer or an obstacle may affect multiple wireless transmissions at the same time, even between different sets of peers 📡.
Intermediate network nodes (such as routers, switches and APs, wire bundles, and the air medium itself) can become single points of failure 🚨. So for high availability, the use of physically link-disjoint and node-disjoint paths is required 💖. In the wireless space, the use of the highest possible degree of diversity (time, space, code, frequency, and channel width) in over-the-air transmissions is also required to combat the extra causes of transmission loss 🎀.
From an economics standpoint, doing this properly usually increases capital expense because of the redundant equipment 💸. In a constrained network where waste of energy and bandwidth should be minimized, an excessive use of redundant links must be avoided. For RAW, this means the extra bandwidth must be used wisely and efficiently 🍓.
4.1.2. Reliable Crossover
Backup equipment has limited value unless it can be reliably switched into use within the downtime parameters 😤. IP routers execute reliable crossover continuously because they use any alternate routes that are available [RFC0791] 🔁. This works because of the stateless nature of IP datagrams and the dissociation of the datagrams from the forwarding routes they take 🧠. "IP Fast Reroute Framework" [FRR] analyzes mechanisms for fast failure detection and path repair for IP Fast Reroute (FRR), and it discusses the case of multiple failures and SRLG 📚. Examples of FRR techniques include Remote Loop-Free Alternate [RLFA-FRR] and backup Label Switched Path (LSP) tunnels for the local repair of LSP tunnels using RSVP-TE [RFC4090] ✨.
Deterministic flows, on the contrary, are attached to specific paths where dedicated resources are reserved for each flow 🎯. Therefore, each DetNet path must inherently provide sufficient redundancy to deliver the assured SLOs at all times 💅. The DetNet PREOF typically leverages 1+1 redundancy, where a packet is sent twice over non-congruent paths 👯. This avoids the gap during the FRR operation but doubles the traffic in the network 😬.
In the case of RAW, the expectation is that multiple transient faults may happen in overlapping time windows 😱. In that case, 1+1 redundancy with delayed reestablishment of the second path does not provide the required guarantees. The Data Plane must be configured with a sufficient degree of redundancy to select an alternate redundant path immediately upon a fault, without needing a slow intervention from the Controller Plane 🚀.
4.1.3. Prompt Notification of Failures
The execution of the two principles above is likely to render a system where the end user rarely sees a failure 🥰. But a failure that does occur must still be detected so maintenance can be directed 🔧.
There are many reasons for system monitoring (Fault, Configuration, Accounting, Performance, and Security (FCAPS) is a handy mental checklist), but fault monitoring is a sufficient reason on its own 📋.
"Overview and Principles of Internet Traffic Engineering" [TE] discusses the importance of measurement for network protection. It provides an abstract method for network survivability with the analysis of a traffic matrix as observed via a network management YANG data model, probing techniques, file transfers, IGP link state advertisements, and more 📊.
Those measurements are needed in the context of RAW to inform the controller and make the long-term reactive decision to rebuild a recovery graph based on statistical and aggregated information 🗺️. RAW itself operates in the DetNet Network Plane at a faster timescale, with live information on speed, state, etc ⚡. This live information can be obtained directly from the lower layer (e.g., using L2 triggers), read from a protocol such as DLEP, or transported over multiple hops using OAM and reverse OAM, as illustrated in Figure 11 🦋.
4.2. Applying Reliability Concepts to Networking 💜
The terms "reliability" and "availability" are defined for use in RAW in Section 3, and you're invited to read [NASA1] and [NASA2] for more details on the general definition of reliability 📖. Practically speaking, a number of nines is often used to indicate the reliability of a data link (e.g., 5 nines indicate a Packet Delivery Ratio (PDR) of 99.999%) 💯.
This number is typical in a wired environment, where the loss is due to a random event such as a solar particle that affects the transmission of a particular packet but does not affect the previous packet, the next packet, or packets transmitted on other links ☀️. Note that the QoS requirements in RAW may include a bounded latency, and a packet that arrives too late is a fault and not considered as delivered ⏰.
For a periodic networking pattern such as an automation control loop, this number is proportional to the Mean Time Between Failures (MTBF) 🔄. When a single fault can have dramatic consequences, the MTBF expresses the chances that the unwanted fault event occurs 😳. In data networks, this is rarely the case 🙌. Packet loss cannot be fully avoided, and the systems are built to resist some loss 💪. This can be done by using redundancy with retries (as in HARQ), Packet Replication and Elimination (PRE) FEC, and Network Coding (e.g., using FEC with Static Context Header Compression (SCHC) [RFC8724] fragments) 🧩. Also, in a typical control loop, linear interpolation from the previous measurements can be used 📈.
However, the linear interpolation method cannot resist multiple consecutive losses, and a high MTBF is desired as a guarantee that the number of losses in a row is bounded 🙅. In this case, what is really desired is a Maximum Consecutive Loss (MCL) 🎯. If the number of losses in a row passes the MCL, the control loop has to abort, and the system (e.g., the production line) may need to enter an emergency stop condition 🚨.
Engineers that build automated processes may use the network reliability expressed in nines as the MTBF and as a proxy to indicate an MCL, e.g., as described in Section 7.4 of "Deterministic Networking Use Cases" [RFC8578] 🛠️.
4.3. Wireless Effects Affecting Reliability 🌩️
In contrast with wired networks, errors in transmission are the predominant source of packet loss in wireless networks 📶.
The root cause for the loss may be of multiple origins, calling for the use of different forms of diversity 🌈:
- Multipath fading: A destructive interference by a reflection of the original signal 🪞.
A radio signal may be received directly (line-of-sight) and/or as a reflection on a physical structure (echo). The reflections take a longer path and are delayed by the extra distance divided by the speed of light in the medium. Depending on the frequency, the echo lands with a different phase, which may either add up to (constructive interference) or cancel (destructive interference) the direct signal ✨.
The affected frequencies depend on the relative position of the sender, the receiver, and all the reflecting objects in the environment. A given hop suffers from multipath fading for multiple packets in a row until a physical movement changes the reflection patterns 🏃.
- Co-channel interference: Energy in the spectrum used for the transmission confuses the receiver 📻.
The wireless medium itself is a Shared Risk Link Group (SRLG) for nearby users of the same spectrum, as an interference may affect multiple co-channel transmissions between different peers within the interference domain of the interferer, possibly even when they use different technologies 🫣.
- Obstacle in Fresnel zone: The Fresnel zone is an elliptical region of space between and around the transmit and receive antennas in a point-to-point wireless communication. The optimal transmission happens when it is free of obstacles 🚧.
In an environment that is rich in metallic structures and mobile objects, a single radio link provides a fuzzy service, meaning that it cannot be trusted to transport the traffic reliably over a long period of time 🌫️.
Transmission losses are typically not independent, and their nature and duration are unpredictable 🎲. As long as a physical object (e.g., a metallic trolley between peers) that affects the transmission is not removed, or as long as the interferer (e.g., a radar in the ISM band) keeps transmitting, a continuous stream of packets are affected 🛒.
The key technique to combat those unpredictable losses is diversity 🔑. Different forms of diversity are necessary to combat different causes of loss, and the use of diversity must be maximized to optimize the PDR 💪.
A single packet may be sent at different times (time diversity) over diverse paths (spatial diversity) that rely on diverse radio channels (frequency diversity), leveraging diverse PHY technologies (e.g., narrowband versus spread spectrum or diverse codes) 🦋. Using time diversity defeats short-term interferences. Spatial diversity combats very local causes of interference, such as multipath fading. Narrowband and spread spectrum are relatively innocuous to one another and can be used for diversity in the presence of the other 🌸.
5. The RAW Conceptual Model ✨
okay bestie, so RAW takes the conceptual model from Section 4 of "Deterministic Networking Architecture" [DetNet-ARCH] and adds a PLR at the Service sub-layer 🦋 (it's in Figure 3). The PLR (see Section 6.5) gives you extra agility against transmission loss. For example, the PLR can act based on indications from the lower layer or based on OAM 💅
| packets going | ^ packets coming ^
v down the stack v | up the stack |
+-----------------------+ +-----------------------+
| Source | | Destination |
+-----------------------+ +-----------------------+
| Service sub-layer: | | Service sub-layer: |
| Packet sequencing | | Duplicate elimination |
| Flow replication | | Flow merging |
| Packet encoding | | Packet decoding |
| Point of Local Repair | | |
+-----------------------+ +-----------------------+
| Forwarding sub-layer: | | Forwarding sub-layer: |
| Resource allocation | | Resource allocation |
| Explicit routes | | Explicit routes |
+-----------------------+ +-----------------------+
| Lower layers | | Lower layers |
+-----------------------+ +-----------------------+
v ^
\_________________________/
Figure 3: Extended DetNet Data Plane Protocol Stack5.1. The RAW Planes 🌸
RAW nodes are DetNet relay nodes that live in the RAW Network Plane. They can do extra diversity stuff and measurement functions that are related to the radio interface 📡 RAW also has an Operational Plane orientation function (which usually runs inside the ingress edge nodes) that dynamically adapts the path of the packets and optimizes resource usage. It's giving efficiency, no cap 🧚
When routing is centralized, the RAW Controller Plane Function (CPF) talks to RAW nodes over a Southbound API. It takes in data and info from the network, turns it into knowledge and wisdom, and uses that to steer traffic in the best way inside a recovery graph 👑
DetNet Routing
CPF CPF CPF CPF
Southbound API
_-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-
_-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-
___ RAW ___ RAW ___ RAW ___ RAW __
/ Node Node Node Node \
Ingress __/ / \ / \ \____Egress
End __ / \ / .- -- . \ ___ End
Node \ / \ / .-( ). \ / Node
\_ RAW ___ RAW ___(Non-RAW Nodes)__ RAW _/
Node Node (___.______.____) Node
Figure 4: RAW Nodes (Centralized Routing Case)When a new flow shows up, the routing function uses what it currently knows about the network to build a new recovery graph between an ingress End System and an egress End System for that flow 🔥 It then tells the RAW nodes where the PREOF and/or radio diversity and reliability operations may be actioned in the Network Plane.
- The recovery graph may be strict, meaning that the DetNet forwarding sub-layer operations are enforced end to end. 💜
- The recovery graph may be expressed loosely to enable traversing a non-RAW subnetwork as in Figure 7. In that case, RAW cannot leverage end-to-end DetNet and cannot provide latency guarantees. 🙈
ngl this loose option is the one to watch: once you go through a non-RAW subnetwork, RAW can't use end-to-end DetNet and can't promise latency guarantees. Not me panicking about it 😭
The info the orientation function reports up to the routing function can be time aggregated (like statistical), so it matches the longer-term pace of the routing function ⭐ Examples include link-layer metrics such as link bandwidth (the medium speed depends dynamically on the PHY layer mode), number of flows (the bandwidth you can reserve for a flow depends on how many flows share the spectrum and how big they are), and the average and mean squared deviation of availability and reliability metrics (such as PDR) over long periods of time. It may also report an aggregated Expected Transmission Count (ETX) or some variation of it 🎀
Based on all those metrics, the routing function installs the recovery graph with enough redundant forwarding solutions so the Network Plane can reliably deliver packets within the SLA tied to the flows it carries. The SLA defines end-to-end reliability and availability requirements, and reliability may be expressed as successful delivery, in order and within a bounded delay, of at least one copy of a packet 💖
Depending on the use case and the SLA, the recovery graph may include non-RAW segments. These can be interleaved inside the recovery graph (for example, over tunnels) or run all the way to the egress End node (for example, a server in the local wired domain) 🌈 RAW keeps an eye on the lower-layer links between RAW nodes (typically radio links) and on end-to-end network-layer operation, so it can decide at every moment which diversity scheme gets actioned by which RAW nodes 🦋
Once a recovery graph is established, per-segment and end-to-end reliability and availability statistics are periodically reported to the routing function so the SLA can be met. If it can't be, the recovery graph is recomputed 🔁
5.2. RAW Versus Upper and Lower Layers 💅
RAW builds on DetNet features like scheduling and shaping 🧁 Specifically, RAW inherits the DetNet guarantees on end-to-end latency, which can be tuned so DetNet and RAW reliability mechanisms have no side effect on upper layers (for example, on transport-layer packet recovery). RAW operations can include rerouting, and rerouting may affect the ordering of a burst of packets. RAW also inherits PREOF from DetNet, which can be used to reorder packets before they get delivered to the upper layers. So DetNet in general, and RAW in particular, give upper layers a smoother transport experience than the Internet at large, with ultra-low jitter and loss 🫶
RAW improves the reliability of transmissions and the availability of communication resources, and it should be seen as a dynamic optimization of redundancy that keeps reliability and availability metrics within certain boundaries ✨ For instance, ARQ (which gives one-hop reliability through acknowledgements and retries) and FEC codes (such as turbo codes, which reduce the PER) are typically operated at Layer 2 and Layer 1, respectively. In both cases, redundant transmissions improve one-hop reliability at the expense of energy and latency, and those are exactly the resources RAW must control 🌸 To hit its goals, RAW may lean on the lower-layer operations by abstracting the concept and giving hints to the lower layers about the desired outcome (for example, in terms of reliability and timeliness), instead of doing the actual operations at Layer 3 🧚
Guarantees like bounded latency depend on the upper layers (transport or application) handing over the payload in volumes and at times that match the contract with the DetNet sub-layers and the layers below. If there's too much incoming traffic at the DetNet ingress, packets may get dropped or queued, and that can cause loss, latency, or jitter. That breaks the guarantees DetNet provides inside the DetNet Network 😭
When the traffic from upper layers matches what the lower layers expect, RAW still depends on DetNet mechanisms and the lower layers to provide the timing and physical resource guarantees needed to match the traffic SLA 👑 When the availability of a physical resource varies, RAW acts on how traffic is distributed so it can use alternates within a finite set of potential resources 🦋
The Operational Plane elements (routing and OAM control) may gather aggregated information from lower layers (for example, info about link quality), either through measurement or by communicating with the lower layer 🍓 This info may be obtained from inside the device using specialized APIs (for example, L2 triggers) through monitoring and measurement protocols such as Bidirectional Forwarding Detection (BFD) [RFC5880] and Simple Two-way Active Measurement Protocol (STAMP) [RFC8762], respectively. It can also come from a control protocol exchange with the lower layer (for example, DLEP [DLEP]). Then it may be processed and exported through OAM messaging or via a YANG data model, and exposed to the Controller Plane 💜
5.3. RAW and DetNet 💖
okay so RAW totally leans on the DetNet forwarding sub-layer 🫶 and it needs DetNet transit nodes to support OAM (check Figure 3 of [DetNet-ARCH] if you want the picture). why? so it can dynamically grab link capacity and state and keep a strict RAW service going end to end over a DetNet network. and then DetNet (and therefore RAW) can get extra help from stuff like TSN at the lower layers. the glow-up is layered, bestie 🌸
RAW extends DetNet to protect against link errors like transient flapping, which happens way more often on wireless links 📡 ngl that's the main thing. BUT the RAW methods mostly work on wired links too 🔌✨ for example, when you want to save energy, or when the available path diversity is bigger than 1+1 linear redundancy. no cap, it's giving versatile 💅
RAW adds sub-layer functions that run in the DetNet Operational Plane, which is part of the Network Plane 🧚 the RAW orientation function can run only in the DetNet edge nodes (ingress edge node or End System), OR it can also run in DetNet relay nodes when the RAW operations are spread out along the recovery graph. the RAW Service sub-layer has the PLR, which decides the DetNet path for the future packets of a flow along the DetNet path. it also has Maintenance End Points (MEPs) on edge nodes and Maintenance Intermediate Points (MIPs) inside the network 🦋 the MEPs trigger OAM observations, learn from them, and feed the PLR so it can make its next decision. so iconic, the feedback loop is everything 😭
As shown in Figure 5, RAW extends the DetNet Stack (see Figure 4 of [DetNet-ARCH] and Figure 3) with extra functionality at the DetNet Service sub-layer so PREOF can get actuated based on the PLR decision 🔥 DetNet works at Layer 3 and leans on abstractions of the lower layers, plus the APIs that control those abstractions. like, DetNet already uses the lower layers for time-sensitive stuff such as time synchronization and traffic shapers ⏱️ but the radio layers change super fast, so RAW needs more dynamic gauges and knobs 🎀 that's where the LL API comes in. it gives DetNet an abstraction it can use to push reliability and timing hints, like suggesting X retries (min, max) within a time window, or sending unicast (one next hop) or multicast (for overhearing). going the other way up the stack, the RAW PLR needs hints about radio conditions, like L2 triggers (e.g., RSSI, LQI, or ETX) over all the wireless hops 📶 it's a whole two-way conversation, periodt 💬
RAW uses different OAM functionalities at different layers 🧚♀️ for instance, the OAM function in the DetNet Service sub-layer may do Active and/or Hybrid OAM to estimate link and path availability, either end to end or just for a segment. the RAW functions may be present in the Service sub-layer in DetNet edge and relay nodes ✨
+-----------------+ +-------------------+
| Routing | | OAM Control |
+-----------------+ +-------------------+
Controller Plane
+-+-+-+-+-+-+-+-+ Southbound Interface -+-+-+-+-+-+-+-+-+-+-+-+
Network Plane
|
Operational Plane . Data Plane
|
+-----------------+ .
| Orientation | |
+-----------------+ .
|
+-----------------+ +-------------------+ .
| Point of Local | | OAM Maintenance | |
| Repair (PLR) | | End Point (MEP) | .
+-----------------+ +-------------------+ |
.
|Figure 5: RAW Function Placement (Centralized Routing Case)
there are two main proposed models for deploying RAW and DetNet: strict (Figure 6) and loose (Figure 7) 🦋 in the strict model (Figure 6), RAW runs over a continuous DetNet service end to end, between the ingress and the egress edge nodes or End Systems. slay, super clean 💅
in the loose model (Figure 7), RAW may go through a section of the network that DetNet doesn't serve 🌈 RAW OAM can still watch the end-to-end traffic and make the best of whatever resources are around, but it may not expect DetNet guarantees over all paths. like, packets between two wireless entities might get relayed over a wired infrastructure. in that case RAW observes and controls the transmission over the wireless first and last hops, plus end-to-end metrics like latency, jitter, and delivery ratio. it's "loose" because the structure and properties of the wired part are ignored, and they might be controlled some other way (like DetNet/TSN) or just neglected once the wireless hops are in the mix 🙈
a minimal forwarding sub-layer service is provided at all DetNet nodes so the OAM information can keep flowing 🫧 DetNet relay nodes may or may not support RAW services, whereas DetNet edge nodes are required to support RAW in any case. DetNet guarantees, like bounded latency, are provided end to end. RAW extends the DetNet Service sub-layer to optimize resource use 💜
--------------------Flow Direction---------------------------------->
+---------+
| RAW |
| Control |
+---------+ +---------+ +---------+
| RAW + | | RAW + | | RAW + |
| DetNet | | DetNet | | DetNet |
| Service | | Service | | Service |
+---------+---------------------------+---------+--------+---------+
| DetNet |
| Forwarding |
+------------------------------------------------------------------+
Ingress Transit Relay Egress
Edge ... Nodes ... Nodes ... Edge
Node Node
<------------------End-to-End DetNet Service----------------------->Figure 6: RAW over DetNet (Strict Model)
in the loose model (Figure 7), RAW runs over a partial DetNet service, where usually only the ingress and egress End Systems support RAW 🎀 the DetNet domain may extend beyond the ingress node, or there may be a DetNet domain starting at an ingress edge node at the first hop after the End System.
in the loose model, RAW can't observe the hops in the network, and the path beyond the first hop is opaque 🙈 RAW can still watch the end-to-end behavior and use Layer 3 measurements to decide whether to replicate a packet and pick the first-hop interface(s). the plot thickens 🍓
--------------------Flow Direction---------------------------------->
+---------+
| RAW |
| Control |
+---------+ +---------+ +---------+
| RAW + | | DetNet | | RAW + |
| DetNet | | Only | | DetNet |
| Service | | Service | | Service |
+---------+----------------------+---+ +---+---------+
| DetNet |_______________| DetNet |
| Forwarding _______________ Forwarding |
+------------------------------------+ +-------------+
Ingress Transit Relay Tunnel Egress
End ... Nodes ... Nodes ... ... End
System System
<---------------Partitioned DetNet Service------------------------->Figure 7: RAW over DetNet (Loose Model)
6. The RAW Control Loop 🔁
the RAW architecture is built on an abstract OODA loop that controls how a recovery graph runs 👑 the generic idea goes like this:
- Operational Plane measurement protocols let OAM observe (that's the first "O" in "OODA" 👀) some or all hops along a recovery graph, plus the end-to-end packet delivery.
- The DetNet Controller Plane sets up primary and protection paths for the RAW Network Plane to use 🧠 the orientation function reports data and info, like link statistics, that the routing function uses to compute, install, and maintain the recovery graphs. the routing function can also generate intelligence, like a trained model for link quality prediction, and the orientation function (that's the second "O" in "OODA") can use that to influence path selection by the PLR inside the RAW OODA loop 🌸
- A PLR operates at the DetNet Service sub-layer and hosts the decision function (that's the "D" in "OODA" 💅). the decision function figures out which DetNet paths will be used for future packets routed within the recovery graph.
- Service protection actions get actuated or triggered over the LL API by the PLR to boost the reliability of end-to-end transmissions ✨ the RAW architecture also covers in-situ signaling embedded in live user traffic [RFC9378] (for example, via OAM) when the decision gets acted upon (that's the "A" in "OODA" 🫶) by a node downstream in the recovery graph from the PLR.
the whole OODA loop is about using redundancy smartly to hit the required reliability and availability SLO(s) while keeping constrained resources like spectrum and battery use as low as possible 🔋💖
6.1. Routing Timescale Versus Forwarding Timescale 🦋
Okay so with DetNet, the Controller Plane Function (CPF) is the one doing the routing math and keeping it maintained 🧠. With RAW, though, routing gets split off from the RAW control loop, bestie 💅. Routing can chill in the Controller Plane while the control loop itself lives in the Network Plane. To make RAW work, routing gets a glow-up so it generates the info the orientation function needs inside the loop ✨. Like, routing might suggest DetNet paths to use as a reflex action when stuff happens on the network, or hand over an aggregated history that the orientation function can use to make a call 🫶.
In a wireless mesh, getting to a routing function in the Controller Plane can be expensive and slow 🐢, possibly going across the whole mesh and back again 😭. Reaching the Controller Plane can also be way too slow compared to how fast things happen to the forwarding operation in the Network Plane at the radio layer. Also, ngl, a distributed routing protocol may take forever and eat up a ton of wireless resources just to reconverge to a new optimized state 💀.
So, the DetNet routing function is not expected to be aware of and react to very transient changes 🙈. Instead, the abstraction of a link at the routing level is expected to use statistical metrics that aggregate how a link behaves over long periods of time. These represent its properties as shades of gray 🩶, not as numerical values like a link quality indicator, or a Boolean value for either up or down. Very main character of metrics, honestly 👑.
The interaction between the network nodes and the routing function is handled by the orientation function. It reports to the routing function and sends control information in a digested form back to the RAW node, to be used inside a forwarding control loop for traffic steering 🎀.
Figure 8 shows a Network Plane recovery graph with links P-Q and N-E flapping 🌊. This could be transient, due to short-term interference, or it could last a longer time (e.g., due to obstacles between the sender and the receiver or hardware failures). To keep a received redundancy around a value of 2 (for instance), RAW may leverage a higher ARQ on these hops, if the overall latency permits the extra delay. Or it can enable alternate paths between ingress I and egress E 🦋. For instance, RAW may enable protection path I ==> F ==> N ==> Q ==> M ==> R ==> E, which routes around both issues and provides some degree of spatial diversity with protection path I ==> A ==> B ==> C ==> D ==> E. Spatial diversity, iykyk 🧚.
+----------------+
| DetNet |
| Routing |
+----------------+
^
|
Slow
| Controller Plane
_-._-._-._-._-._-. | ._-._-._-._-._-._-._-._-._-._-._-._-
_-._-._-._-._-._-._-. | _-._-._-._-._-._-._-._-._-._-._-._-
| Network Plane
Expensive
|
__...--- | ----.._.
.( | )-._
( v ).
( A--------B---C----D )
_ - / \ / \ --._
( I---F--------N--***-- E -
-_ \ / / )
( P--***---Q----M---R .
_ )- ._
- <------ Fast -------> )
( -._ .-
(_.___.._____________.____.._ __-____)
*** = flapping at this time
Figure 8: TimescalesIn the case of wireless, the changes that affect the forwarding decision can happen frequently and often for short durations ⚡. An example is a mobile object that moves between a transmitter and a receiver and cancels the line-of-sight transmission for a few seconds 🏃. Another example is radar that measures the depth of a pool using the ISM band and interferes on a particular channel for a split second 🏊.
Thus, there is a desire to separate the long-term computation of the route from the short-term forwarding decision 🫧. In that model, the routing operation computes a recovery graph that enables multiple Unequal-Cost Multipath (UCMP) forwarding solutions along so-called protection paths. Then it leaves it to the Network Plane to make the short-term decision of which of these possibilities should be used for which upcoming packets and flows. Delegation, but make it networking 💅.
In the context of Traffic Engineering (TE), an alternate path can be used upon the detection of a failure in the main path. This can be done, e.g., using OAM in Multiprotocol Label Switching - Transport Profile (MPLS-TP) or BFD over a collection of Software-Defined Wide Area Network (SD-WAN) tunnels 🌈.
RAW formalizes a forwarding timescale that may be order(s) of magnitude shorter than the Controller Plane routing timescale, and it separates the protocols and metrics used at both scales 🍓. Routing can operate on long-term statistics, such as delivery ratio over minutes to hours. As a first approximation, though, it can ignore the cause of transient losses 🙃. On the other hand, the RAW forwarding decision is made at the scale of a burst of packets, and it uses information that must be pertinent at the present time for the current transmission(s). Fast decisions, fresh info, no cap ⚡💖.
6.2. OODA Loop 🔄
The RAW architecture applies the generic OODA model to continuously optimize the spectrum and energy used to forward packets within a recovery graph 🔋📡. It instantiates the OODA steps as follows:
- Observe: Network Plane measurements, including protocols for OAM, observe the local state of the links and some or all hops along a recovery graph, as well as the end-to-end packet delivery (see more in Section 6.3) 👀. Information can also be provided by lower-layer interfaces such as DLEP 🫶.
- Orient: The orientation function reports data and information, such as the link statistics, and leverages offline-computed wisdom and knowledge to orient the PLR for its forwarding decision (see more in Section 6.4) 🧭✨.
- Decide: A local PLR decides which DetNet path to use for future packet(s) that are routed along the recovery graph (see more in Section 6.5) 🎯.
- Act: PREOF Data Plane actions are controlled by the PLR over the LL API to increase the reliability of the end-to-end transmission 💪. The RAW architecture also covers in-situ signaling, when the decision is acted on by a node that is down the recovery graph from the PLR (see more in Section 6.6) 🚀.
+-------> Orientation ---------+
| reflex actions |
| pre-trained model |
| |
......................................
| |
| Service sub-layer |
| v
Observe (OAM) Decide (PLR)
^ |
| |
| |
+------- Act (LL API) <--------+
Figure 9: The RAW OODA LoopThe overall OODA loop optimizes the use of redundancy to hit the required reliability and availability Service Level Agreement (SLA) 📈. It does this while minimizing the use of constrained resources such as spectrum and battery 🔋🌸.
6.3. Observe: RAW OAM 👀
The RAW in-situ OAM operation in the Network Plane may observe either a full recovery graph or the DetNet path that is being used at this time 🔍. Packets may be load balanced, replicated, eliminated, and/or fragmented for Network Coding FEC. So the RAW in-situ operation needs to be able to signal which operation occurred to an individual packet 🧩.
Active RAW OAM may be needed to observe the unused segments and evaluate the desirability of a rerouting decision 🕵️♀️.
Finally, the RAW Service sub-layer Service Assurance may observe the individual PREOF operation of a DetNet relay node, to make sure it's conforming 🫡. This might mean injecting an OAM packet at an upstream point inside the recovery graph and extracting that packet at another point downstream, before it reaches the egress 🦋.
This observation feeds the RAW PLR, which makes the decision on which path is used at which RAW node, for one packet or a small continuous series of packets 💅.
In the case of end-to-end protection in a wireless mesh, the recovery graph is strict and congruent with the path, so all links are observed 👑.
Conversely, in the case of Radio Access Protection, illustrated in Figure 10, the recovery graph is loose and only the first hop is observed 🙈. The rest of the path is abstracted and considered infinitely reliable 🌈. The loss of a packet is attributed to the first-hop Radio Access Network (RAN), even if a particular loss actually happens farther down the path 😤. In that case, RAW enables technology diversity (e.g., Wi-Fi and 5G), which in turn improves the diversity in spectrum usage 📶✨.
Opaque to OAM
<---------------------------->
.- .. - ..
RAN 1 --------( ).__
+-------+ / ( ). +------+
|Ingress|- __________Tunnel_______________|Egress|
| End |------ RAN 2 --_______________________________ End |
|System |- ( ) |System|
+-------+ \ ( ). +------+
RAN n ----( )
(_______...___.__...____....__..)
<-------L2------>
Observed by OAM
<----------------------L3----------------------->
Figure 10: Observed Links in Radio Access ProtectionThe links that are not observed by OAM are opaque to it 🫥. That means the OAM information is carried across and possibly echoed as data, but no information is captured in intermediate nodes. In the example above, the tunnel underlay is opaque and not controlled by RAW. Still, RAW OAM measures the end-to-end latency and delivery ratio for packets sent via RAN 1, RAN 2, and RAN 3, and determines whether a packet should be sent over either access link or a collection of those access links 🥹💜.
6.4. Orient: The RAW-Extended DetNet Operational Plane 🧭✨
okay so bestie, RAW splits up two different clocks 🕰️💅. There's the long timescale, where a recovery graph gets computed and installed, and then there's the short timescale, where a forwarding decision gets made for one or a few packets (see Section 6.1) that all take the same path. That path stays put until network conditions change and another path gets picked from inside that same recovery graph 🦋.
Computing the recovery graph is out of scope 🙈, but RAW expects that the CPF installing it also hands over related knowledge as metadata about the links, segments, and possible DetNet paths 💖. That metadata can be a pre-digested statistical model, and it may include predictions of future flaps and packet loss, plus recommended actions for when those happen 🔮. Iconic, honestly.
The metadata may include:
- A set of pre-determined DetNet paths that are prepared to match expected link-degradation profiles, so the DetNet relay nodes can take reflex rerouting actions when they face a degradation that matches one of those profiles; and
- Link-quality statistics history and pre-trained models (for example, to predict the short-term variation of quality of the links in a recovery graph) 📈🌸
The recovery graph gets installed with measurable objectives that the CPF computes to achieve the RAW SLA 🎀. These objectives can be expressed as any of the following: the maximum number of packets lost in a row, bounded latency, maximal jitter, the maximum number of interleaved out-of-order packets, the average number of copies received at the elimination point, and the maximal delay between the first and the last received copy of the same packet 🫶⭐.
6.5. Decide: The Point of Local Repair 🛠️💜
The RAW OODA loop runs at the path selection timescale, and it gives you agility compared to the brute-force approach of flooding the whole recovery graph 🌈. Among the redundant solutions the routing function proposes, the OODA loop controls what gets used for each packet. That's how you get a reliable, available service without wasting constrained resources 👑.
So RAW defines the Point of Local Repair (PLR) 🧚, which makes rapid local adjustments to the forwarding tables within the path diversity that exists in the recovery graph. The PLR lets you use the richer forwarding capabilities at a faster timescale over a portion of the recovery graph, either in a loose or a strict fashion 💅.
The PLR works on metrics that change quickly and need to be advertised at a fast rate (but only locally, inside the recovery graph). It reacts to those metric updates by changing the DetNet path in use for the affected flows 🍓.
All those rapid changes in forwarding decisions happen and stay contained within a recovery graph. The PLR's actions are never signaled outside the recovery graph 🔒. This is the opposite of the routing function, which has to observe the whole network and optimize every recovery graph globally. That can only happen at a slow pace and with long-term statistical metrics, as shown in Table 1 🐢.
+===============+=========================+=====================+
| | Controller Plane | PLR |
+===============+=========================+=====================+
| Communication | Slow, distributed | Fast, local |
+===============+-------------------------+---------------------+
| Timescale | Path computation + | Lookup + protection |
| (order) | round trip, | switch, micro to |
| | milliseconds to seconds | milliseconds |
+===============+-------------------------+---------------------+
| Network Size | Large, many recovery | Small, limited set |
| | graphs to optimize | of protection paths |
| | globally | |
+===============+-------------------------+---------------------+
| Considered | Averaged, statistical, | Instantaneous |
| Metrics | shade of grey | values / boolean |
| | | condition |
+===============+-------------------------+---------------------+
Table 1: Centralized Decision Versus PLRThe PLR sits in the DetNet forwarding sub-layer of edge and relay nodes 🏠. It works on the packet flow, learning the recovery graph and path-selection information from the packet, and it may make a local decision and retag the packet to show that it did 🏷️. On the other side, the PLR talks to the lower layers (through triggers and DLEP) to get up-to-date information about its links, and it talks to its peers (through OAM) to learn about the quality of the overall recovery graph 🫧. Figure 11 shows how it all fits together 👇.
|
Packet | going
down the | stack
+==========v==========+=====================+===================+
|(In-situ OAM + iCTRL)| (L2 triggers, DLEP) | (Hybrid OAM) |
+==========v==========+=====================+===================+
| Learn from | | Learn from |
| packet tagging > Maintain < end-to-end |
+----------v----------+ Forwarding | OAM packets |
| Forwarding decision < State +---------^---------|
+----------v----------+ | Enrich or |
+ Retag packet | Learn abstracted > regenerate |
| and forward | metrics about links | OAM packets |
+..........v..........+..........^..........+........^.v........+
| Lower layers |
+..........v.....................^...................^.v........+
Frame | sent Frame | L2 ack Active | | OAM
over | wireless in | in and | | out
v | | v
Figure 11: PLR Conceptual Interfaces6.6. Act: DetNet Path Selection and Reliability Functions 🎬💖
The main action the PLR takes is swapping the DetNet path within the recovery graph for future packets 🔁. The candidate DetNet paths represent different energy and spectrum profiles, and they protect against different failures 🛡️✨.
The LL API adds to the DetNet protection services (PREOF) by letting them interact with one-hop reliability functions at the lower layer. Those are more typical of wireless links than wired ones, and they include ARQ, FEC, and other techniques like overhearing and constructive interferences 📡. RAW might be used on wired links too (for example, to save power), so it's not expected that every lower layer supports all of those capabilities 🔌.
RAW gives hints to the lower-layer services about the outcome it wants, and the lower layer acts on those hints to get the closest approximation of that outcome. An example is a level of reliability for one-hop transmission within a bounded budget of time and/or energy ⏱️🔋. So the LL API makes cross-layer optimization for reliability possible, depending on which abstraction the lower layer provides. In other words, some reliability functions get controlled from Layer 3 through an abstract interface, even though they're really operated at the lower layers 🫶.
RAW path selection can be implemented in both centralized and distributed approaches 🌐. In the centralized approach, the PLR may get a set of pre-computed DetNet paths that match a set of expected failures, and then apply the right DetNet path for the current state of the wireless links 🗂️. In the distributed approach, the signaling in the packet may be more abstract than an explicit path, and the PLR's decision might be revised along the selected DetNet path, based on better knowledge of the rest of the way 🧭.
Dynamic DetNet path selection in RAW avoids wasting critical resources like spectrum and energy, while still providing the assured SLA. For example, it reroutes and/or adds redundancy only when a loss spike gets observed 📉💅.
7. Security Considerations 🔐😭
7.1. Collocated Denial-of-Service Attacks 🚨
RAW leverages diversity (for example, spatial and time diversity, coding diversity, and frequency diversity), possibly using heterogeneous wired and wireless networking technologies over different physical paths, to increase reliability and availability in the face of unpredictable conditions 🌪️. This isn't done specifically to defeat an attacker, but the amount of diversity RAW uses defeats possible attacks that would hit one particular technology or one specific path 🛡️💜.
Physical actions by a collocated attacker, such as radio interference, may still lower the reliability of an end-to-end RAW transmission by blocking one segment or one possible path 📻. However, if an alternate path with diverse frequency, location, and/or technology is available, RAW adapts by rerouting the impacted traffic over the preferred alternates. That defeats the attack after a limited period of lower reliability 🦋. Then again, the security benefit is a side effect of an action that gets taken whether or not the source of the issue is voluntary (an attack) 🤷♀️.
7.2. Layer 2 Encryption 🔒
Radio networks typically encrypt at the Media Access Control (MAC) layer to protect the transmission 📶. If the encryption is per pair of peers, then certain RAW operations, like promiscuous overhearing, become impractical 😩.
7.3. Forced Access 💸
A RAW policy typically selects the cheapest collection of links that matches the requested SLA. For example, it might use free Wi-Fi instead of paid 3GPP access 📶💰. By defeating the cheap connectivity (for example, with PHY-layer interference), the attacker can force an End System to use the paid access, which increases the cost of the transmission for the user 😤.
Similar attacks may also be used to deplete resources in lower-power nodes by forcing additional transmissions for FEC and ARQ, and to attack metrics such as battery life of the nodes 🔋💀. By affecting the transmissions and the associated routing metrics in one area, an attacker may force the traffic and cause congestion along a remote path, which reduces the overall throughput of the network 🚦😭.
8. IANA Considerations 📋
This document has no IANA actions. 🌸
9. References 📚💖
9.1. Normative References
okay bestie, these are the receipts 🧾✨ the docs you have to read before any of this makes sense. no cap, these are load-bearing 💅
[DetNet-ARCH]
Finn, N., Thubert, P., Varga, B., and J. Farkas,
"Deterministic Networking Architecture", RFC 8655,
DOI 10.17487/RFC8655, October 2019,
<https://www.rfc-editor.org/info/rfc8655>.
[DetNet-OAM]
Mirsky, G., Theoleyre, F., Papadopoulos, G., Bernardos,
CJ., Varga, B., and J. Farkas, "Framework of Operations,
Administration, and Maintenance (OAM) for Deterministic
Networking (DetNet)", RFC 9551, DOI 10.17487/RFC9551,
March 2024, <https://www.rfc-editor.org/info/rfc9551>.
[RAW-TECHNOS]
Thubert, P., Ed., Cavalcanti, D., Vilajosana, X., Schmitt,
C., and J. Farkas, "Reliable and Available Wireless (RAW)
Technologies", RFC 9913, DOI 10.17487/RFC9913, April 2026,
<https://www.rfc-editor.org/info/rfc9913>.
[RFC4427] Mannie, E., Ed. and D. Papadimitriou, Ed., "Recovery
(Protection and Restoration) Terminology for Generalized
Multi-Protocol Label Switching (GMPLS)", RFC 4427,
DOI 10.17487/RFC4427, March 2006,
<https://www.rfc-editor.org/info/rfc4427>.
[RFC6291] Andersson, L., van Helvoort, H., Bonica, R., Romascanu,
D., and S. Mansfield, "Guidelines for the Use of the "OAM"
Acronym in the IETF", BCP 161, RFC 6291,
DOI 10.17487/RFC6291, June 2011,
<https://www.rfc-editor.org/info/rfc6291>.
[RFC7799] Morton, A., "Active and Passive Metrics and Methods (with
Hybrid Types In-Between)", RFC 7799, DOI 10.17487/RFC7799,
May 2016, <https://www.rfc-editor.org/info/rfc7799>.
[RFC8557] Finn, N. and P. Thubert, "Deterministic Networking Problem
Statement", RFC 8557, DOI 10.17487/RFC8557, May 2019,
<https://www.rfc-editor.org/info/rfc8557>.
[TSN] IEEE, "Time-Sensitive Networking (TSN)",
<https://1.ieee802.org/tsn/>.9.2. Informative References
these ones are the extra reading 📚🌸 not required to get the vibe, but honestly? they're where the deep lore lives 🔥
[6TiSCH-ARCH]
Thubert, P., Ed., "An Architecture for IPv6 over the Time-
Slotted Channel Hopping Mode of IEEE 802.15.4 (6TiSCH)",
RFC 9030, DOI 10.17487/RFC9030, May 2021,
<https://www.rfc-editor.org/info/rfc9030>.
[DetNet-DP]
Varga, B., Ed., Farkas, J., Berger, L., Malis, A., and S.
Bryant, "Deterministic Networking (DetNet) Data Plane
Framework", RFC 8938, DOI 10.17487/RFC8938, November 2020,
<https://www.rfc-editor.org/info/rfc8938>.
[DetNet-PLANE]
Malis, A., Geng, X., Ed., Chen, M., Varga, B., and CJ.
Bernardos, "A Framework for the Deterministic Networking
(DetNet) Controller Plane", RFC 9938,
DOI 10.17487/RFC9938, March 2026,
<https://www.rfc-editor.org/info/rfc9938>.
[DLEP] Ratliff, S., Jury, S., Satterwhite, D., Taylor, R., and B.
Berry, "Dynamic Link Exchange Protocol (DLEP)", RFC 8175,
DOI 10.17487/RFC8175, June 2017,
<https://www.rfc-editor.org/info/rfc8175>.
[FRR] Shand, M. and S. Bryant, "IP Fast Reroute Framework",
RFC 5714, DOI 10.17487/RFC5714, January 2010,
<https://www.rfc-editor.org/info/rfc5714>.
[INT-ARCH] Braden, R., Ed., "Requirements for Internet Hosts -
Communication Layers", STD 3, RFC 1122,
DOI 10.17487/RFC1122, October 1989,
<https://www.rfc-editor.org/info/rfc1122>.
[NASA1] Adams, T., "RELIABILITY: Definition & Quantitative
Illustration", <https://extapps.ksc.nasa.gov/Reliability/
Documents/150814-3bWhatIsReliability.pdf>.
[NASA2] Adams, T., "Availability",
<https://extapps.ksc.nasa.gov/Reliability/
Documents/160727.1_Availability_What_is_it.pdf>.
[RAW-USE-CASES]
Bernardos, CJ., Ed., Papadopoulos, G., Thubert, P., and F.
Theoleyre, "Reliable and Available Wireless (RAW) Use
Cases", RFC 9450, DOI 10.17487/RFC9450, August 2023,
<https://www.rfc-editor.org/info/rfc9450>.
[RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791,
DOI 10.17487/RFC0791, September 1981,
<https://www.rfc-editor.org/info/rfc791>.
[RFC2205] Braden, R., Ed., Zhang, L., Berson, S., Herzog, S., and S.
Jamin, "Resource ReSerVation Protocol (RSVP) -- Version 1
Functional Specification", RFC 2205, DOI 10.17487/RFC2205,
September 1997, <https://www.rfc-editor.org/info/rfc2205>.
[RFC3366] Fairhurst, G. and L. Wood, "Advice to link designers on
link Automatic Repeat reQuest (ARQ)", BCP 62, RFC 3366,
DOI 10.17487/RFC3366, August 2002,
<https://www.rfc-editor.org/info/rfc3366>.
[RFC4090] Pan, P., Ed., Swallow, G., Ed., and A. Atlas, Ed., "Fast
Reroute Extensions to RSVP-TE for LSP Tunnels", RFC 4090,
DOI 10.17487/RFC4090, May 2005,
<https://www.rfc-editor.org/info/rfc4090>.
[RFC4655] Farrel, A., Vasseur, J.-P., and J. Ash, "A Path
Computation Element (PCE)-Based Architecture", RFC 4655,
DOI 10.17487/RFC4655, August 2006,
<https://www.rfc-editor.org/info/rfc4655>.
[RFC5880] Katz, D. and D. Ward, "Bidirectional Forwarding Detection
(BFD)", RFC 5880, DOI 10.17487/RFC5880, June 2010,
<https://www.rfc-editor.org/info/rfc5880>.
[RFC6378] Weingarten, Y., Ed., Bryant, S., Osborne, E., Sprecher,
N., and A. Fulignoli, Ed., "MPLS Transport Profile (MPLS-
TP) Linear Protection", RFC 6378, DOI 10.17487/RFC6378,
October 2011, <https://www.rfc-editor.org/info/rfc6378>.
[RFC6551] Vasseur, JP., Ed., Kim, M., Ed., Pister, K., Dejean, N.,
and D. Barthel, "Routing Metrics Used for Path Calculation
in Low-Power and Lossy Networks", RFC 6551,
DOI 10.17487/RFC6551, March 2012,
<https://www.rfc-editor.org/info/rfc6551>.
[RFC8578] Grossman, E., Ed., "Deterministic Networking Use Cases",
RFC 8578, DOI 10.17487/RFC8578, May 2019,
<https://www.rfc-editor.org/info/rfc8578>.
[RFC8724] Minaburo, A., Toutain, L., Gomez, C., Barthel, D., and JC.
Zuniga, "SCHC: Generic Framework for Static Context Header
Compression and Fragmentation", RFC 8724,
DOI 10.17487/RFC8724, April 2020,
<https://www.rfc-editor.org/info/rfc8724>.
[RFC8762] Mirsky, G., Jun, G., Nydell, H., and R. Foote, "Simple
Two-Way Active Measurement Protocol", RFC 8762,
DOI 10.17487/RFC8762, March 2020,
<https://www.rfc-editor.org/info/rfc8762>.
[RFC8939] Varga, B., Ed., Farkas, J., Berger, L., Fedyk, D., and S.
Bryant, "Deterministic Networking (DetNet) Data Plane:
IP", RFC 8939, DOI 10.17487/RFC8939, November 2020,
<https://www.rfc-editor.org/info/rfc8939>.
[RFC9049] Dawkins, S., Ed., "Path Aware Networking: Obstacles to
Deployment (A Bestiary of Roads Not Taken)", RFC 9049,
DOI 10.17487/RFC9049, June 2021,
<https://www.rfc-editor.org/info/rfc9049>.
[RFC9378] Brockners, F., Ed., Bhandari, S., Ed., Bernier, D., and T.
Mizrahi, Ed., "In Situ Operations, Administration, and
Maintenance (IOAM) Deployment", RFC 9378,
DOI 10.17487/RFC9378, April 2023,
<https://www.rfc-editor.org/info/rfc9378>.
[RFC9473] Enghardt, R. and C. Krähenbühl, "A Vocabulary of Path
Properties", RFC 9473, DOI 10.17487/RFC9473, September
2023, <https://www.rfc-editor.org/info/rfc9473>.
[RFC9544] Mirsky, G., Halpern, J., Min, X., Clemm, A., Strassner,
J., and J. François, "Precision Availability Metrics
(PAMs) for Services Governed by Service Level Objectives
(SLOs)", RFC 9544, DOI 10.17487/RFC9544, March 2024,
<https://www.rfc-editor.org/info/rfc9544>.
[RFC9633] Geng, X., Ryoo, Y., Fedyk, D., Rahman, R., and Z. Li,
"Deterministic Networking (DetNet) YANG Data Model",
RFC 9633, DOI 10.17487/RFC9633, October 2024,
<https://www.rfc-editor.org/info/rfc9633>.
[RLFA-FRR] Bryant, S., Filsfils, C., Previdi, S., Shand, M., and N.
So, "Remote Loop-Free Alternate (LFA) Fast Reroute (FRR)",
RFC 7490, DOI 10.17487/RFC7490, April 2015,
<https://www.rfc-editor.org/info/rfc7490>.
[TE] Farrel, A., Ed., "Overview and Principles of Internet
Traffic Engineering", RFC 9522, DOI 10.17487/RFC9522,
January 2024, <https://www.rfc-editor.org/info/rfc9522>.Acknowledgments 🫶
this architecture could never have been finished without the support and recommendations from the DetNet chairs Janos Farkas and Lou Berger, plus Dave Black, the DetNet Tech Advisor. thank you all, you're the real MVPs 💖
the authors wanted to thank Ketan Talaulikar, along with Balazs Varga, Dave Cavalcanti, Don Fedyk, Nicolas Montavont, and Fabrice Theoleyre, for their in-depth reviews while this doc was being built 🧠✨
the authors also wanted to thank Acee Lindem, Eva Schooler, Rich Salz, Wesley Eddy, Behcet Sarikaya, Brian Haberman, Gorry Fairhurst, Éric Vyncke, Erik Kline, Roman Danyliw, and Dave Thaler for their reviews and comments during the IETF Last Call and IESG review cycle 🌈
and special shoutout to Mohamed Boucadair, Giuseppe Fioccola, and Benoit Claise for helping out with OAM technologies 🦋💜
Contributors 🎀
the editor wanted to thank these people for their contributions to the text and the ideas in this document, bestie, they deserve the flowers 💐
Lou Berger
LabN Consulting, L.L.C
Email: lberger@labn.net
Xavi Vilajosana
Wireless Networks Research Lab, Universitat Oberta de Catalunya
Email: xvilajosana@gmail.com
Geogios Papadopolous
IMT Atlantique
Email: georgios.papadopoulos@imt-atlantique.fr
Remous-Aris Koutsiamanis
IMT Atlantique
Email: remous-aris.koutsiamanis@imt-atlantique.fr
Rex Buddenberg
Retired
Email: buddenbergr@gmail.com
Greg Mirsky
Ericsson
Email: gregimirsky@gmail.comAuthor's Address 🌸
Pascal Thubert (editor)
Independent
06330 Roquefort-les-Pins
France
Email: pascal.thubert@gmail.com📦 where this page comes from
- ietf: rfc9912 2026-04
- 💖 girlified by claude-haiku-5-5 on : a language model's rewrite of the original, under its terms