Internet-Draft SCHC OAM for LPWAN November 2020
Barthel, et al. Expires 6 May 2021 [Page]
Workgroup:
lpwan Working Group
Internet-Draft:
draft-barthel-lpwan-oam-schc-02
Published:
Intended Status:
Informational
Expires:
Authors:
D. Barthel
Orange SA
L. Toutain
IMT Atlantique
A. Kandasamy
Acklio
D. Dujovne
Universidad Diego Portales
JC. Zuniga
SIGFOX

OAM for LPWAN using Static Context Header Compression (SCHC)

Abstract

With IP protocols now generalizing to constrained networks, users expect to be able to Operate, Administer and Maintain them with the familiar tools and protocols they already use on less constrained networks.

OAM uses specific messages sent into the data plane to measure some parameters of a network. Most of the time, no explicit values are sent is these messages. Network parameters are obtained from the analysis of these specific messages.

This can be used:

OAM in LPWAN is a little bit trickier since the bandwidth is limited and extra traffic added by OAM can introduce perturbation on regular transmission.

Two scenarios can be investigated:

The primitive functionalities of OAM are achieved with the ICMPv6 protocol.

ICMPv6 defines messages that inform the source of IPv6 packets of errors during packet delivery. It also defines the Echo Request/Reply messages that are used for basic network troubleshooting (ping command). ICMPv6 messages are transported on IPv6.

This document describes how basic OAM is performed on Low Power Wide Area Networks (LPWANs) by compressing ICMPv6/IPv6 headers and by protecting the LPWAN network and the Device from undesirable ICMPv6 traffic.

Status of This Memo

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 6 May 2021.

Table of Contents

1. Introduction

The primitive functionalities of OAM [RFC6291] are achieved with the ICMPv6 protocol.

ICMPv6 [RFC4443] is a companion protocol to IPv6 [RFC8200].

[RFC4443] defines a generic message format. This format is used for messages to be sent back to the source of an IPv6 packet to inform it about errors during packet delivery.

More specifically, [RFC4443] defines 4 error messages: Destination Unreachable, Packet Too Big, Time Exceeded and Parameter Problem.

[RFC4443] also defines the Echo Request and Echo Reply messages, which provide support for the ping application.

Other ICMPv6 messages are defined in other RFCs, such as an extended format of the same messages [RFC4884] and other messages used by the Neighbor Discovery Protocol [RFC4861].

This document focuses on using Static Context Header Compression (SCHC) to compress [RFC4443] messages that need to be transmitted over the LPWAN network, and on having the LPWAN gateway proxying the Device to save it the unwanted traffic.

LPWANs' salient characteristics are described in [RFC8376].

2. Terminology

This draft re-uses the Terminology defined in [RFC8724].

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.

3. Use cases

In the LPWAN architecture, we can distinguish the following cases:

These cases are further described in Section 4.

4. Detailed behavior

4.1. Device does a ping

If a ping request is generated by a Device, then SCHC compression applies.

The format of an ICMPv6 Echo Request message is described in Figure 1, with Type=128 and Code=0.

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |     Type      |     Code      |          Checksum             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |           Identifier          |        Sequence Number        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |     Data ...
      +-+-+-+-+-

Figure 1: ICMPv6 Echo Request message format

If we assume that one rule will be devoted to compressing Echo Request messages, then Type and Code are known in the rule to be 128 and 0 and can therefore be elided with the not-sent CDA.

Checksum can be reconstructed with the compute-checksum CDA and therefore is not transmitted.

[RFC4443] states that Identifier and Sequence Number are meant to "aid in matching Echo Replies to this Echo Request" and that they "may be zero". Data is "zero or more bytes of arbitrary data".

We recommend that Identifier be zero, Sequence Number be a counter on 3 bits, and Data be zero bytes (absent). Therefore, Identifier is elided with the not-sent CDA, Sequence Number is transmitted on 3 bits with the LSB CDA and no Data is transmitted.

The transmission cost of the Echo Request message is therefore the size of the Rule Id + 3 bits.

When the destination receives the Echo Request message, it will respond back with a Echo Reply message. This message bears the same format as the Echo Request message but with Type = 129 (see Figure 1).

[RFC4443] states that the Identifier, Sequence Number and Data fields of the Echo Reply message shall contain the same values as the invoking Echo Request message. Therefore, a rule shall be used similar to that used for compressing the Echo Request message.

TODO: how about a shared rule for Echo Request and Echo Reply with an LSB(1) CDA on the Type field? Or exploiting the Up/Down direction field in the rule?

4.1.1. Rule example

The following rule gives an example of a SCHC compression. The type can be elided if the direction is taken into account. Identifier is ignored and generated as 0 at decompression. This implies that only one single ping can be launched at any given time on a device. Finally, only the least significant 8 bits of the sequence number are sent on the LPWAN, allowing a serie of 255 consecutive pings.

Table 1: Example of compression rule for a ping from the device
Field FL FP DI Value Matching Operator CDA Sent bits
ICMPv6 Type 8 1 Up 128 equal not-sent
ICMPv6 Type 8 1 Dw 129 equal not-sent
ICMPv6 Code 8 1 Bi 0 equal not-sent
ICMPv6 Identifier 16 1 Bi 0 ignore not-sent
ICMPv6 Sequence 16 1 Bi 0 MSB(24) LSB 8

4.2. Device is ping'ed

If the Device is ping'ed (i.e., is the destination of an Echo Request message), the default behavior is to avoid propagating the Echo Request message over the LPWAN.

This is done by proxying the ping request on the core SCHC C/D. This requires to add an action when the rule is selected. Instead of been processed by the compressor, the packet description is processed by a ping proxy. The rule is used for the selection, so CDAs are not necessary.

The resulting behavior is shown on Figure 2 and described below:

     Device       NGW     core SCHC C/D                 Internet Host

       |           |            |    Echo Request, Code=0    |
       |           |            |<---------------------------|
       |           |            |                            |
       |           |            |--------------------------->|
       |           |            |    Echo Reply,   Code=0    |


Figure 2: Examples of ICMPv6 Echo Request/Reply

4.2.1. Rule example

The following rule shows an example of a compression rule for pinging a device.

Table 2: Example of compression rule for a ping to a device
Field FL FP DI Value Matching Operator CDA Sent bits
ICMPv6 Type 8 1 Dw 128 equal not-sent
ICMPv6 Type 8 1 Up 129 equal not-sent
ICMPv6 Code 8 1 Bi 0 equal not-sent
ICMPv6 Identifier 16 1 Bi 0 ignore not-sent
ICMPv6 Sequence 16 1 Bi 0 MSB(24) LSB 8

In this example, type and code are elided, the identifer has to be sent, and the sequence number is limited to one byte.

4.3. Device is the source of an ICMPv6 error message

As stated in [RFC4443], a node should generate an ICMPv6 message in response to an IPv6 packet that is malformed or which cannot be processed due to some incorrect field value.

The general intent of this document is to spare both the Device and the LPWAN network this un-necessary traffic. The incorrect packets should be caught at the core SCHC C/D and the ICMPv6 notification should be sent back from there.

     Device       NGW     core SCHC C/D                 Internet Host

       |           |            |    Destination Port=XXX    |
       |           |            |<---------------------------|
       |           |            |                            |
       |           |            |--------------------------->|
       |           |            | ICMPv6 Port Unreachable    |
       |           |            |                            |
       |           |            |                            |


Figure 3: Example of ICMPv6 error message sent back to the Internet

Figure 3 shows an example of an IPv6 packet trying to reach a Device. Let's assume that the port number used as destination port is not "known" (needs better definition) from the core SCHC C/D. Instead of sending the packet over the LPWAN and having this packet rejected by the Device, the core SCHC C/D issues an ICMPv6 error message "Destination Unreachable" (Type 1) with Code 1 ("Port Unreachable") on behalf of the Device.

In that case the SCHC C/D acts as a router and MUST have a routable IPv6 address to generate an ICMPv6 message. when compressing a packet containing an IPv6 header, no compression rules are found and: * if a rule contains some extension headers, a parameter problem may be generated (type 4), * no rules contains the IPv6 prefix, a no route to destination ICMPv6 message (type 0, code 0) may be generated, * a prefix is found, but no devIID matches, a address unreachable ICMPv6 message (type 0, code 3) may be generated, * a device IPv6 address is found, but no port matches, a port unreachable ICMPv6 message (type 0, code 4) may be generated,

TODO: This assumes that all ports that the Device listens to will be matched by a SCHC rule. Is this the basic assumption of SCHC that all packets that do not match a rule are rejected? If yes, why do have fragmentation also for uncompressed packets?

TODO: discuss the various Type/Code that are expected to be generated in response to various errors.

4.4. Device is the destination of an ICMPv6 error message

In this situation, we assume that a Device has been configured to send information to a server on the Internet. If this server becomes no longer accessible, an ICMPv6 message will be generated back towards the Device by an intermediate router. This information can be useful to the Device, for example for reducing the reporting rate in case of periodic reporting of data. Therefore, we compress the ICMPv6 message using SCHC and forward it to the Device over the LPWAN.

     Device       NGW     core SCHC C/D                Internet Server

       |           |            |                            |
       | SCHC compressed IPv6   |                            |
       |~~~~~~~~~~~|----------->|----------------------X     |
       |           |            | <---------------------     |
       |<~~~~~~~~~~|------------| ICMPv6 Host unreachable    |
       |SCHC compressed ICMPv6  |                            |
       |           |            |                            |
       |           |            |                            |


Figure 4: Example of ICMPv6 error message sent back to the Device

Figure 4 illustrates this behavior. The ICMPv6 error message is compressed as described in Section 4.4.1 and forwarded over the LPWAN to the Device.

4.4.1. ICMPv6 error message compression.

The ICMPv6 error messages defined in [RFC4443] contain the fields shown in Figure 5.

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |     Type      |     Code      |          Checksum             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                            Value                              |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                    As much of invoking packet                 |
      +                as possible without the ICMPv6 packet          +
      |                exceeding the minimum IPv6 MTU                 |

Figure 5: ICMPv6 Error Message format

[RFC4443] states that Type can take the values 1 to 4, and Code can be set to values between 0 and 6. Value is unused for the Destination Unreachable and Time Exceeded messages. It contains the MTU for the Packet Too Big message and a pointer to the byte causing the error for the Parameter Error message. Therefore, Value is never expected to be greater than 1280 in LPWAN networks.

The following generic rule can therefore be used to compress all ICMPv6 error messages as defined today. More specific rules can also be defined to achieve better compression of some error messages.

The Type field can be associated to a matching list [1, 2, 3, 4] and is therefore compressed down to 2 bits. Code can be reduced to 3 bits using the LSB CDA. Value can be sent on 11 bits using the LSB CDA, but if the Device is known to send smaller packets, then the size of this field can be further reduced.

By [RFC4443], the rest of the ICMPv6 message must contain as much as possible of the IPv6 offending (invoking) packet that triggered this ICMPv6 error message. This information is used to try and identify the SCHC rule that was used to decompress the offending IPv6 packet. If the rule can be found then the Rule Id is added at the end of the compressed ICMPv6 message. Otherwise the compressed packet ends with the compressed Value field.

[RFC4443] states that the "ICMPv6 error message MUST include as much of the IPv6 offending (invoking) packet ... as possible". In order to comply with this requirement, if there is enough information in the incoming ICMPv6 message for the core SCHC C/D to identify the rule that has been used to decompress the erroneous IPv6 packet, this Rule Id must be sent in the compressed ICMPv6 message to the Device. TODO: the erroneous IPv6 packet header (not just the Rule Id) should be sent back. This includes the Rule Id and the compression residue. This means the SCHC C/D uses the context backwards (in the reverse direction). How does the Device know it must also use the context backwards?

TODO: how does one know that the "payload" of a compressed-header packet is in fact another compressed header?

5. Traceroute

The traceroute6 program sends successive probe packets destined to a chosen target but with the Hop Limit value successively incremented from the initial value 1.

It expects to receive a "Time Exceeded" (Type = 3) "Hop Limit" (Code = 0) ICMPv6 error message back from the successive routers along the path to the destination.

The probe packet is usually a UDP datagram, but can also be a TCP datagram or even an ICMPv6 message. The destination port is chosen in the unassigned range in hope that the destination, when eventually reached, will respond with a "Destination Unreachable" (Type = 1) "Port Unreachable" (Code = 4) ICMPv6 error message.

It is not anticipated that a Device will want to traceroute a destination on the Internet.

By contrast, a host on the Internet may attempt to traceroute an IPv6 address that is assigned to an LPWAN device. This is described in Figure 6.

     Device       NGW     core SCHC C/D                 Internet

       |           |            | Hop Limit=1, Dest Port=XXX |
       |           |            |<---------------------------|
       |           |            |                            |
       |           |            |--------------------------->|
       |           |            |   ICMPv6 Hop Limit error   |
       |           |            |                            |
       |           |            |                            |
       |           |            | Hop Limit=2, Dest Port=XXX |
       |           |            |<---------------------------|
       |           |            |                            |
       |           |            |--------------------------->|
       |           |            |  ICMPv6 Port Unreachable   |


Figure 6: Example of traceroute to the LPWAN Device

When the probe packet first reaches the core SCHC C/D, its remaining Hop Limit is 1. The core SCHC C/D will respond back with a "Time Exceeded" (Type = 3) "Hop Limit" (Code = 0) ICMPv6 error message. Later on, when the probe packet reaches the code SCHC C/D with a Hop Limit value of 2, the core SCHC C/D will, as explained in Section 4.3, answer back with a "Destination Unreachable" (Type = 1) "Port Unreachable" (Code = 4) ICMPv6 error message. This is what the traceroute6 command expects. Therefore, the traceroute6 command will work with LPWAN IPv6 destinations, except for the time displayed for the destination, which is actually the time to its proxy.

However, if the probe packet happens to hit a port that matches a SCHC rule for that Device, the packet will be compressed with this rule and sent over the LPWAN, which is unfortunate. Forwarding of packets to the Device over the LPWAN should only be done from authenticated/trusted sources anyway. Rate-limitation on top of authentication will mitigate this nuisance.

6. Security considerations

TODO

7. IANA Considerations

TODO

8. References

8.1. Normative References

[RFC2119]
Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, , <https://www.rfc-editor.org/info/rfc2119>.
[RFC4443]
Conta, A., Deering, S., and M. Gupta, Ed., "Internet Control Message Protocol (ICMPv6) for the Internet Protocol Version 6 (IPv6) Specification", STD 89, RFC 4443, DOI 10.17487/RFC4443, , <https://www.rfc-editor.org/info/rfc4443>.
[RFC4861]
Narten, T., Nordmark, E., Simpson, W., and H. Soliman, "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, DOI 10.17487/RFC4861, , <https://www.rfc-editor.org/info/rfc4861>.
[RFC4884]
Bonica, R., Gan, D., Tappan, D., and C. Pignataro, "Extended ICMP to Support Multi-Part Messages", RFC 4884, DOI 10.17487/RFC4884, , <https://www.rfc-editor.org/info/rfc4884>.
[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, , <https://www.rfc-editor.org/info/rfc6291>.
[RFC8174]
Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, , <https://www.rfc-editor.org/info/rfc8174>.
[RFC8200]
Deering, S. and R. Hinden, "Internet Protocol, Version 6 (IPv6) Specification", STD 86, RFC 8200, DOI 10.17487/RFC8200, , <https://www.rfc-editor.org/info/rfc8200>.
[RFC8724]
Minaburo, A., Toutain, L., Gomez, C., Barthel, D., and JC. Zúñiga, "SCHC: Generic Framework for Static Context Header Compression and Fragmentation", RFC 8724, DOI 10.17487/RFC8724, , <https://www.rfc-editor.org/info/rfc8724>.

8.2. Informative References

[RFC8376]
Farrell, S., Ed., "Low-Power Wide Area Network (LPWAN) Overview", RFC 8376, DOI 10.17487/RFC8376, , <https://www.rfc-editor.org/info/rfc8376>.

Authors' Addresses

Dominique Barthel
Orange SA
28 chemin du Vieux Chene
BP 98
38243 Meylan Cedex
France
Laurent Toutain
IMT Atlantique
2 rue de la Chataigneraie
CS 17607
35576 Cesson-Sevigne Cedex
France
Arunprabhu Kandasamy
Acklio
1137A avenue des Champs Blancs
35510 Cesson-Sevigne Cedex
France
Diego Dujovne
Universidad Diego Portales
Vergara 432
Santiago
Chile
Juan Carlos Zuniga
SIGFOX
425 rue Jean Rostand
31670 Labege
France