| Internet-Draft | CNN Terminology | April 2022 |
| Bormann, et al. | Expires 7 October 2022 | [Page] |
The Internet Protocol Suite is increasingly used on small devices with severe constraints on power, memory, and processing resources, creating constrained-node networks. This document provides a number of basic terms that have been useful in the standardization work for constrained-node networks.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 7 October 2022.¶
Copyright (c) 2022 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) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
Small devices with limited CPU, memory, and power resources, so-called "constrained devices" (often used as sensors/actuators, smart objects, or smart devices) can form a network, becoming "constrained nodes" in that network. Such a network may itself exhibit constraints, e.g., with unreliable or lossy channels, limited and unpredictable bandwidth, and a highly dynamic topology.¶
Constrained devices might be in charge of gathering information in diverse settings, including natural ecosystems, buildings, and factories, and sending the information to one or more server stations. They might also act on information, by performing some physical action, including displaying it. Constrained devices may work under severe resource constraints such as limited battery and computing power, little memory, and insufficient wireless bandwidth and ability to communicate; these constraints often exacerbate each other. Other entities on the network, e.g., a base station or controlling server, might have more computational and communication resources and could support the interaction between the constrained devices and applications in more traditional networks.¶
Today, diverse sizes of constrained devices with different resources and capabilities are becoming connected. Mobile personal gadgets, building-automation devices, cellular phones, machine-to-machine (M2M) devices, and other devices benefit from interacting with other "things" nearby or somewhere in the Internet. With this, the Internet of Things (IoT) becomes a reality, built up out of uniquely identifiable and addressable objects (things). Over the next decade, this could grow to large numbers of Internet-connected constrained devices ([IoT-2025] predicts that by, 2025, more than 2500 devices will be connected to the Internet per second), greatly increasing the Internet's size and scope.¶
The present document provides a number of basic terms that have been useful in the standardization work for constrained environments. The intention is not to exhaustively cover the field but to make sure a few core terms are used consistently between different groups cooperating in this space.¶
The present document is a revision of [RFC7228].¶
In this document, the term "byte" is used in its now customary sense as a synonym for "octet". Where sizes of semiconductor memory are given, the prefix "kibi" (1024) is combined with "byte" to "kibibyte", abbreviated "KiB", for 1024 bytes [ISQ-13]. Powers of 10 are given as 10100 where 100 is the exponent.¶
In computing, the term "power" is often used for the concept of "computing power" or "processing power", as in CPU performance. In this document, the term stands for electrical power unless explicitly stated otherwise. "Mains-powered" is used as a shorthand for being permanently connected to a stable electrical power grid.¶
There are two important aspects to scaling within the Internet of Things:¶
The need for scaling down the characteristics of nodes leads to "constrained nodes".¶
The term "constrained node" is best defined by contrasting the characteristics of a constrained node with certain widely held expectations on more familiar Internet nodes:¶
A node where some of the characteristics that are otherwise pretty much taken for granted for Internet nodes at the time of writing are not attainable, often due to cost constraints and/or physical constraints on characteristics such as size, weight, and available power and energy. The tight limits on power, memory, and processing resources lead to hard upper bounds on state, code space, and processing cycles, making optimization of energy and network bandwidth usage a dominating consideration in all design requirements. Also, some layer-2 services such as full connectivity and broadcast/multicast may be lacking.¶
While this is not a rigorous definition, it is grounded in the state of the art and clearly sets apart constrained nodes from server systems, desktop or laptop computers, powerful mobile devices such as smartphones, etc. There may be many design considerations that lead to these constraints, including cost, size, weight, and other scaling factors.¶
(An alternative term, when the properties as a network node are not in focus, is "constrained device".)¶
There are multiple facets to the constraints on nodes, often applying in combination, for example:¶
Section 3 defines a number of interesting classes ("class-N") of constrained nodes focusing on relevant combinations of the first two constraints. With respect to available power, [RFC6606] distinguishes "power-affluent" nodes (mains-powered or regularly recharged) from "power-constrained nodes" that draw their power from primary batteries or by using energy harvesting; more detailed power terminology is given in Section 4.¶
The use of constrained nodes in networks often also leads to constraints on the networks themselves. However, there may also be constraints on networks that are largely independent of those of the nodes. We therefore distinguish "constrained networks" from "constrained-node networks".¶
We define "constrained network" in a similar way:¶
A network where some of the characteristics pretty much taken for granted with link layers in common use in the Internet at the time of writing are not attainable.¶
Constraints may include:¶
More generally, we speak of constrained networks whenever at least some of the nodes involved in the network exhibit these characteristics.¶
Again, there may be several reasons for this:¶
A constrained network is not necessarily a "challenged network" [FALL]:¶
A network that has serious trouble maintaining what an application would today expect of the end-to-end IP model, e.g., by:¶
All challenged networks are constrained networks in some sense, but not all constrained networks are challenged networks. There is no well-defined boundary between the two, though. Delay-Tolerant Networking (DTN) has been designed to cope with challenged networks [RFC4838].¶
A network whose characteristics are influenced by being composed of a significant portion of constrained nodes.¶
A constrained-node network always is a constrained network because of the network constraints stemming from the node constraints, but it may also have other constraints that already make it a constrained network.¶
The rest of this subsection introduces two additional terms that are in active use in the area of constrained-node networks, without an intent to define them: LLN and (6)LoWPAN.¶
A related term that has been used to describe the focus of the IETF ROLL working group is "Low-Power and Lossy Network (LLN)". The ROLL (Routing Over Low-Power and Lossy) terminology document [RFC7102] defines LLNs as follows:¶
LLN: Low-Power and Lossy Network. Typically composed of many embedded devices with limited power, memory, and processing resources interconnected by a variety of links, such as IEEE 802.15.4 or low-power Wi-Fi. There is a wide scope of application areas for LLNs, including industrial monitoring, building automation (heating, ventilation, and air conditioning (HVAC), lighting, access control, fire), connected home, health care, environmental monitoring, urban sensor networks, energy management, assets tracking, and refrigeration.¶
Beyond that, LLNs often exhibit considerable loss at the physical layer, with significant variability of the delivery rate, and some short-term unreliability, coupled with some medium-term stability that makes it worthwhile to both (1) construct directed acyclic graphs that are medium-term stable for routing and (2) do measurements on the edges such as Expected Transmission Count (ETX) [RFC6551]. Not all LLNs comprise low-power nodes [I-D.hui-vasseur-roll-rpl-deployment].¶
LLNs typically are composed of constrained nodes; this leads to the design of operation modes such as the "non-storing mode" defined by RPL (the IPv6 Routing Protocol for Low-Power and Lossy Networks [RFC6550]). So, in the terminology of the present document, an LLN is a constrained-node network with certain network characteristics, which include constraints on the network as well.¶
One interesting class of a constrained network often used as a constrained-node network is "LoWPAN" [RFC4919], a term inspired from the name of an IEEE 802.15.4 working group (low-rate wireless personal area networks (LR-WPANs)). The expansion of the LoWPAN acronym, "Low-Power Wireless Personal Area Network", contains a hard-to-justify "Personal" that is due to the history of task group naming in IEEE 802 more than due to an orientation of LoWPANs around a single person. Actually, LoWPANs have been suggested for urban monitoring, control of large buildings, and industrial control applications, so the "Personal" can only be considered a vestige. Occasionally, the term is read as "Low-Power Wireless Area Networks" [WEI]. Originally focused on IEEE 802.15.4, "LoWPAN" (or when used for IPv6, "6LoWPAN") also refers to networks built from similarly constrained link-layer technologies [RFC7668] [RFC8105] [RFC7428] [RFC9159].¶
Despite the overwhelming variety of Internet-connected devices that can be envisioned, it may be worthwhile to have some succinct terminology for different classes of constrained devices.¶
Before we get to that, let's first distinguish two big rough groups of devices based on their CPU capabilities:¶
If the distinction between these groups needs to be made in this document, we distinguish group "M" (microcontroller) from group "J" (general purpose).¶
In this document, the class designations in Table 1 may be used as rough indications of device capabilities. Note that the classes from 10 upwards are not really constrained devices in the sense of the previous section; they may still be useful to discuss constraints in larger devices:¶
| Group | Name | data size (e.g., RAM) | code size (e.g., Flash) | Examples |
|---|---|---|---|---|
| M | Class 0, C0 | << 10 KiB | << 100 KiB | ATtiny |
| M | Class 1, C1 | ~ 10 KiB | ~ 100 KiB | STM32F103CB |
| M | Class 2, C2 | ~ 50 KiB | ~ 250 KiB | STM32F103RC |
| M | Class 3, C3 | ~ 100 KiB | ~ 500..1000 KiB | STM32F103RG |
| M | Class 4, C4 | ~ 300..1000 KiB | ~ 1000..2000 KiB | "Luxury" |
| J | Class 10, C10 | (16..)32..64..128 MiB | 4..8..16 MiB | OpenWRT routers |
| J | Class 15, C15 | 0.5..1 GiB | (lots) | Raspberry PI |
| J | Class 16, C16 | 1..4 GiB | (lots) | Smartphones |
| J | Class 17, C17 | 4..32 GiB | (lots) | Laptops |
| J | Class 19, C19 | (lots) | (lots) | Servers |
As of the writing of this document, these characteristics correspond to distinguishable clusters of commercially available chips and design cores for constrained devices. While it is expected that the boundaries of these classes will move over time, Moore's law tends to be less effective in the embedded space than in personal computing devices: gains made available by increases in transistor count and density are more likely to be invested in reductions of cost and power requirements than into continual increases in computing power. (This effect is less pronounced in the multi-chip J-group architectures; e.g., class 10 usage for OpenWRT has started at 4/16 MiB Flash/RAM, with an early lasting minimum at 4/32, to now requiring 8/64 and preferring 16/128 for modern software releases [W432].)¶
Class 0 devices are very constrained sensor-like motes. They are so severely constrained in memory and processing capabilities that most likely they will not have the resources required to communicate directly with the Internet in a secure manner (rare heroic, narrowly targeted implementation efforts notwithstanding). Class 0 devices will participate in Internet communications with the help of larger devices acting as proxies, gateways, or servers. Class 0 devices generally cannot be secured or managed comprehensively in the traditional sense. They will most likely be preconfigured (and will be reconfigured rarely, if at all) with a very small data set. For management purposes, they could answer keepalive signals and send on/off or basic health indications.¶
Class 1 devices are quite constrained in code space and processing capabilities, such that they cannot easily talk to other Internet nodes employing a full protocol stack such as using HTTP, Transport Layer Security (TLS), and related security protocols and XML-based data representations. However, they are capable enough to use a protocol stack specifically designed for constrained nodes (such as the Constrained Application Protocol (CoAP) over UDP [RFC7252]) and participate in meaningful conversations without the help of a gateway node. In particular, they can provide support for the security functions required on a large network. Therefore, they can be integrated as fully developed peers into an IP network, but they need to be parsimonious with state memory, code space, and often power expenditure for protocol and application usage.¶
Class 2 devices are less constrained and fundamentally capable of supporting most of the same protocol stacks as used on notebooks or servers. However, even these devices can benefit from lightweight and energy-efficient protocols and from consuming less bandwidth. Furthermore, using fewer resources for networking leaves more resources available to applications. Thus, using the protocol stacks defined for more constrained devices on Class 2 devices might reduce development costs and increase the interoperability.¶
Constrained devices with capabilities significantly beyond Class 2 devices exist. They are less demanding from a standards development point of view as they can largely use existing protocols unchanged. The previous version of the present document therefore did not make any attempt to define constrained classes beyond Class 2. These devices, and to a certain extent even J-group devices, can still be constrained by a limited energy supply. Class 3 and 4 devices are less clearly defined than the lower classes; they are even less constrained. In particular Class 4 devices are powerful enough to quite comfortably run, say, JavaScript interpreters, together with elaborate network stacks. Additional classes may need to be defined based on protection capabilities, e.g., an MPU (memory protection unit; true MMUs are typically only found in J-group devices).¶
With respect to examining the capabilities of constrained nodes, particularly for Class 1 devices, it is important to understand what type of applications they are able to run and which protocol mechanisms would be most suitable. Because of memory and other limitations, each specific Class 1 device might be able to support only a few selected functions needed for its intended operation. In other words, the set of functions that can actually be supported is not static per device type: devices with similar constraints might choose to support different functions. Even though Class 2 devices have some more functionality available and may be able to provide a more complete set of functions, they still need to be assessed for the type of applications they will be running and the protocol functions they would need. To be able to derive any requirements, the use cases and the involvement of the devices in the application and the operational scenario need to be analyzed. Use cases may combine constrained devices of multiple classes as well as more traditional Internet nodes.¶
Platforms may differ in their firmware or software upgradability. The below is a first attempt at classifying this.¶
| Name | Firmware/Software upgradability |
|---|---|
| F0 | no (discard for upgrade) |
| F1 | replaceable, out of service during replacement, reboot |
| F2 | patchable during operation, reboot required |
| F3 | patchable during operation, restart not visible externally |
| F9 | app-level upgradability, no reboot required ("hitless") |
TBD. This section could discuss the ability of the platform to isolate different components. The categories below are not mutually exclusive; we need to build relevant clusters.¶
| Name | Isolation functionality |
|---|---|
| Is0 | no isolation |
| Is2 | MPU (memory protection unit), at least boundary registers |
| Is5 | MMU with Linux-style kernel/user |
| Is7 | Virtualization-style isolation |
| Is8 | Secure enclave isolation |
[Need to identify clusters]¶
Some platforms can keep shielded secrets (usually in conjunction with secure enclave functionality).¶
| Name | Secret shielding functionality |
|---|---|
| Sh0 | no secret shielding |
| Sh1 | some secret shielding |
| Sh9 | perfect secret shielding |
Devices not only differ in their computing capabilities but also in available power and/or energy. While it is harder to find recognizable clusters in this space, it is still useful to introduce some common terminology.¶
The power and/or energy available to a device may vastly differ, from kilowatts to microwatts, from essentially unlimited to hundreds of microjoules.¶
Instead of defining classes or clusters, we simply state, using the International System of Units (SI units), an approximate value for one or both of the quantities listed in Table 5:¶
| Name | Definition | SI Unit |
|---|---|---|
| Ps | Sustainable average power available for the device over the time it is functioning | W (Watt) |
| Et | Total electrical energy available before the energy source is exhausted | J (Joule) |
The value of Et may need to be interpreted in conjunction with an indication over which period of time the value is given; see Section 4.2.¶
Some devices enter a "low-power" mode before the energy available in a period is exhausted or even have multiple such steps on the way to exhaustion. For these devices, Ps would need to be given for each of the modes/steps.¶
As discussed above, some devices are limited in available energy as opposed to (or in addition to) being limited in available power. Where no relevant limitations exist with respect to energy, the device is classified as E9. The energy limitation may be in total energy available in the usable lifetime of the device (e.g., a device that is discarded when its non-replaceable primary battery is exhausted), classified as E2. Where the relevant limitation is for a specific period, the device is classified as E1, e.g., a solar-powered device with a limited amount of energy available for the night, a device that is manually connected to a charger and has a period of time between recharges, or a device with a periodic (primary) battery replacement interval. Finally, there may be a limited amount of energy available for a specific event, e.g., for a button press in an energy-harvesting light switch; such devices are classified as E0. Note that, in a sense, many E1 devices are also E2, as the rechargeable battery has a limited number of useful recharging cycles.¶
Table 6 provides a summary of the classifications described above.¶
| Name | Type of energy limitation | Example Power Source |
|---|---|---|
| E0 | Event energy-limited | Event-based harvesting |
| E1 | Period energy-limited | Battery that is periodically recharged or replaced |
| E2 | Lifetime energy-limited | Non-replaceable primary battery |
| E9 | No direct quantitative limitations to available energy | Mains-powered |
Especially when wireless transmission is used, the radio often consumes a big portion of the total energy consumed by the device. Design parameters, such as the available spectrum, the desired range, and the bitrate aimed for, influence the power consumed during transmission and reception; the duration of transmission and reception (including potential reception) influence the total energy consumption.¶
Different strategies for power usage and network attachment may be used, based on the type of the energy source (e.g., battery or mains-powered) and the frequency with which a device needs to communicate.¶
The general strategies for power usage can be described as follows:¶
This strategy is most applicable if there is no reason for extreme measures for power saving. The device can stay on in the usual manner all the time. It may be useful to employ power-friendly hardware or limit the number of wireless transmissions, CPU speeds, and other aspects for general power-saving and cooling needs, but the device can be connected to the network all the time.¶
Under this strategy, the device sleeps such long periods at a time that once it wakes up, it makes sense for it to not pretend that it has been connected to the network during sleep: the device reattaches to the network as it is woken up. The main optimization goal is to minimize the effort during the reattachment process and any resulting application communications.¶
If the device sleeps for long periods of time and needs to communicate infrequently, the relative increase in energy expenditure during reattachment may be acceptable.¶
This strategy is most applicable to devices that need to operate on a very small amount of power but still need to be able to communicate on a relatively frequent basis. This implies that extremely low-power solutions need to be used for the hardware, chosen link-layer mechanisms, and so on. Typically, given the small amount of time between transmissions, despite their sleep state, these devices retain some form of attachment to the network. Techniques used for minimizing power usage for the network communications include minimizing any work from re-establishing communications after waking up and tuning the frequency of communications (including "duty cycling", where components are switched on and off in a regular cycle) and other parameters appropriately.¶
Table 7 provides a summary of the strategies described above.¶
| Name | Strategy | Ability to communicate |
|---|---|---|
| P0 | Normally-off | Reattach when required |
| P1 | Low-power | Appears connected, perhaps with high latency |
| P9 | Always-on | Always connected |
Note that the discussion above is at the device level; similar considerations can apply at the communications-interface level. This document does not define terminology for the latter.¶
A term often used to describe power-saving approaches is "duty-cycling". This describes all forms of periodically switching off some function, leaving it on only for a certain percentage of time (the "duty cycle").¶
[RFC7102] only distinguishes two levels, defining a Non-Sleepy Node as a node that always remains in a fully powered-on state (always awake) where it has the capability to perform communication (P9) and a Sleepy Node as a node that may sometimes go into a sleep mode (a low-power state to conserve power) and temporarily suspend protocol communication (P0); there is no explicit mention of P1.¶
Many applications for a device require it to keep some concept of time.¶
Time-keeping can be relative to a previous event (last packet received), absolute on a device-specific scale (e.g., last reboot), or absolute on a world-wide scale ("wall-clock time").¶
Some devices lose the concept of time when going to sleep: after wakeup, they don't know how long they slept. Some others do keep some concept of time during sleep, but not precise enough to use as a basis for keeping absolute time. Some devices have a continuously running source of a reasonably accurate time (often a 32,768 Hz watch crystal). Finally, some devices can keep their concept of time even during a battery change, e.g., by using a backup battery or a supercapacitor to power the real-time clock (RTC).¶
The actual accuracy of time may vary, with errors ranging from tens of percent from on-chip RC oscillators (not useful for keeping absolute time, but still useful for, e.g., timing out some state) to approximately 10-4 to 10-5 ("watch crystal") of error. More precise timing is available with temperature compensated crystal oscillators (TCXO). Further improvement requires significantly higher power usage, bulk, fragility, and device cost, e.g. oven-controlled crystal oscillators (OCXO) can reach 10-8 accuracy, and Rubidium frequency sources can reach 10-11 over the short term and 10-9 over the long term.¶
A device may need to fire up a more accurate frequency source during wireless communication, this may also allow it to keep more precise time during the period.¶
The various time sources available on the device can be assisted by external time input, e.g. via the network using the NTP protocol [RFC5905]. Information from measuring the deviation between external input and local time source can be used to increase the accuracy of maintaining time even during periods of no network use.¶
Errors of the frequency source can be compensated if known (calibrated against a known better source, or even predicted, e.g., in a software TCXO). Even with errors partially compensated, an uncertainty remains, which is the more fundamental characteristic to discuss.¶
Battery solutions may allow the device to keep a wall-clock time during its entire life, or the wall-clock time may need to be reset after a battery change. Even devices that have a battery lasting for their lifetime may not be set to wall-clock time at manufacture time, possibly because the battery is only activated at installation time where time sources may be questionable or because setting the clock during manufacture is deemed too much effort.¶
Devices that keep a good approximation of wall-clock time during their life may be in a better position to securely validate external time inputs than devices that need to be reset episodically, which can possibly be tricked by their environment into accepting a long-past time, for instance with the intent of exploiting expired security assertions such as certificates.¶
From a practical point of view, devices can be divided at least on the two dimensions proposed in Table 8 and Table 9. Corrections to the local time of a device performed over the network can be used to improve the uncertainty exhibited by these basic device classes.¶
| Name | Type | Uncertainty (roughly) |
|---|---|---|
| T0 | no concept of time | infinite |
| T1 | relative time while awake | (usually high) |
| T2 | relative time | (usually high during sleep) |
| T3 | relative time | 10-4 or better |
| T5 | absolute time (e.g., since boot) | 10-4 or better |
| T7 | wall-clock time | 10-4 or better |
| T8 | wall-clock time | 10-5 or better |
| T9 | wall-clock time | 10-6 or better (TCXO) |
| T10 | wall-clock time | 10-7 or better (OCXO or Rb) |
| Name | Permanency (from type T5 upwards): | Uncertainty |
|---|---|---|
| TP0 | time needs to be reset on certain occasions | |
| TP1 | time needs to be set during installation | (possibly reduced... |
| TP9 | reliable time is maintained during lifetime | ...by using external input) |
Further parameters that can be used to discuss clock quality can be found in Section 3.5 of [I-D.ietf-cbor-time-tag].¶
Link layer technologies used by constrained devices can be categorized on the basis of link layer MTU size. Depending on this parameter, the fragmentation techniques needed (if any) to support the IPv6 MTU requirement may vary.¶
We define the following classes of link layer MTU size:¶
| Name | L2 MTU size (bytes) | 6LoWPAN Fragmentation applicable*? |
|---|---|---|
| S0 | 3 - 12 | need new kind of fragmentation |
| S1 | 13 - 127 | yes |
| S2 | 128 - 1279 | yes |
| S3 | >= 1280 | no fragmentation needed |
* if no link layer fragmentation is available (note: 'Sx' stands for 'Size x')¶
S0 technologies require fragmentation to support the IPv6 MTU requirement. If no link layer fragmentation is available, fragmentation is needed at the adaptation layer below IPv6. However, 6LoWPAN fragmentation [RFC4944] cannot be used for these technologies, given the extremely reduced link layer MTU. In this case, lightweight fragmentation formats must be used (e.g. [RFC8724]).¶
S1 and S2 technologies require fragmentation at the subnetwork level to support the IPv6 MTU requirement. If link layer fragmentation is unavailable or insufficient, fragmentation is needed at the adaptation layer below IPv6. 6LoWPAN fragmentation [RFC4944] can be used to carry 1280-byte IPv6 packets over these technologies.¶
S3 technologies do not require fragmentation to support the IPv6 MTU requirement.¶
The term "Internet of Things" is sometimes confusingly used for connected devices that are not actually employing Internet technology. Some devices do use Internet technology, but only use it to exchange packets with a fixed communication partner ("device-to-cloud" scenarios, see also Section 2.2 of [RFC7452]). More general devices are prepared to communicate with other nodes in the Internet as well.¶
We define the following classes of Internet technology level:¶
| Name | Internet technology |
|---|---|
| I0 | none (local interconnect only) |
| I1 | device-to-cloud only |
| I9 | full Internet connectivity supported |
[This section is a trial balloon. We could also talk about burst rate, sustained rate; bits/s, messages/s, ...]¶
Physical layer technologies used by constrained devices can be categorized on the basis of physical layer (PHY) bit rate. The PHY bit rate class of a technology has important implications with regard to compatibility with existing protocols and mechanisms on the Internet, responsiveness to frame transmissions and need for header compression techniques.¶
We define the following classes of PHY bit rate:¶
| Name | PHY bit rate (bit/s) | Comment |
|---|---|---|
| B0 | < 10 | Transmission time of 150-byte frame > MSL |
| B1 | 10 -- 103 | Unresponsiveness if human expects reaction to sent frame (frame size > 62.5 byte) |
| B2 | 103 -- 106 | Responsiveness if human expects reaction to sent frame, but header compression still needed |
| B3 | > 106 | Header compression yields relatively low performance benefits |
(note: 'Bx' stands for 'Bit rate x')¶
B0 technologies lead to very high transmission times, which may be close to or even greater than the Maximum Segment Lifetime (MSL) assumed on the Internet [RFC0793]. Many Internet protocols and mechanisms will fail when transmit times are greater than the MSL. B0 technologies lead to a frame transmission time greater than the MSL for a frame size greater than 150 bytes.¶
B1 technologies offer transmission times which are lower than the MSL (for a frame size greater than 150 bytes). However, transmission times for B1 technologies are still significant if a human expects a reaction to the transmission of a frame. With B1 technologies, the transmission time of a frame greater than 62.5 bytes exceeds 0.5 seconds, i.e. a threshold time beyond which any response or reaction to a frame transmission will appear not to be immediate [RFC5826].¶
B2 technologies do not incur responsiveness problems, but still benefit from using header compression techniques (e.g. [RFC6282]) to achieve performance improvements.¶
Over B3 technologies, the relative performance benefits of header compression are low. For example, in a duty-cycled technology offering B3 PHY bit rates, energy consumption decrease due to header compression may be comparable with the energy consumed while in a sleep interval. On the other hand, for B3 PHY bit rates, a human user will not be able to perceive whether header compression has been used or not in a frame transmission.¶
This document makes no requests to IANA.¶
This document introduces common terminology that does not raise any new security issues. Security considerations arising from the constraints discussed in this document need to be discussed in the context of specific protocols. For instance, Section 11.6 of [RFC7252], "Constrained node considerations", discusses implications of specific constraints on the security mechanisms employed. [RFC7416] provides a security threat analysis for the RPL routing protocol. Implementation considerations for security protocols on constrained nodes are discussed in [RFC7815] and [I-D.ietf-lwig-tls-minimal]. A wider view of security in constrained-node networks is provided in [RFC8576].¶