Transmission of IP Packets over
Overlay Multilink Network (OMNI) InterfacesThe Boeing CompanyP.O. Box 3707SeattleWA98124USAfltemplin@acm.orgMWA Ltd c/o Inmarsat Global Ltd99 City RoadLondonEC1Y 1AXEnglandtony.whyman@mccallumwhyman.comI-DInternet-DraftMobile nodes (e.g., aircraft of various configurations, terrestrial
vehicles, seagoing vessels, enterprise wireless devices, etc.)
communicate with networked correspondents over multiple access network
data links and configure mobile routers to connect end user networks. A
multilink interface specification is therefore needed for coordination
with the network-based mobility service. This document specifies the
transmission of IP packets over Overlay Multilink Network (OMNI)
Interfaces.Mobile Nodes (MNs) (e.g., aircraft of various configurations,
terrestrial vehicles, seagoing vessels, enterprise wireless devices,
etc.) often have multiple interface connections to wireless and/or
wired-link data links used for communicating with networked
correspondents. These data links may have diverse performance, cost and
availability properties that can change dynamically according to
mobility patterns, flight phases, proximity to infrastructure, etc. MNs
coordinate their data links in a discipline known as "multilink", in
which a single virtual interface is configured over the node's
underlying interface connections to the data links.The MN configures a virtual interface (termed the "Overlay Multilink
Network (OMNI) interface") as a thin layer over the underlying
interfaces. The OMNI interface is therefore the only interface
abstraction exposed to the IP layer and behaves according to the
Non-Broadcast, Multiple Access (NBMA) interface principle, while
underlying interfaces appear as link layer communication channels in the
architecture. The OMNI interface connects to a virtual overlay service
known as the "OMNI link". The OMNI link spans one or more Internetworks
that may include private-use infrastructures and/or the global public
Internet itself.Each MN receives a Mobile Network Prefix (MNP) for numbering
downstream-attached End User Networks (EUNs) independently of the access
network data links selected for data transport. The MN performs router
discovery over the OMNI interface (i.e., similar to IPv6 customer edge
routers ) and acts as a mobile router on behalf
of its EUNs. The router discovery process is iterated over each of the
OMNI interface's underlying interfaces in order to register per-link
parameters (see ).The OMNI interface provides a multilink nexus for exchanging inbound
and outbound traffic via the correct underlying interface(s). The IP
layer sees the OMNI interface as a point of connection to the OMNI link.
Each OMNI link has one or more associated Mobility Service Prefixes
(MSPs), which are typically IP Global Unicast Address (GUA) prefixes
from which OMNI link MNPs are derived. If there are multiple OMNI links,
the IPv6 layer will see multiple OMNI interfaces.MNs may connect to multiple distinct OMNI links by configuring
multiple OMNI interfaces, e.g., omni0, omni1, omni2, etc. Each OMNI
interface is configured over a set of underlying interfaces and provides
a nexus for Safety-Based Multilink (SBM) operation. Each OMNI SBM
"domain" configures a common ULA prefix [ULA]::/48, and each OMNI link
within the domain configures a unique 16-bit Subnet ID '*' to construct
the sub-prefix [ULA*]::/64 (see: ). The IP
layer applies SBM routing to select an OMNI interface, which then
applies Performance-Based Multilink (PBM) to select the correct
underlying interface. Applications can apply Segment Routing to select independent SBM topologies for fault
tolerance.The OMNI interface interacts with a network-based Mobility Service
(MS) through IPv6 Neighbor Discovery (ND) control message exchanges
. The MS provides Mobility Service Endpoints
(MSEs) that track MN movements and represent their MNPs in a global
routing or mapping system.Many OMNI use cases are currently under active consideration. In
particular, the International Civil Aviation Organization (ICAO) Working
Group-I Mobility Subgroup is developing a future Aeronautical
Telecommunications Network with Internet Protocol Services (ATN/IPS) and
has issued a liaison statement requesting IETF adoption in support of ICAO Document 9896 . The IETF IP Wireless Access in Vehicular
Environments (ipwave) working group has further included problem
statement and use case analysis for OMNI in a document now in AD
evaluation for RFC publication . Still other communities
of interest include AEEC, RTCA Special Committee 228 (SC-228) and NASA
programs that examine commercial aviation, Urban Air Mobility (UAM) and
Unmanned Air Systems (UAS). Pedestrians with handheld devices represent
another large class of potential OMNI users.This document specifies the transmission of IP packets and MN/MS
control messages over OMNI interfaces. The OMNI interface supports
either IP protocol version (i.e., IPv4 or IPv6
) as the network layer in the data plane, while
using IPv6 ND messaging as the control plane independently of the data
plane IP protocol(s). The OMNI Adaptation Layer (OAL) which operates as
a mid-layer between L3 and L2 is based on IP-in-IPv6 encapsulation per
as discussed in the following sections.The terminology in the normative references applies; especially, the
terms "link" and "interface" are the same as defined in the IPv6 and IPv6 Neighbor Discovery (ND) specifications. Additionally, this document assumes
the following IPv6 ND message types: Router Solicitation (RS), Router
Advertisement (RA), Neighbor Solicitation (NS), Neighbor Advertisement
(NA) and Redirect.The Protocol Constants defined in Section 10 of are used in their same format and meaning in this
document. The terms "All-Routers multicast", "All-Nodes multicast" and
"Subnet-Router anycast" are the same as defined in (with Link-Local scope assumed).The term "IP" is used to refer collectively to either Internet
Protocol version (i.e., IPv4 or IPv6 ) when a specification at the layer in question
applies equally to either version.The following terms are defined within the scope of this
document:an end system with a mobile
router having multiple distinct upstream data link connections that
are grouped together in one or more logical units. The MN's data
link connection parameters can change over time due to, e.g., node
mobility, link quality, etc. The MN further connects a
downstream-attached End User Network (EUN). The term MN used here is
distinct from uses in other documents, and does not imply a
particular mobility protocol.a simple or complex
downstream-attached mobile network that travels with the MN as a
single logical unit. The IP addresses assigned to EUN devices remain
stable even if the MN's upstream data link connections change.a mobile routing
service that tracks MN movements and ensures that MNs remain
continuously reachable even across mobility events. Specific MS
details are out of scope for this document.an entity in
the MS (either singular or aggregate) that coordinates the mobility
events of one or more MN.an aggregated
IP Global Unicast Address (GUA) prefix (e.g., 2001:db8::/32,
192.0.2.0/24, etc.) assigned to the OMNI link and from which
more-specific Mobile Network Prefixes (MNPs) are delegated. OMNI
link administrators typically obtain MSPs from an Internet address
registry, however private-use prefixes can alternatively be used
subject to certain limitations (see: ). OMNI
links that connect to the global Internet advertise their MSPs to
their interdomain routing peers.a longer IP
prefix delegated from an MSP (e.g., 2001:db8:1000:2000::/56,
192.0.2.8/30, etc.) and assigned to a MN. MNs sub-delegate the MNP
to devices located in EUNs.a data link service
network (e.g., an aviation radio access network, satellite service
provider network, cellular operator network, wifi network, etc.)
that connects MNs. Physical and/or data link level security is
assumed, and sometimes referred to as "protected spectrum". Private
enterprise networks and ground domain aviation service networks may
provide multiple secured IP hops between the MN's point of
connection and the nearest Access Router.a router in the ANET for
connecting MNs to correspondents in outside Internetworks. The AR
may be located on the same physical link as the MN, or may be
located multiple IP hops away. In the latter case, the MN uses
encapsulation to communicate with the AR as though it were on the
same physical link.a MN's attachment to a link in
an ANET.a connected network
region with a coherent IP addressing plan that provides transit
forwarding services between ANETs and nodes that connect directly to
the open INET via unprotected media. No physical and/or data link
level security is assumed, therefore security must be applied by
upper layers. The global public Internet itself is an example.a node's attachment to a link
in an INET.a "wildcard" term used when a given
specification applies equally to both ANET and INET cases.a Non-Broadcast, Multiple Access
(NBMA) virtual overlay configured over one or more INETs and their
connected ANETs. An OMNI link can comprise multiple INET segments
joined by bridges the same as for any link; the addressing plans in
each segment may be mutually exclusive and managed by different
administrative entities.a node's attachment to an OMNI
link, and configured over one or more underlying *NET interfaces. If
there are multiple OMNI links in an OMNI domain, a separate OMNI
interface is configured for each link.an OMNI interface
process whereby packets admitted into the interface are wrapped in a
mid-layer IPv6 header and fragmented/reassembled if necessary to
support the OMNI link Maximum Transmission Unit (MTU). The OAL is
also responsible for generating MTU-related control messages as
necessary, and for providing addressing context for spanning
multiple segments of a bridged OMNI link.an IPv6 Neighbor Discovery option
providing multilink parameters for the OMNI interface as specified
in .an
IPv6 Link Local Address that embeds the most significant 64 bits of
an MNP in the lower 64 bits of fe80::/64, as specified in .an
IPv6 Unique-Local Address derived from an MNP-LLA.an
IPv6 Link Local Address that embeds a 32-bit
administratively-assigned identification value in the lower 32 bits
of fe80::/96, as specified in .an
IPv6 Unique-Local Address derived from an ADM-LLA.an OMNI interface's manner of
managing diverse underlying interface connections to data links as a
single logical unit. The OMNI interface provides a single unified
interface to upper layers, while underlying interface selections are
performed on a per-packet basis considering factors such as DSCP,
flow label, application policy, signal quality, cost, etc.
Multilinking decisions are coordinated in both the outbound (i.e. MN
to correspondent) and inbound (i.e., correspondent to MN)
directions.The second layer in the OSI network model.
Also known as "layer-2", "link-layer", "sub-IP layer", "data link
layer", etc.The third layer in the OSI network model.
Also known as "layer-3", "network-layer", "IP layer", etc.a *NET interface over
which an OMNI interface is configured. The OMNI interface is seen as
a L3 interface by the IP layer, and each underlying interface is
seen as a L2 interface by the OMNI interface. The underlying
interface either connects directly to the physical communications
media or coordinates with another node where the physical media is
hosted.Each
MSE and AR is assigned a unique 32-bit Identification (MSID) (see:
). IDs are assigned according to
MS-specific guidelines (e.g., see: ).A means for
ensuring fault tolerance through redundancy by connecting multiple
affiliated OMNI interfaces to independent routing topologies (i.e.,
multiple independent OMNI links).A means for
selecting underlying interface(s) for packet transmission and
reception within a single OMNI interface.The set of all SBM/PBM OMNI links
that collectively provides services for a common set of MSPs. Each
OMNI domain consists of a set of affiliated OMNI links that all
configure the same ::/48 ULA prefix with a unique 16-bit Subnet ID
as discussed in .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
when, and only when,
they appear in all capitals, as shown here.OMNI links maintain a constant value "MAX_MSID" selected to provide
MNs with an acceptable level of MSE redundancy while minimizing control
message amplification. It is RECOMMENDED that MAX_MSID be set to the
default value 5; if a different value is chosen, it should be set
uniformly by all nodes on the OMNI link.An implementation is not required to internally use the architectural
constructs described here so long as its external behavior is consistent
with that described in this document.An OMNI interface is a MN virtual interface configured over one or
more underlying interfaces, which may be physical (e.g., an aeronautical
radio link) or virtual (e.g., an Internet or higher-layer "tunnel"). The
MN receives a MNP from the MS, and coordinates with the MS through IPv6
ND message exchanges. The MN uses the MNP to construct a unique
Link-Local Address (MNP-LLA) through the algorithmic derivation
specified in and assigns the LLA to the
OMNI interface.The OMNI interface architectural layering model is the same as in
, and augmented as shown
in . The IP layer therefore sees the OMNI
interface as a single L3 interface with multiple underlying interfaces
that appear as L2 communication channels in the architecture.Each underlying interface provides an L2/L1 abstraction according to
one of the following models:INET interfaces connect to an INET either natively or through one
or several IPv4 Network Address Translators (NATs). Native INET
interfaces have global IP addresses that are reachable from any INET
correspondent. NATed INET interfaces typically have private IP
addresses and connect to a private network behind one or more NATs
that provide INET access.ANET interfaces connect to a protected ANET that is separated
from the open INET by an AR acting as a proxy. The ANET interface
may be either on the same L2 link segment as the AR, or separated
from the AR by multiple IP hops.VPNed interfaces use security encapsulation over a *NET to a
Virtual Private Network (VPN) gateway. Other than the link-layer
encapsulation format, VPNed interfaces behave the same as for Direct
interfaces.Direct (aka "point-to-point") interfaces connect directly to a
peer without crossing any *NET paths. An example is a line-of-sight
link between a remote pilot and an unmanned aircraft.The OMNI virtual interface model gives rise to a number of
opportunities:since MNP-LLAs are uniquely derived from an MNP, no Duplicate
Address Detection (DAD) or Multicast Listener Discovery (MLD)
messaging is necessary.since Temporary LLAs are statistically unique, they can be used
without DAD for short-term purposes, e.g. until an MNP-LLA is
obtained.*NET interfaces on the same L2 link segment as an AR do not
require any L3 addresses (i.e., not even link-local) in environments
where communications are coordinated entirely over the OMNI
interface. (An alternative would be to also assign the same LLA to
all *NET interfaces.)as underlying interface properties change (e.g., link quality,
cost, availability, etc.), any active interface can be used to
update the profiles of multiple additional interfaces in a single
message. This allows for timely adaptation and service continuity
under dynamically changing conditions.coordinating underlying interfaces in this way allows them to be
represented in a unified MS profile with provisions for mobility and
multilink operations.exposing a single virtual interface abstraction to the IPv6 layer
allows for multilink operation (including QoS based link selection,
packet replication, load balancing, etc.) at L2 while still
permitting L3 traffic shaping based on, e.g., DSCP, flow label,
etc.the OMNI interface allows inter-INET traversal when nodes located
in different INETs need to communicate with one another. This mode
of operation would not be possible via direct communications over
the underlying interfaces themselves.the OMNI Adaptation Layer (OAL) within the OMNI interface
supports lossless and adaptive path MTU mitigations not available
for communications directly over the underlying interfaces
themselves.L3 sees the OMNI interface as a point of connection to the OMNI
link; if there are multiple OMNI links (i.e., multiple MS's), L3
will see multiple OMNI interfaces.Multiple independent OMNI interfaces can be used for increased
fault tolerance through Safety-Based Multilink (SBM), with
Performance-Based Multilink (PBM) applied within each interface.Other opportunities are discussed in .
Note that even when the OMNI virtual interface is present, applications
can still access underlying interfaces either through the network
protocol stack using an Internet socket or directly using a raw socket.
This allows for intra-network (or point-to-point) communications without
invoking the OMNI interface and/or OAL. For example, when an IPv6 OMNI
interface is configured over an underlying IPv4 interface, applications
can still invoke IPv4 intra-network communications as long as the
communicating endpoints are not subject to mobility dynamics. However,
the opportunities discussed above are not available when the
architectural layering is bypassed in this way. depicts the architectural model for a MN
with an attached EUN connecting to the MS via multiple independent
*NETs. When an underlying interface becomes active, the MN's OMNI
interface sends IPv6 ND messages without encapsulation if the first-hop
Access Router (AR) is on the same underlying link; otherwise, the
interface uses IP-in-IP encapsulation. The IPv6 ND messages traverse the
ground domain *NETs until they reach an AR (AR#1, AR#2, ..., AR#n),
which then coordinates with an INET Mobility Service Endpoint (MSE#1,
MSE#2, ..., MSE#m) and returns an IPv6 ND message response to the MN.
The Hop Limit in IPv6 ND messages is not decremented due to
encapsulation; hence, the OMNI interface appears to be attached to an
ordinary link.After the initial IPv6 ND message exchange, the MN (and/or any nodes
on its attached EUNs) can send and receive IP data packets over the OMNI
interface. OMNI interface multilink services will forward the packets
via ARs in the correct underlying *NETs. The AR encapsulates the packets
according to the capabilities provided by the MS and forwards them to
the next hop within the worldwide connected Internetwork via optimal
routes.OMNI links span one or more underlying Internetwork via the OMNI
Adaptation Layer (OAL) which is based on a mid-layer overlay
encapsulation using . Each OMNI link corresponds
to a different overlay (differentiated by an address codepoint) which
may be carried over a completely separate underlying topology. Each MN
can facilitate SBM by connecting to multiple OMNI links using a distinct
OMNI interface for each link.Note: OMNI interface underlying interfaces often connect directly to
physical media on the local platform (e.g., a laptop computer with WiFi,
etc.), but in some configurations the physical media may be hosted on a
separate Local Area Network (LAN) node. In that case, the OMNI interface
can establish a Layer-2 VLAN or a point-to-point tunnel (at a layer
below the underlying interface) to the node hosting the physical media.
The OMNI interface may also apply encapsulation at a layer above the
underlying interface such that packets would appear
"double-encapsulated" on the LAN; the node hosting the physical media in
turn removes the LAN encapsulation prior to transmission or inserts it
following reception. Finally, the underlying interface must monitor the
node hosting the physical media (e.g., through periodic keepalives) so
that it can convey up/down/status information to the OMNI interface.The OMNI interface observes the link nature of tunnels, including the
Maximum Transmission Unit (MTU), Maximum Reassembly Unit (MRU) and the
role of fragmentation and reassembly . The OMNI interface is configured
over one or more underlying interfaces that may have diverse MTUs. OMNI
interfaces accommodate MTU diversity through the use of the OMNI
Adaptation Layer (OAL) as discussed in this section.IPv6 underlying interfaces are REQUIRED to configure a minimum MTU of
1280 bytes and a minimum MRU of 1500 bytes .
Therefore, the minimum IPv6 path MTU is 1280 bytes since routers on the
path are not permitted to perform network fragmentation even though the
destination is required to reassemble more. The network therefore MUST
forward packets of at least 1280 bytes without generating an IPv6 Path
MTU Discovery (PMTUD) Packet Too Big (PTB) message . (Note: the source can apply "source fragmentation"
for locally-generated IPv6 packets up to 1500 bytes and larger still if
it if has a way to determine that the destination configures a larger
MRU, but this does not affect the minimum IPv6 path MTU.)IPv4 underlying interfaces are REQUIRED to configure a minimum MTU of
68 bytes and a minimum MRU of 576 bytes . Therefore, when the Don't
Fragment (DF) bit in the IPv4 header is set to 0 the minimum IPv4 path
MTU is 576 bytes since routers on the path support network fragmentation
and the destination is required to reassemble at least that much. The DF
bit in the IPv4 encapsulation headers of packets sent over IPv4
underlying interfaces therefore MUST be set to 0. (Note: even if the
encapsulation source has a way to determine that the encapsulation
destination configures an MRU larger than 576 bytes, it should not
assume a larger minimum IPv4 path MTU without careful consideration of
the issues discussed in .)In network paths where IPv6/IPv4 protocol translation or IPv6-in-IPv4
encapsulation may be prevalent, it may be prudent for the OAL to always
assume the IPv4 minimum path MTU (i.e., 576 bytes) regardless of the
underlying interface IP protocol version. Always assuming the IPv4
minimum path MTU even for IPv6 underlying interfaces may produce more
fragments and additional header overhead, but will always interoperate
and never run the risk of presenting an IPv4 interface with a packet
that exceeds its MRU.The OMNI interface configures both an MTU and MRU of 9180 bytes ; the size is therefore not a reflection of the
underlying interface MTUs, but rather determines the largest packet the
OMNI interface can forward or reassemble. The OMNI interface uses the
OMNI Adaptation Layer (OAL) to admit packets from the network layer that
are no larger than the OMNI interface MTU while generating ICMPv4
Fragmentation Needed or ICMPv6 Path MTU
Discovery (PMTUD) Packet Too Big (PTB) messages
as necessary. This document refers to both of these ICMPv4/ICMPv6
message types simply as "PTBs", and introduces a distinction between PTB
"hard" and "soft" errors as discussed below.For IPv4 packets with DF=0, the network layer performs IPv4
fragmentation if necessary then admits the packets/fragments into the
OMNI interface; these fragments will be reassembled by the final
destination. For IPv4 packets with DF=1 and IPv6 packets, the network
layer admits the packet if it is no larger than the OMNI interface MTU;
otherwise, it drops the packet and returns a PTB hard error message to
the source.For each admitted IP packet/fragment, the OMNI interface internally
employs the OAL when necessary by inserting a mid-layer IPv6 header
between the inner IP packet/fragment and any outer IP encapsulation
headers per . (The OAL does not decrement the
inner IP Hop Limit/TTL during encapsulation since the insertion occurs
at a layer below IP forwarding.) The OAL then calculates the 32-bit CRC
over the entire mid-layer packet and writes the value in a trailing
4-octet field at the end of the packet. Next, the OAL fragments this
mid-layer IPv6 packet, forwards the fragments (using *NET encapsulation
if necessary), and returns an internally-generated PTB soft error
message (subject to rate limiting) if it deems the packet too large
according to factors such as link performance characteristics,
reassembly congestion, etc. This ensures that the path MTU is adaptive
and reflects the current path used for a given data flow.The OAL operates with respect to both the minimum IPv6 and IPv4 path
MTUs as follows:When an OMNI interface sends a packet toward a final destination
via an ANET peer, it sends without OAL encapsulation if the packet
(including any outer-layer ANET encapsulations) is no larger than
the underlying interface MTU for on-link ANET peers or the minimum
ANET path MTU for peers separated by multiple IP hops. Otherwise,
the OAL inserts an IPv6 header per with
source address set to the node's own Mobile Network Prefix
Unique-Local Address (MNP-ULA) (see: )
and destination set to either the Administrative ULA (ADM-ULA) of
the ANET peer or the MNP-ULA corresponding to the final destination
(see below). The OAL then calculates and appends the trailing 32-bit
CRC, then uses IPv6 fragmentation to break the packet into a minimum
number of non-overlapping fragments where the largest fragment size
(including both the OMNI and any outer-layer ANET encapsulations) is
determined by the underlying interface MTU for on-link ANET peers or
the minimum ANET path MTU for peers separated by multiple IP hops,
and the smallest fragment size is no smaller than 256 bytes. The OAL
then encapsulates the fragments in any ANET headers and sends them
to the ANET peer, which either reassembles before forwarding if the
OAL destination is its own ADM-ULA or forwards the fragments toward
the final destination without first reassembling otherwise.When an OMNI interface sends a packet toward a final destination
via an INET interface, it sends packets (including any outer-layer
INET encapsulations) no larger than the minimum INET path MTU
without OAL encapsulation if the destination is reached via an INET
address within the same OMNI link segment. Otherwise, the OAL
inserts an IPv6 header per with source
address set to the node's ULA, destination set to the ULA of the
next hop OMNI node toward the final destination and (if necessary)
with an OMNI Routing Header (ORH) (see: ) with final segment
addressing information. The OAL then calculates and appends the
trailing 32-bit CRC, then uses IPv6 fragmentation to break the
packet into a minimum number of non-overlapping fragments where the
largest fragment size (including both the OMNI and outer-layer INET
encapsulations) is the minimum INET path MTU, and the smallest
fragment size is no smaller than 256 bytes. The OAL then
encapsulates the fragments in any INET headers and sends them to the
OMNI link neighbor, which reassembles before forwarding toward the
final destination.In light of the above considerations, the OAL uses a Maximum Fragment
Size (MaxFS) of 576 bytes. Each fragment must accommodate 40 bytes for
the OAL header, plus 8 bytes for the OAL fragment header, plus 20 bytes
for a *NET IPv4 header, plus 8 bytes for a *NET UDP header. This leaves
500 bytes for the inner IP header plus data in the initial fragment, 500
bytes for data per intermediate fragment, and up to 500 bytes for the
remaining data plus trailing CRC in the final fragment. The OAL further
uses a Minimum Fragment Size (MinFS) of 256 bytes (i.e., slightly less
than half the minimum IPv4 path MTU). Hence, OAL fragmentation
algorithms must produce fragments that are no larger than MaxFS and no
smaller than MinFS, and any received fragments that are larger than
MaxFS or smaller than MinFS are dropped unconditionally.Note that an exception occurs when two ANET peers share a common link
(which could be a tunnel) with a large MTU such as 1500 bytes or larger.
In that case, fragmentation may use this larger MTU as MaxFS as long as
the receiving ANET peer reassembles (and possibly also refragments)
before forwarding. This is important for accommodating links where
performance is highly dependent on maximum use of the available link
MTU, e.g. for wireless aviation data links. Additionally, in order to
set the correct context for reassembly, the OMNI interface that inserts
the OAL header MUST also be the one that inserts the IPv6 Fragment
Header Identification value. While not strictly required, sending all
fragments of the same fragmented OAL packet consecutively over the same
underlying interface with minimal inter-fragment delay may increase the
likelihood of successful reassembly.Ordinary PTB messages with ICMPv4 header "unused" field or ICMPv6
header Code field value 0 are hard errors that always indicate that a
packet has been dropped due to a real MTU restriction. However, the OAL
can also forward large packets via encapsulation and fragmentation while
at the same time returning PTB soft error messages (subject to rate
limiting) indicating that a forwarded packet was uncomfortably large.
The OMNI interface can therefore continuously forward large packets
without loss while returning PTB soft error messages recommending a
smaller size. Original sources that receive the soft errors in turn
reduce the size of the packets they send, i.e., the same as for hard
errors.The OAL sets the ICMPv4 header "unused" field or ICMPv6 header Code
field to the value 1 in PTB soft error messages. The OAL sets the PTB
destination address to the source address of the original packet, and
sets the source address to the MNP Subnet Router Anycast address of the
MN (i.e., whether the MN was the source or target of the original
packet). The OAL then sets the MTU field to a value no smaller than 576
for ICMPv4 or 1280 for ICMPv6, and returns the PTB soft error to the
original source.When the original source receives the PTB, it reduces its path MTU
estimate the same as for hard errors but does not regard the message as
a loss indication. (If the original source does not recognize the soft
error code, it regards the PTB the same as a hard error but should heed
the retransmission advice given in suggesting
retransmission based on normal packetization layer retransmission
timers.) This document therefore updates and . Furthermore,
implementations of must be aware that PTB hard
or soft errors may arrive at any time even if after a successful MTU
probe (this is the same consideration as for an ordinary path
fluctuation following a successful probe).In summary, the OAL supports continuous transmission and reception of
packets of various sizes in the face of dynamically changing network
conditions. Moreover, since PTB soft errors do not indicate loss,
original sources that receive soft errors can quickly scan for path MTU
increases without waiting for the minimum 10 minutes specified for
loss-oriented PTB hard errors . The OAL therefore provides a lossless and adaptive
service that accommodates MTU diversity especially well-suited for
dynamic multilink environments.Note: An OMNI interface that reassembles OAL fragments may experience
congestion-oriented loss in its reassembly cache and can optionally send
PTB soft errors to the original source and/or ICMP "Time Exceeded"
messages to the source of the OAL fragments. In environments where the
messages may contribute to unacceptable additional congestion, however,
the OMNI interface can simply regard the loss as an ordinary unreported
congestion event for which the original source will eventually
compensate.Note: When the network layer forwards an IPv4 packet/fragment with
DF=0 into the OMNI interface, the interface can optionally perform
(further) IPv4 fragmentation before invoking the OAL so that the
fragments will be reassembled by the final destination. When the network
layer performs IPv6 fragmentation for locally-generated IPv6 packets,
the OMNI interface typically invokes the OAL without first applying
(further) IPv6 fragmentation; the network layer should therefore
fragment to the minimum IPv6 path MTU (or smaller still) to push the
reassembly burden to the final destination and avoid receiving PTB soft
errors from the OMNI interface. Aside from these non-normative
guidelines, the manner in which any IP fragmentation is invoked prior to
OAL encapsulation/fragmentation is an implementation matter.Note: Inclusion of the 32-bit CRC prior to fragmentation assumes that
the receiving OAL will discard any packets with incorrect CRC values
following reassembly. The 32-bit CRC is sufficient to detect reassembly
misassociations for packet sizes up to the OMNI interface MTU 9180 but
may not be sufficient to detect errors for larger sizes .Note: Some underlying interface types (e.g., VPNs) may already
provide their own robust fragmentation and reassembly services even
without OAL encapsulation. In those cases, the OAL can invoke the
inherent underlying interface schemes instead while employing PTB soft
errors in the same fashion as described above. Other underlying
interface properties such as header/message compression can also be
harnessed in a similar fashion.Note: Applications can dynamically tune the size of the packets they
to send to produce the best possible throughput and latency, with the
understanding that these parameters may change over time due to factors
such as congestion, mobility, network path changes, etc. The receipt or
absence of soft errors should be seen as hints of when increasing or
decreasing packet sizes may be beneficial.Note: Any OAL extension headers such as the ORH could cause fragments
to exceed MaxFS bytes. Extension header sizes must therefore be included
in the per-fragment encapsulation overhead.As discussed in Section 3.7 of , there are
four basic threats concerning IPv6 fragmentation; each of which is
addressed by effective mitigations as follows:Overlapping fragment attacks - reassembly of overlapping
fragments is forbidden by ; therefore,
this threat does not apply to the OAL.Resource exhaustion attacks - this threat is mitigated by
providing a sufficiently large OAL reassembly cache and
instituting “fast discard” of incomplete reassemblies
that may be part of a buffer exhaustion attack. The reassembly
cache should be sufficiently large so that a sustained attack does
not cause excessive loss of good reassemblies but not so large
that (timer-based) data structure management becomes
computationally expensive. The cache should also be indexed based
on the arrival underlying interface such that congestion
experienced over a first underlying interface does not cause
discard of incomplete reassemblies for uncongested underlying
interfaces.Attacks based on predictable fragment identification values -
this threat is mitigated by selecting a suitably random ID value
per .Evasion of Network Intrusion Detection Systems (NIDS) - this
threat is mitigated by disallowing “tiny fragments”
per the OAL fragmentation procedures specified above.Additionally, IPv4 fragmentation includes a 16-bit
Identification (IP ID) field with only 65535 unique values such that
at high data rates the field could wrap and apply to new packets while
the fragments of old packets using the same ID are still alive in the
network . However, since the largest OAL
fragment that will be sent via an IPv4 *NET path is 576 bytes any IPv4
fragmentation would occur only on links with an IPv4 MTU smaller than
this size, and recommendations suggest that
such links will have low data rates. Since IPv6 provides a 32-bit
Identification value, IP ID wraparound at high data rates is not a
concern for IPv6 fragmentation.The OMNI interface transmits IPv6 packets according to the native
frame format of each underlying interface. For example, for
Ethernet-compatible interfaces the frame format is specified in , for aeronautical radio interfaces the frame format
is specified in standards such as ICAO Doc 9776 (VDL Mode 2 Technical
Manual), for tunnels over IPv6 the frame format is specified in , etc.OMNI nodes are assigned OMNI interface IPv6 Link-Local Addresses
(LLAs) through pre-service administrative actions. "MNP-LLAs" embed the
MNP assigned to the mobile node, while "ADM-LLAs" include an
administratively-unique ID that is guaranteed to be unique on the link.
LLAs are configured as follows:IPv6 MNP-LLAs encode the most-significant 64 bits of a MNP within
the least-significant 64 bits of the IPv6 link-local prefix
fe80::/64, i.e., in the LLA "interface identifier" portion. The
prefix length for the LLA is determined by adding 64 to the MNP
prefix length. For example, for the MNP 2001:db8:1000:2000::/56 the
corresponding MNP-LLA is fe80::2001:db8:1000:2000/120.IPv4-compatible MNP-LLAs are constructed as fe80::ffff:[IPv4],
i.e., the interface identifier consists of 16 '0' bits, followed by
16 '1' bits, followed by a 32bit IPv4 address/prefix. The prefix
length for the LLA is determined by adding 96 to the MNP prefix
length. For example, the IPv4-Compatible MN OMNI LLA for
192.0.2.0/24 is fe80::ffff:192.0.2.0/120 (also written as
fe80::ffff:c000:0200/120).ADM-LLAs are assigned to ARs and MSEs and MUST be managed for
uniqueness. The lower 32 bits of the LLA includes a unique integer
"MSID" value between 0x00000001 and 0xfeffffff, e.g., as in fe80::1,
fe80::2, fe80::3, etc., fe80::feffffff. The ADM-LLA prefix length is
determined by adding 96 to the MSID prefix length. For example, if
the prefix length for MSID 0x10012001 is 16 then the ADM-LLA prefix
length is set to 112 and the LLA is written as fe80::1001:2001/112.
The "zero" address for each ADM-LLA prefix is the Subnet-Router
anycast address for that prefix ; for
example, the Subnet-Router anycast address for fe80::1001:2001/112
is simply fe80::1001:2000. The MSID range 0xff000000 through
0xffffffff is reserved for future use.Temporary LLAs are constructed per and used by MNs for the
short-term purpose of procuring an actual MNP-LLA upon startup or
(re)connecting to the network. MNs may use Temporary LLAs as the
IPv6 source address of an RS message in order to request a MNP-LLA
from the MS.Since the prefix 0000::/8 is "Reserved by the IETF" , no MNPs can be allocated from that block ensuring
that there is no possibility for overlap between the various LLA
constructs discussed above.Since MNP-LLAs are based on the distribution of administratively
assured unique MNPs, and since ADM-LLAs are guaranteed unique through
administrative assignment, OMNI interfaces set the autoconfiguration
variable DupAddrDetectTransmits to 0 .Temporary LLAs employ optimistic DAD principles since they are probabilistically unique and their use
is short-duration in nature.Note: If future protocol extensions relax the 64-bit boundary in IPv6
addressing, the additional prefix bits of an MNP could be encoded in
bits 16 through 63 of the MNP-LLA. (The most-significant 64 bits would
therefore still be in bits 64-127, and the remaining bits would appear
in bits 16 through 48.) However, the analysis provided in suggests that the 64-bit boundary will remain in the
IPv6 architecture for the forseeable future.Note: Even though this document honors the 64-bit boundary in IPv6
addressing per , it suggests prefix lengths
longer than /64 for routing purposes. This effectively extends IPv6
routing determination into the interface identifier portion of the IPv6
address, but it does not redefine the 64-bit boundary.OMNI domains use IPv6 Unique-Local Addresses (ULAs) as the source and
destination addresses in OAL IPv6 encapsulation headers. ULAs are only
routable within the scope of a an OMNI domain, and are derived from the
IPv6 Unique Local Address prefix fc00::/7 followed by the L bit set to 1
(i.e., as fd00::/8) followed by a 40-bit pseudo-random Global ID to
produce the prefix [ULA]::/48, which is then followed by a 16-bit Subnet
ID then finally followed by a 64 bit Interface ID as specified in
Section 3 of . The statistic uniqueness of the
40-bit pseudo-random Global ID allows different OMNI domains to be
joined together in the future without requiring OMNI link
renumbering.Each OMNI link instance is identified by a value between 0x0000 and
0xfeff in bits 48-63 of [ULA]::/48 (the values 0xff00 through 0xffff are
reserved for future use). For example, OMNI ULAs associated with
instance 0 are configured from the prefix [ULA]:0000::/64, instance 1
from [ULA]:0001::/64, instance 2 from [ULA]:0002::/64, etc. ULAs and
their associated prefix lengths are configured in correspondence with
LLAs through stateless prefix translation where "MNP-ULAs" are assigned
in correspondence to MNP-LLAs and "ADM-ULAs" are assigned in
correspondence to ADM-LLAs. For example, for OMNI link instance
[ULA]:1010::/64:the MNP-ULA corresponding to the MNP-LLA fe80::2001:db8:1:2 with
a 56-bit MNP length is derived by copying the lower 64 bits of the
LLA into the lower 64 bits of the ULA as [ULA]:1010:2001:db8:1:2/120
(where, the ULA prefix length becomes 64 plus the IPv6 MNP
length).the MNP-ULA corresponding to fe80::ffff:192.0.2.0 with a 28-bit
MNP length is derived by simply writing the LLA interface ID into
the lower 64 bits as [ULA]:1010:0:ffff:192.0.2.0/124 (where, the ULA
prefix length is 64 plus 32 plus the IPv4 MNP length).the ADM-ULA corresponding to fe80::1000/112 is simply
[ULA]:1010::1000/112.the ADM-ULA corresponding to fe80::/128 is simply
[ULA]:1010::/128.the Temporary ULA corresponding to a Temporary LLA is simply
[ULA]:1010:[64-bit Temporary Interface ID]/128.etc.Each OMNI interface assigns the Anycast ADM-ULA specific to the OMNI
link instance. For example, the OMNI interface connected to instance 3
assigns the Anycast address [ULA]:0003::/128. Routers that configure
OMNI interfaces advertise the OMNI service prefix (e.g.,
[ULA]:0003::/64) into the local routing system so that applications can
direct traffic according to SBM requirements.The ULA presents an IPv6 address format that is routable within the
OMNI routing system and can be used to convey link-scoped IPv6 ND
messages across multiple hops using IPv6 encapsulation . The OMNI link extends across one or more underling
Internetworks to include all ARs and MSEs. All MNs are also considered
to be connected to the OMNI link, however OAL encapsulation is omitted
whenever possible to conserve bandwidth (see: ).Each OMNI link can be subdivided into "segments" that often
correspond to different administrative domains or physical partitions.
OMNI nodes can use IPv6 Segment Routing when
necessary to support efficient packet forwarding to destinations located
in other OMNI link segments. A full discussion of Segment Routing over
the OMNI link appears in .Note: IPv6 ULAs taken from the prefix fc00::/7 followed by the L bit
set to 0 (i.e., as fc00::/8) are never used for OMNI OAL addressing,
however the range could be used for MSP and MNP addressing under certain
limiting conditions (see: ).OMNI domains use IP Global Unicast Address (GUA) prefixes as Mobility Service Prefixes (MSPs) from which Mobile
Network Prefixes (MNP) are delegated to Mobile Nodes (MNs).For IPv6, GUA prefixes are assigned by IANA
and/or an associated regional assigned numbers authority such that the
OMNI domain can be interconnected to the global IPv6 Internet without
causing inconsistencies in the routing system. An OMNI domain could
instead use ULAs with the 'L' bit set to 0 (i.e., from the prefix
fc00::/8), however this would require
IPv6 Network Address Translation (NAT) if the domain were ever connected
to the global IPv6 Internet.For IPv4, GUA prefixes are assigned by IANA
and/or an associated regional assigned numbers authority such that the
OMNI domain can be interconnected to the global IPv4 Internet without
causing routing inconsistencies. An OMNI domain could instead use
private IPv4 prefixes (e.g., 10.0.0.0/8, etc.) ,
however this would require IPv4 NAT if the domain were ever connected to
the global IPv4 Internet.OMNI interfaces maintain a neighbor cache for tracking per-neighbor
state and use the link-local address format specified in . OMNI interface IPv6 Neighbor Discovery (ND)
messages sent over physical underlying
interfaces without encapsulation observe the native underlying interface
Source/Target Link-Layer Address Option (S/TLLAO) format (e.g., for
Ethernet the S/TLLAO is specified in ). OMNI
interface IPv6 ND messages sent over underlying interfaces via
encapsulation do not include S/TLLAOs which were intended for encoding
physical L2 media address formats and not encapsulation IP addresses.
Furthermore, S/TLLAOs are not intended for encoding additional interface
attributes needed for multilink coordination. Hence, this document does
not define an S/TLLAO format but instead defines a new option type
termed the "OMNI option" designed for these purposes.MNs such as aircraft typically have many wireless data link types
(e.g. satellite-based, cellular, terrestrial, air-to-air directional,
etc.) with diverse performance, cost and availability properties. The
OMNI interface would therefore appear to have multiple L2 connections,
and may include information for multiple underlying interfaces in a
single IPv6 ND message exchange. OMNI interfaces use an IPv6 ND option
called the OMNI option formatted as shown in :In this format:Type is set to TBD. If multiple OMNI option instances appear in
the same IPv6 ND message, the first instance is processed and all
other instances are ignored.Length is set to the number of 8 octet blocks in the option.T is a 1-bit flag set to 1 for Temporary LLAs (otherwise, set to
0) and Preflen is a 7 bit field that determines the length of prefix
associated with an LLA. Values 1 through 127 specify a prefix
length, while the value 0 indicates "unspecified". For IPv6 ND
messages sent from a MN to the MS, T and Preflen apply to the IPv6
source LLA and provide the length that the MN is requesting or
asserting to the MS. For IPv6 ND messages sent from the MS to the
MN, T and Preflen apply to the IPv6 destination LLA and indicate the
length that the MS is granting to the MN. For IPv6 ND messages sent
between MS endpoints, T is set to 0 and Preflen provides the length
associated with the source/target MN that is subject of the ND
message.S/T-ifIndex corresponds to the ifIndex value for source or target
underlying interface used to convey this IPv6 ND message. OMNI
interfaces MUST number each distinct underlying interface with an
ifIndex value between '1' and '255' that represents a MN-specific
8-bit mapping for the actual ifIndex value assigned by network
management (the ifIndex value '0' is
reserved for use by the MS). For RS and NS messages, S/T-ifIndex
corresponds to the source underlying interface the message
originated from. For RA and NA messages, S/T-ifIndex corresponds to
the target underlying interface that the message is destined to.Sub-Options is a Variable-length field, of length such that the
complete OMNI Option is an integer multiple of 8 octets long.
Contains one or more Sub-Options, as described in .The OMNI option includes zero or more Sub-Options. Each consecutive
Sub-Option is concatenated immediately after its predecessor. All
Sub-Options except Pad1 (see below) are in type-length-value (TLV)
encoded in the following format: Sub-Type is a 1-octet field that encodes the Sub-Option type.
Sub-Options defined in this document are:Sub-Types 253 and 254 are reserved for experimentation,
as recommended in .Sub-Length is a 1-octet field that encodes the length of the
Sub-Option Data (i.e., ranging from 0 to 255 octets).Sub-Option Data is a block of data with format determined by
Sub-Type.During processing, unrecognized Sub-Options are ignored and
the next Sub-Option processed until the end of the OMNI option is
reached.The following Sub-Option types and formats are defined in this
document:Sub-Type is set to 0. If multiple instances appear in the
same OMNI option all are processed.No Sub-Length or Sub-Option Data follows (i.e., the
"Sub-Option" consists of a single zero octet).Sub-Type is set to 1. If multiple instances appear in the
same OMNI option all are processed.Sub-Length is set to N (from 0 to 255) being the number of
padding octets that follow.Sub-Option Data consists of N zero-valued octets.The Interface Attributes sub-option provides L2 forwarding
information for the multilink conceptual sending algorithm discussed
in . The L2 information is used for
selecting among potentially multiple candidate underlying interfaces
that can be used to forward packets to the neighbor based on factors
such as DSCP preferences and link quality. Interface Attributes
further include link-layer address information to be used for either
OAL encapsulation or direct UDP/IP encapsulation (when OAL
encapsulation can be avoided). The Interface Attributes format and
contents are given in below:Sub-Type is set to 2. If multiple instances with different
ifIndex values appear in the same OMNI option all are processed;
if multiple instances with the same ifIndex value appear, the
first is processed and all others are ignored.Sub-Length is set to N (from 4 to 255) that encodes the
number of Sub-Option Data octets that follow.Sub-Option Data contains an "Interface Attribute" option
encoded as follows (note that the first four octets must be
present):ifIndex is set to an 8-bit integer value corresponding to
a specific underlying interface the same as specified above
for the OMNI option header S/T-ifIndex. An OMNI option may
include multiple Interface Attributes Sub-Options, with each
distinct ifIndex value pertaining to a different underlying
interface. The OMNI option will often include an Interface
Attributes Sub-Option with the same ifIndex value that
appears in the S/T-ifIndex. In that case, the actual
encapsulation address of the received IPv6 ND message should
be compared with the L2ADDR encoded in the Sub-Option (see
below); if the addresses are different (or, if L2ADDR
absent) the presence of a Network Address Translator (NAT)
is indicated.ifType is set to an 8-bit integer value corresponding to
the underlying interface identified by ifIndex. The value
represents an OMNI interface-specific 8-bit mapping for the
actual IANA ifType value registered in the 'IANAifType-MIB'
registry [http://www.iana.org].Provider ID is set to an OMNI interface-specific 8-bit ID
value for the network service provider associated with this
ifIndex.Link encodes a 4-bit link metric. The value '0' means the
link is DOWN, and the remaining values mean the link is UP
with metric ranging from '1' ("lowest") to '15'
("highest").R is reserved for future use.API - a 3-bit "Address/Preferences/Indexed" code that
determines the contents of the remainder of the sub-option
as follows:When the most significant bit (i.e., "Address") is
set to 1, the SRT, FMT, LHS and L2ADDR fields are
included immediately following the API code; else, they
are omitted.When the next most significant bit (i.e.,
"Preferences") is set to 1, a preferences block is
included next; else, it is omitted. (Note that if
"Address" is set the preferences block immediately
follows L2ADDR; else, it immediately follows the API
code.)When a preferences block is present and the least
significant bit (i.e., "Indexed") is set to 0, the block
is encoded in "Simplex" form as shown in ; else it is encoded in
"Indexed" form as discussed below.When API indicates that an "Address" is included, the
following fields appear in consecutive order (else, they are
omitted):SRT - a 5-bit Segment Routing Topology prefix length
value that (when added to 96) determines the prefix
length to apply to the ULA formed from concatenating
[ULA*]::/96 with the 32 bit LHS MSID value that follows.
For example, the value 16 corresponds to the prefix
length 112.FMT - a 3-bit "Framework/Mode/Type" code
corresponding to the included Link Layer Address as
follows:When the most significant bit (i.e., "Framework")
is set to 0, L2ADDR is the INET encapsulation
address of a Proxy/Server; otherwise, it is the
address for the Source/Target itselfWhen the next most significant bit (i.e., "Mode")
is set to 0, the Source/Target L2ADDR is on the open
INET; otherwise, it is (likely) located behind a
Network Address Translator (NAT).When the least significant bit (i.e., "Type") is
set to 0, L2ADDR includes a UDP Port Number followed
by an IPv4 address; else, a UDP Port Number followed
by an IPv6 address.LHS - the 32 bit MSID of the Last Hop Server/Proxy on
the path to the target. When SRT and LHS are both set to
0, the LHS is considered unspecified in this IPv6 ND
message. When SRT is set to 0 and LHS is non-zero, the
prefix length is set to 128. SRT and LHS together
provide guidance to the OMNI interface forwarding
algorithm. Specifically, if SRT/LHS is located in the
local OMNI link segment then the OMNI interface can
encapsulate according to FMT/L2ADDR; else, it must
forward according to the OMNI link spanning tree. See
for further
discussion.Link Layer Address (L2ADDR) - Formatted according to
FMT, and identifies the link-layer address (i.e., the
encapsulation address) of the source/target. The UDP
Port Number appears in the first two octets and the IP
address appears in the next 4 octets for IPv4 or 16
octets for IPv6. The Port Number and IP address are
recorded in ones-compliment "obfuscated" form per . The OMNI interface forwarding
algorithm uses FMT/L2ADDR to determine the encapsulation
address for forwarding when SRT/LHS is located in the
local OMNI link segment.When API indicates that "Preferences" are included, a
preferences block appears as the remainder of the Sub-Option
as a series of Bitmaps and P[*] values. In "Simplex" form,
the index for each singleton Bitmap octet is inferred from
its sequential position (i.e., 0, 1, 2, ...) as shown in
. In "Indexed" form, each
Bitmap is preceded by an Index octet that encodes a value
"i" = (0 - 255) as the index for its companion Bitmap as
follows:The preferences consist of a first (simplex/indexed)
Bitmap (i.e., "Bitmap(i)") followed by 0-8 single-octet
blocks of 2-bit P[*] values, followed by a second Bitmap
(i), followed by 0-8 blocks of P[*] values, etc. Reading
from bit 0 to bit 7, the bits of each Bitmap(i) that are set
to '1'' indicate the P[*] blocks from the range P[(i*32)]
through P[(i*32) + 31] that follow; if any Bitmap(i) bits
are '0', then the corresponding P[*] block is instead
omitted. For example, if Bitmap(0) contains 0xff then the
block with P[00]-P[03], followed by the block with
P[04]-P[07], etc., and ending with the block with
P[28]-P[31] are included (as shown in ). The next Bitmap(i) is then
consulted with its bits indicating which P[*] blocks follow,
etc. out to the end of the Sub-Option.Each 2-bit P[*] field is set to the value '0'
("disabled"), '1' ("low"), '2' ("medium") or '3' ("high") to
indicate a QoS preference for underlying interface selection
purposes. Not all P[*] values need to be included in the
OMNI option of each IPv6 ND message received. Any P[*]
values represented in an earlier OMNI option but omitted in
the current OMNI option remain unchanged. Any P[*] values
not yet represented in any OMNI option default to
"medium".The first 16 P[*] blocks correspond to the 64
Differentiated Service Code Point (DSCP) values P[00] -
P[63] . Any additional P[*] blocks
that follow correspond to "pseudo-DSCP" traffic classifier
values P[64], P[65], P[66], etc. See Appendix A for further
discussion and examples.Sub-Type is set to 3. If multiple instances appear in the
same OMNI option all are processed, i.e., even if the same
ifIndex value appears multiple times.Sub-Length is set to N (the number of Sub-Option Data octets
that follow).Sub-Option Data contains a 1-octet ifIndex encoded exactly as
specified in , followed by an N-1 octet
traffic selector formatted per
beginning with the "TS Format" field. The largest traffic
selector for a given ifIndex is therefore 254 octets.Sub-Type is set to 4. If multiple instances appear in the
same OMNI option all are processed. Only the first MAX_MSID
values processed (whether in a single instance or multiple) are
retained and all other MSIDs are ignored.Sub-Length is set to 4n.A list of n 4-octet MSIDs is included in the following 4n
octets. The Anycast MSID value '0' in an RS message MS-Register
sub-option requests the recipient to return the MSID of a nearby
MSE in a corresponding RA response.Sub-Type is set to 5. If multiple instances appear in the
same IPv6 OMNI option all are processed. Only the first MAX_MSID
values processed (whether in a single instance or multiple) are
retained and all other MSIDs are ignored.Sub-Length is set to 4n.A list of n 4 octet MSIDs is included in the following 4n
octets. The Anycast MSID value '0' is ignored in MS-Release
sub-options, i.e., only non-zero values are processed.Sub-Type is set to 6. If multiple instances appear in the
same OMNI option the first is processed and all others are
ignored.Sub-Length is set to N (i.e., the length of the encoded
Network Access Identifier (NAI)).An NAI up to 255 octets in length is coded per .Sub-Type is set to 7. If multiple instances appear in the
same OMNI option the first is processed and all others are
ignored.Sub-Length is set to N (i.e., the length of the encoded Geo
Coordinates).A set of Geo Coordinates up to 255 octets in length (format
TBD). Includes Latitude/Longitude at a minimum; may also include
additional attributes such as altitude, heading, speed,
etc.).Sub-Type is set to 8. If multiple instances appear in the
same OMNI option the first is processed and all others are
ignored.Sub-Length is set to N (i.e., the length of the option
beginning with the DUID-Type and continuing to the end of the
type-specific body).DUID-Type is a two-octet field coded in network byte order
that determines the format and contents of the type-specific
body according to Section 11 of .
DUID-Type 4 in particular corresponds to the Universally Unique
Identifier (UUID) which will occur in
common operational practice.A type-specific DUID body up to 253 octets in length follows,
formatted according to DUID-type. For example, for type 4 the
body consists of a 128-bit UUID selected according to .Sub-Type is set to 9. If multiple instances appear in the
same OMNI option the first is processed and all others are
ignored.Sub-Length is set to N (i.e., the length of the DHCPv6
message beginning with 'msg-type' and continuing to the end of
the DHCPv6 options). The length of the entire DHCPv6 message is
therefore restricted to 255 octets.'msg-type' and 'transaction-id' are coded according to
Section 8 of .A set of DHCPv6 options coded according to Section 21 of
follows.The multicast address mapping of the native underlying interface
applies. The mobile router on board the MN also serves as an IGMP/MLD
Proxy for its EUNs and/or hosted applications per while using the L2 address of the AR as the L2
address for all multicast packets.The MN uses Multicast Listener Discovery (MLDv2) to coordinate with the AR, and *NET L2 elements use
MLD snooping .The MN's IPv6 layer selects the outbound OMNI interface according to
SBM considerations when forwarding data packets from local or EUN
applications to external correspondents. Each OMNI interface maintains a
neighbor cache the same as for any IPv6 interface, but with additional
state for multilink coordination. Each OMNI interface maintains default
routes via ARs discovered as discussed in , and
may configure more-specific routes discovered through means outside the
scope of this specification.After a packet enters the OMNI interface, one or more outbound
underlying interfaces are selected based on PBM traffic attributes, and
one or more neighbor underlying interfaces are selected based on the
receipt of Interface Attributes sub-options in IPv6 ND messages (see:
). Underlying interface selection for the
nodes own local interfaces are based on attributes such as DSCP,
application port number, cost, performance, message size, etc. OMNI
interface multilink selections could also be configured to perform
replication across multiple underlying interfaces for increased
reliability at the expense of packet duplication. The set of all
Interface Attributes received in IPv6 ND messages determine the
multilink forwarding profile for selecting the neighbor's underlying
interfaces.When the OMNI interface sends a packet over a selected outbound
underlying interface, the OAL includes or omits a mid-layer
encapsulation header as necessary as discussed in and as determined by the L2 address information
received in Interface Attributes. The OAL also performs encapsulation
when the nearest AR is located multiple hops away as discussed in .OMNI interface multilink service designers MUST observe the BCP
guidance in Section 15 in terms of implications
for reordering when packets from the same flow may be spread across
multiple underlying interfaces having diverse properties.MNs may connect to multiple independent OMNI links concurrently in
support of SBM. Each OMNI interface is distinguished by its Anycast
ULA (e.g., [ULA]:0002::, [ULA]:1000::, [ULA]:7345::, etc.). The MN
configures a separate OMNI interface for each link so that multiple
interfaces (e.g., omni0, omni1, omni2, etc.) are exposed to the IPv6
layer. A different Anycast ULA is assigned to each interface, and the
MN injects the service prefixes for the OMNI link instances into the
EUN routing system.Applications in EUNs can use Segment Routing to select the desired
OMNI interface based on SBM considerations. The Anycast ULA is written
into the IPv6 destination address, and the actual destination (along
with any additional intermediate hops) is written into the Segment
Routing Header. Standard IP routing directs the packets to the MN's
mobile router entity, and the Anycast ULA identifies the OMNI
interface to be used for transmission to the next hop. When the MN
receives the message, it replaces the IPv6 destination address with
the next hop found in the routing header and transmits the message
over the OMNI interface identified by the Anycast ULA.Multiple distinct OMNI links can therefore be used to support fault
tolerance, load balancing, reliability, etc. The architectural model
is similar to Layer 2 Virtual Local Area Networks (VLANs).After an AR has registered an MNP for a MN (see: ), the AR will forward packets destined to an address
within the MNP to the MN. The MN will under normal circumstances then
forward the packet to the correct destination within its internal
networks.If at some later time the MN loses state (e.g., after a reboot), it
may begin returning packets destined to an MNP address to the AR as
its default router. The AR therefore must drop any packets originating
from the MN and destined to an address within the MN's registered MNP.
To do so, the AR institutes the following check:if the IP destination address belongs to a neighbor on the same
OMNI interface, and if the link-layer source address is the same
as one of the neighbor's link-layer addresses, drop the
packet.MNs interface with the MS by sending RS messages with OMNI options
under the assumption that one or more AR on the *NET will process the
message and respond. The MN then configures default routes for the OMNI
interface via the discovered ARs as the next hop. The manner in which
the *NET ensures AR coordination is link-specific and outside the scope
of this document (however, considerations for *NETs that do not provide
ARs that recognize the OMNI option are discussed in ).For each underlying interface, the MN sends an RS message with an
OMNI option to coordinate with MSEs identified by MSID values. Example
MSID discovery methods are given in and include
data link login parameters, name service lookups, static configuration,
a static "hosts" file, etc. The MN can also send an RS with an
MS-Register suboption that includes the Anycast MSID value '0', i.e.,
instead of or in addition to any non-zero MSIDs. When the AR receives an
RS with a MSID '0', it selects a nearby MSE (which may be itself) and
returns an RA with the selected MSID in an MS-Register suboption. The AR
selects only a single wildcard MSE (i.e., even if the RS MS-Register
suboption included multiple '0' MSIDs) while also soliciting the MSEs
corresponding to any non-zero MSIDs.MNs configure OMNI interfaces that observe the properties discussed
in the previous section. The OMNI interface and its underlying
interfaces are said to be in either the "UP" or "DOWN" state according
to administrative actions in conjunction with the interface connectivity
status. An OMNI interface transitions to UP or DOWN through
administrative action and/or through state transitions of the underlying
interfaces. When a first underlying interface transitions to UP, the
OMNI interface also transitions to UP. When all underlying interfaces
transition to DOWN, the OMNI interface also transitions to DOWN.When an OMNI interface transitions to UP, the MN sends RS messages to
register its MNP and an initial set of underlying interfaces that are
also UP. The MN sends additional RS messages to refresh lifetimes and to
register/deregister underlying interfaces as they transition to UP or
DOWN. The MN sends initial RS messages over an UP underlying interface
with its MNP-LLA as the source and with destination set to All-Routers
multicast (ff02::2) . The RS messages include an
OMNI option per with a Preflen assertion,
Interface Attributes appropriate for underlying interfaces,
MS-Register/Release sub-options containing MSID values, and with any
other necessary OMNI sub-options (e.g., a DUID suboption as an identity
for the MN). The S/T-ifIndex field is set to the index of the underlying
interface over which the RS message is sent.ARs process IPv6 ND messages with OMNI options and act as an MSE
themselves and/or as a proxy for other MSEs. ARs receive RS messages and
create a neighbor cache entry for the MN, then coordinate with any MSEs
named in the Register/Release lists in a manner outside the scope of
this document. When an MSE processes the OMNI information, it first
validates the prefix registration information then injects/withdraws the
MNP in the routing/mapping system and caches/discards the new Preflen,
MNP and Interface Attributes. The MSE then informs the AR of
registration success/failure, and the AR returns an RA message to the MN
with an OMNI option per .The AR returns the RA message via the same underlying interface of
the MN over which the RS was received, and with destination address set
to the MNP-LLA (i.e., unicast), with source address set to its own LLA,
and with an OMNI option with S/T-ifIndex set to the value included in
the RS. The OMNI option also includes a Preflen confirmation, Interface
Attributes, MS-Register/Release and any other necessary OMNI sub-options
(e.g., a DUID suboption as an identity for the AR). The RA also includes
any information for the link, including RA Cur Hop Limit, M and O flags,
Router Lifetime, Reachable Time and Retrans Timer values, and includes
any necessary options such as:PIOs with (A; L=0) that include MSPs for the link .RIOs with more-specific routes.an MTU option that specifies the maximum acceptable packet size
for this underlying interface.The AR MAY also send periodic and/or event-driven unsolicited RA
messages per . In that case, the S/T-ifIndex
field in the OMNI header of the unsolicited RA message identifies the
target underlying interface of the destination MN.The AR can combine the information from multiple MSEs into one or
more "aggregate" RAs sent to the MN in order conserve *NET bandwidth.
Each aggregate RA includes an OMNI option with MS-Register/Release
sub-options with the MSEs represented by the aggregate. If an aggregate
is sent, the RA message contents must consistently represent the
combined information advertised by all represented MSEs. Note that since
the AR uses its own ADM-LLA as the RA source address, the MN determines
the addresses of the represented MSEs by examining the
MS-Register/Release OMNI sub-options.When the MN receives the RA message, it creates an OMNI interface
neighbor cache entry for each MSID that has confirmed MNP registration
via the L2 address of this AR. If the MN connects to multiple *NETs, it
records the additional L2 AR addresses in each MSID neighbor cache entry
(i.e., as multilink neighbors). The MN then configures a default route
via the MSE that returned the RA message, and assigns the Subnet Router
Anycast address corresponding to the MNP (e.g., 2001:db8:1:2::) to the
OMNI interface. The MN then manages its underlying interfaces according
to their states as follows:When an underlying interface transitions to UP, the MN sends an
RS over the underlying interface with an OMNI option. The OMNI
option contains at least one Interface Attribute sub-option with
values specific to this underlying interface, and may contain
additional Interface Attributes specific to other underlying
interfaces. The option also includes any MS-Register/Release
sub-options.When an underlying interface transitions to DOWN, the MN sends an
RS or unsolicited NA message over any UP underlying interface with
an OMNI option containing an Interface Attribute sub-option for the
DOWN underlying interface with Link set to '0'. The MN sends an RS
when an acknowledgement is required, or an unsolicited NA when
reliability is not thought to be a concern (e.g., if redundant
transmissions are sent on multiple underlying interfaces).When the Router Lifetime for a specific AR nears expiration, the
MN sends an RS over the underlying interface to receive a fresh RA.
If no RA is received, the MN can send RS messages to an alternate
MSID in case the current MSID has failed. If no RS messages are
received even after trying to contact alternate MSIDs, the MN marks
the underlying interface as DOWN.When a MN wishes to release from one or more current MSIDs, it
sends an RS or unsolicited NA message over any UP underlying
interfaces with an OMNI option with a Release MSID. Each MSID then
withdraws the MNP from the routing/mapping system and informs the AR
that the release was successful.When all of a MNs underlying interfaces have transitioned to DOWN
(or if the prefix registration lifetime expires), any associated
MSEs withdraw the MNP the same as if they had received a message
with a release indication.The MN is responsible for retrying each RS exchange up to
MAX_RTR_SOLICITATIONS times separated by RTR_SOLICITATION_INTERVAL
seconds until an RA is received. If no RA is received over an UP
underlying interface (i.e., even after attempting to contact alternate
MSEs), the MN declares this underlying interface as DOWN.The IPv6 layer sees the OMNI interface as an ordinary IPv6 interface.
Therefore, when the IPv6 layer sends an RS message the OMNI interface
returns an internally-generated RA message as though the message
originated from an IPv6 router. The internally-generated RA message
contains configuration information that is consistent with the
information received from the RAs generated by the MS. Whether the OMNI
interface IPv6 ND messaging process is initiated from the receipt of an
RS message from the IPv6 layer is an implementation matter. Some
implementations may elect to defer the IPv6 ND messaging process until
an RS is received from the IPv6 layer, while others may elect to
initiate the process proactively. Still other deployments may elect to
administratively disable the ordinary RS/RA messaging used by the IPv6
layer over the OMNI interface, since they are not required to drive the
internal RS/RA processing. (Note that this same logic applies to IPv4
implementations that employ ICMP-based Router Discovery per .)Note: The Router Lifetime value in RA messages indicates the time
before which the MN must send another RS message over this underlying
interface (e.g., 600 seconds), however that timescale may be
significantly longer than the lifetime the MS has committed to retain
the prefix registration (e.g., REACHABLETIME seconds). ARs are therefore
responsible for keeping MS state alive on a shorter timescale than the
MN is required to do on its own behalf.Note: On multicast-capable underlying interfaces, MNs should send
periodic unsolicited multicast NA messages and ARs should send periodic
unsolicited multicast RA messages as "beacons" that can be heard by
other nodes on the link. If a node fails to receive a beacon after a
timeout value specific to the link, it can initiate a unicast exchange
to test reachability.Note: if an AR acting as a proxy forwards a MN's RS message to
another node acting as an MSE using UDP/IP encapsulation, it must use a
distinct UDP source port number for each MN. This allows the MSE to
distinguish different MNs behind the same AR at the link-layer, whereas
the link-layer addresses would otherwise be indistinguishable.On some *NETs, a MN may be located multiple IP hops away from the
nearest AR. Forwarding through IP multihop *NETs is conducted through
the application of a routing protocol (e.g., a Mobile Ad-hoc Network
(MANET) routing protocol over omni-directional wireless interfaces, an
inter-domain routing protocol in an enterprise network, etc.). These
*NETs could be either IPv6-enabled or IPv4-only, while IPv4-only *NETs
could be either multicast-capable or unicast-only (note that for
IPv4-only *NETs the following procedures apply for both single-hop and
multihop cases).A MN located potentially multiple *NET hops away from the nearest
AR prepares an RS message with source address set to either its
MNP-LLA or a Temporary LLA, and with destination set to link-scoped
All-Routers multicast the same as discussed above. For IPv6-enabled
*NETs, the MN then encapsulates the message in an IPv6 header with
source address set to the ULA corresponding to the LLA source address
and with destination set to either a unicast or anycast ADM-ULA. For
IPv4-only *NETs, the MN instead encapsulates the RS message in an IPv4
header with source address set to the node's own IPv4 address and with
destination address set to either the unicast IPv4 address of an AR
or an IPv4 anycast address reserved for OMNI.
The MN then sends the encapsulated RS message via the *NET interface,
where it will be forwarded by zero or more intermediate *NET hops.When an intermediate *NET hop that participates in the routing
protocol receives the encapsulated RS, it forwards the message
according to its routing tables (note that an intermediate node could
be a fixed infrastructure element or another MN). This process repeats
iteratively until the RS message is received by a penultimate *NET hop
within single-hop communications range of an AR, which forwards the
message to the AR.When the AR receives the message, it decapsulates the RS and
coordinates with the MS the same as for an ordinary link-local RS,
since the inner Hop Limit will not have been decremented by the
multihop forwarding process. The AR then prepares an RA message with
source address set to its own ADM-LLA and destination address set to
the LLA of the original MN, then encapsulates the message in an
IPv4/IPv6 header with source address set to its own IPv4/ULA address
and with destination set to the encapsulation source of the RS.The AR then forwards the message to an *NET node within
communications range, which forwards the message according to its
routing tables to an intermediate node. The multihop forwarding
process within the *NET continues repetitively until the message is
delivered to the original MN, which decapsulates the message and
performs autoconfiguration the same as if it had received the RA
directly from the AR as an on-link neighbor.Note: An alternate approach to multihop forwarding via IPv6
encapsulation would be to statelessly translate the IPv6 LLAs into
ULAs and forward the messages without encapsulation. This would
violate the requirement that certain IPv6 ND
messages must use link-local addresses and must not be accepted if
received with Hop Limit less than 255. This document therefore
advocates encapsulation since the overhead is nominal considering the
infrequent nature and small size of IPv6 ND messages. Future documents
may consider encapsulation avoidance through translation while
updating .Note: An alternate approach to multihop forwarding via IPv4
encapsulation would be to employ IPv6/IPv4 protocol translation.
However, for IPv6 ND messages the LLAs would be truncated due to
translation and the OMNI Router and Prefix Discovery services would
not be able to function. The use of IPv4 encapsulation is therefore
indicated.Note: An IPv4 anycast address for OMNI in IPv4 networks could be
part of a new IPv4 /24 prefix allocation, but this may be difficult to
obtain given IPv4 address exhaustion. An alternative would be to
re-purpose the prefix 192.88.99.0 which has been set aside from its
former use by .When a MN sends an RS message with an OMNI option via an underlying
interface to an AR, the MN must convey its knowledge of its
currently-associated MSEs. Initially, the MN will have no associated
MSEs and should therefore include an MS-Register sub-option with the
single MSID value 0 which requests the AR to select and assign an MSE.
The AR will then return an RA message with source address set to the
ADM-LLA of the selected MSE.As the MN activates additional underlying interfaces, it can
optionally include an MS-Register sub-option with MSID value 0, or
with non-zero MSIDs for MSEs discovered from previous RS/RA exchanges.
The MN will thus eventually begin to learn and manage its currently
active set of MSEs, and can register with new MSEs or release from
former MSEs with each successive RS/RA exchange. As the MN's MSE
constituency grows, it alone is responsible for including or omitting
MSIDs in the MS-Register/Release lists it sends in RS messages. The
inclusion or omission of MSIDs determines the MN's interface to the MS
and defines the manner in which MSEs will respond. The only limiting
factor is that the MN should include no more than MAX_MSID values in
each list per each IPv6 ND message, and should avoid duplication of
entries in each list unless it wants to increase likelihood of control
message delivery.When an AR receives an RS message sent by a MN with an OMNI option,
the option will contain zero or more MS-Register and MS-Release
sub-options containing MSIDs. After processing the OMNI option, the AR
will have a list of zero or more MS-Register MSIDs and a list of zero
or more of MS-Release MSIDs. The AR then processes the lists as
follows:For each list, retain the first MAX_MSID values in the list and
discard any additional MSIDs (i.e., even if there are duplicates
within a list).Next, for each MSID in the MS-Register list, remove all
matching MSIDs from the MS-Release list.Next, proceed according to whether the AR's own MSID or the
value 0 appears in the MS-Register list as follows:If yes, send an RA message directly back to the MN and send
a proxy copy of the RS message to each additional MSID in the
MS-Register list with the MS-Register/Release lists omitted.
Then, send a uNA message to each MSID in the MS-Release list
with the MS-Register/Release lists omitted and with an OMNI
header with S/T-ifIndex set to 0.If no, send a proxy copy of the RS message to each
additional MSID in the MS-Register list with the MS-Register
list omitted. For the first MSID, include the original
MS-Release list; for all other MSIDs, omit the MS-Release
list.Each proxy copy of the RS message will include an OMNI option
and encapsulation header with the ADM-ULA of the AR as the source and
the ADM-ULA of the Register MSE as the destination. When the Register
MSE receives the proxy RS message, if the message includes an
MS-Release list the MSE sends a uNA message to each additional MSID in
the Release list. The Register MSE then sends an RA message back to
the (Proxy) AR wrapped in an OMNI encapsulation header with source and
destination addresses reversed, and with RA destination set to the
MNP-LLA of the MN. When the AR receives this RA message, it sends a
proxy copy of the RA to the MN.Each uNA message (whether send by the first-hop AR or by a Register
MSE) will include an OMNI option and an encapsulation header with the
ADM-ULA of the Register MSE as the source and the ADM-ULA of the
Release ME as the destination. The uNA informs the Release MSE that
its previous relationship with the MN has been released and that the
source of the uNA message is now registered. The Release MSE must then
note that the subject MN of the uNA message is now "departed", and
forward any subsequent packets destined to the MN to the Register
MSE.Note that it is not an error for the MS-Register/Release lists to
include duplicate entries. If duplicates occur within a list, the AR
will generate multiple proxy RS and/or uNA messages - one for each
copy of the duplicate entries.When a MN is not pre-provisioned with an MNP-LLA (or, when multiple
MNPs are needed), it will require the AR to select MNPs on its behalf
and set up the correct routing state within the MS. The DHCPv6 service
supports this requirement.When an MN needs to have the AR select MNPs, it sends an RS message
with a Temporary LLA as the source and with DHCPv6 Message suboption
containing a Client Identifier, one or more IA_PD options and a Rapid
Commit option. The MN also sets the 'msg-type' field to "Solicit", and
includes a 3-octet 'transaction-id'.When the AR receives the RS message, it extracts the DHCPv6 message
from the OMNI option. The AR then acts as a "Proxy DHCPv6 Client" in a
message exchange with the locally-resident DHCPv6 server, which
delegates MNPs and returns a DHCPv6 Reply message with PD parameters.
(If the AR wishes to defer creation of MN state until the DHCPv6 Reply
is received, it can instead act as a Lightweight DHCPv6 Relay Agent
per by encapsulating the DHCPv6 message in a
Relay-forward/reply exchange with Relay Message and Interface ID
options.)When the AR receives the DHCPv6 Reply, it adds routes to the
routing system and creates MNP-LLAs based on the delegated MNPs. The
AR then sends an RA back to the MN with the DHCPv6 Reply message
included in an OMNI DHCPv6 message sub-option. If the RS message
source address was a Temporary address, the AR includes one of the
(newly-created) MNP-LLAs as the RA destination address. The MN then
creates a default route, assigns Subnet Router Anycast addresses and
uses the RA destination address as its primary MNP-LLA. The MN will
then use this primary MNP-LLA as the source address of any IPv6 ND
messages it sends as long as it retains ownership of the MNP.Note: The single-octet OMNI sub-option length field restricts the
DHCPv6 Message sub-option to a maximum of 255 octets for both the RS
and RA messages. This provides sufficient room for the DHCPv6 message
header, a Client/Server Identifier option, a Rapid Commit option, at
least 3 Identity Association for Prefix Delegation (IA_PD) options and
any other supporting DHCPv6 options. A MN requiring more DHCPv6-based
configuration information than this can either perform multiple
independent RS/RA exchanges (with each exchange providing a subset of
the total configuration information) or simply perform an actual
DHCPv6 message exchange in addition to any RS/RA exchanges.Note: After a MN performs a DHCPv6-based prefix registration
exchange with a first AR, it would need to repeat the exchange with
each additional MSE it registers with. In that case, the MN supplies
the MNP delegations received from the first AR in the IA_PD fields of
a DHCPv6 message when it engages the additional MSEs.If the *NET link model is multiple access, the AR is responsible for
assuring that address duplication cannot corrupt the neighbor caches of
other nodes on the link. When the MN sends an RS message on a multiple
access *NET link, the AR verifies that the MN is authorized to use the
address and returns an RA with a non-zero Router Lifetime only if the MN
is authorized.After verifying MN authorization and returning an RA, the AR MAY
return IPv6 ND Redirect messages to direct MNs located on the same *NET
link to exchange packets directly without transiting the AR. In that
case, the MNs can exchange packets according to their unicast L2
addresses discovered from the Redirect message instead of using the
dogleg path through the AR. In some *NET links, however, such direct
communications may be undesirable and continued use of the dogleg path
through the AR may provide better performance. In that case, the AR can
refrain from sending Redirects, and/or MNs can ignore them.*NETs SHOULD deploy ARs in Virtual Router Redundancy Protocol (VRRP)
configurations so that service continuity is
maintained even if one or more ARs fail. Using VRRP, the MN is unaware
which of the (redundant) ARs is currently providing service, and any
service discontinuity will be limited to the failover time supported by
VRRP. Widely deployed public domain implementations of VRRP are
available.MSEs SHOULD use high availability clustering services so that
multiple redundant systems can provide coordinated response to failures.
As with VRRP, widely deployed public domain implementations of high
availability clustering services are available. Note that
special-purpose and expensive dedicated hardware is not necessary, and
public domain implementations can be used even between lightweight
virtual machines in cloud deployments.In environments where fast recovery from MSE failure is required, ARs
SHOULD use proactive Neighbor Unreachability Detection (NUD) in a manner
that parallels Bidirectional Forwarding Detection (BFD) to track MSE reachability. ARs can then quickly
detect and react to failures so that cached information is
re-established through alternate paths. Proactive NUD control messaging
is carried only over well-connected ground domain networks (i.e., and
not low-end *NET links such as aeronautical radios) and can therefore be
tuned for rapid response.ARs perform proactive NUD for MSEs for which there are currently
active MNs on the *NET. If an MSE fails, ARs can quickly inform MNs of
the outage by sending multicast RA messages on the *NET interface. The
AR sends RA messages to MNs via the *NET interface with an OMNI option
with a Release ID for the failed MSE, and with destination address set
to All-Nodes multicast (ff02::1) .The AR SHOULD send MAX_FINAL_RTR_ADVERTISEMENTS RA messages separated
by small delays . Any MNs on the *NET interface
that have been using the (now defunct) MSE will receive the RA messages
and associate with a new MSE.When a MN connects to an *NET link for the first time, it sends an RS
message with an OMNI option. If the first hop AR recognizes the option,
it returns an RA with its ADM-LLA as the source, the MNP-LLA as the
destination and with an OMNI option included. The MN then engages the AR
according to the OMNI link model specified above. If the first hop AR is
a legacy IPv6 router, however, it instead returns an RA message with no
OMNI option and with a non-OMNI unicast source LLA as specified in . In that case, the MN engages the *NET according to
the legacy IPv6 link model and without the OMNI extensions specified in
this document.If the *NET link model is multiple access, there must be assurance
that address duplication cannot corrupt the neighbor caches of other
nodes on the link. When the MN sends an RS message on a multiple access
*NET link with an LLA source address and an OMNI option, ARs that
recognize the option ensure that the MN is authorized to use the address
and return an RA with a non-zero Router Lifetime only if the MN is
authorized. ARs that do not recognize the option instead return an RA
that makes no statement about the MN's authorization to use the source
address. In that case, the MN should perform Duplicate Address Detection
to ensure that it does not interfere with other nodes on the link.An alternative approach for multiple access *NET links to ensure
isolation for MN / AR communications is through L2 address mappings as
discussed in . This arrangement imparts a
(virtual) point-to-point link model over the (physical) multiple access
link.OMNI interfaces configured over IPv6-enabled underlying interfaces on
the open Internet without an OMNI-aware first-hop AR receive RA messages
that do not include an OMNI option, while OMNI interfaces configured
over IPv4-only underlying interfaces do not receive any (IPv6) RA
messages at all. OMNI interfaces that receive RA messages without an
OMNI option configure addresses, on-link prefixes, etc. on the
underlying interface that received the RA according to standard IPv6 ND
and address resolution conventions . OMNI interfaces configured over IPv4-only underlying
interfaces configure IPv4 address information on the underlying
interfaces using mechanisms such as DHCPv4 .OMNI interfaces configured over underlying interfaces that connect to
the open Internet can apply security services such as VPNs to connect to
an MSE, or can establish a direct link to an MSE through some other
means (see ). In environments where an explicit
VPN or direct link may be impractical, OMNI interfaces can instead use
UDP/IP encapsulation and HMAC-based message authentication per .After establishing a VPN or preparing for UDP/IP encapsulation, OMNI
interfaces send control plane messages to interface with the MS,
including RS/RA messages used according to and
Neighbor Solicitation (NS) and Neighbor Advertisement (NA) messages used
for address resolution / route optimization (see: ). The control plane messages must
be authenticated while data plane messages are delivered the same as for
ordinary best-effort Internet traffic with basic source address-based
data origin verification. Data plane communications via OMNI interfaces
that connect over the open Internet without an explicit VPN should
therefore employ transport- or higher-layer security to ensure integrity
and/or confidentiality.OMNI interfaces in the open Internet are often located behind Network
Address Translators (NATs). The OMNI interface accommodates NAT
traversal using UDP/IP encapsulation and the mechanisms discussed in
.In some use cases, it is desirable, beneficial and efficient for the
MN to receive a constant MNP that travels with the MN wherever it moves.
For example, this would allow air traffic controllers to easily track
aircraft, etc. In other cases, however (e.g., intelligent transportation
systems), the MN may be willing to sacrifice a modicum of efficiency in
order to have time-varying MNPs that can be changed every so often to
defeat adversarial tracking.The prefix delegation services discussed in
allows OMNI MNs that desire time-varying MNPs to obtain short-lived
prefixes to use a Temporary LLA as the source address of an RS message
with an OMNI option with DHCPv6 Option sub-options. The MN would then be
obligated to renumber its internal networks whenever its MNP (and
therefore also its OMNI address) changes. This should not present a
challenge for MNs with automated network renumbering services, however
presents limits for the durations of ongoing sessions that would prefer
to use a constant address.The IANA is instructed to allocate an official Type number TBD from
the registry "IPv6 Neighbor Discovery Option Formats" for the OMNI
option. Implementations set Type to 253 as an interim value .The IANA is instructed to assign a new Code value "1" in the "ICMPv6
Code Fields: Type 2 - Packet Too Big" registry. The registry should read
as follows:The IANA is instructed to allocate one Ethernet unicast address TBD2
(suggest 00-00-5E-00-52-14 ) in the registry
"IANA Ethernet Address Block - Unicast Use".The OMNI option also defines an 8-bit Sub-Type field, for which IANA
is instructed to create and maintain a new registry entitled "OMNI
option Sub-Type values". Initial values for the OMNI option Sub-Type
values registry are given below; future assignments are to be made
through Expert Review .Security considerations for IPv4 , IPv6 and IPv6 Neighbor Discovery
apply. OMNI interface IPv6 ND messages SHOULD include Nonce and
Timestamp options when transaction confirmation
and/or time synchronization is needed.OMNI interfaces configured over secured ANET interfaces inherit the
physical and/or link-layer security properties (i.e., "protected
spectrum") of the connected ANETs. OMNI interfaces configured over open
INET interfaces can use symmetric securing services such as VPNs or can
by some other means establish a direct link. When a VPN or direct link
may be impractical, however, an asymmetric security service such as the
authentication option specified in or other
protocol control message security mechanisms may be necessary. While the
OMNI link protects control plane messaging, applications must still
employ end-to-end transport- or higher-layer security services to
protect the data plane.The Mobility Service MUST provide strong network layer security for
control plane messages and forwarding path integrity for data plane
messages. In one example, the AERO service constructs a spanning tree
between mobility service elements and secures the links in the spanning
tree with network layer security mechanisms such as IPsec or Wireguard. Control plane messages are then
constrained to travel only over the secured spanning tree paths and are
therefore protected from attack or eavesdropping. Since data plane
messages can travel over route optimized paths that do not strictly
follow the spanning tree, however, end-to-end transport- or higher-layer
security services are still required.Security considerations for specific access network interface types
are covered under the corresponding IP-over-(foo) specification (e.g.,
, , etc.).Security considerations for IPv6 fragmentation and reassembly are
discussed in .Draft -29 is implemented in the recently tagged AERO/OMNI 3.0.0
internal release, and Draft -30 is now tagged as the AERO/OMNI 3.0.1.
Newer specification versions will be tagged in upcoming releases. First
public release expected before the end of 2020.The first version of this document was prepared per the consensus
decision at the 7th Conference of the International Civil Aviation
Organization (ICAO) Working Group-I Mobility Subgroup on March 22, 2019.
Consensus to take the document forward to the IETF was reached at the
9th Conference of the Mobility Subgroup on November 22, 2019. Attendees
and contributors included: Guray Acar, Danny Bharj, Francois
D´Humieres, Pavel Drasil, Nikos Fistas, Giovanni Garofolo,
Bernhard Haindl, Vaughn Maiolla, Tom McParland, Victor Moreno, Madhu
Niraula, Brent Phillips, Liviu Popescu, Jacky Pouzet, Aloke Roy, Greg
Saccone, Robert Segers, Michal Skorepa, Michel Solery, Stephane Tamalet,
Fred Templin, Jean-Marc Vacher, Bela Varkonyi, Tony Whyman, Fryderyk
Wrobel and Dongsong Zeng.The following individuals are acknowledged for their useful comments:
Michael Matyas, Madhu Niraula, Greg Saccone, Stephane Tamalet, Eric
Vyncke. Pavel Drasil, Zdenek Jaron and Michal Skorepa are recognized for
their many helpful ideas and suggestions. Madhuri Madhava Badgandi,
Katherine Tran, and Vijayasarathy Rajagopalan are acknowledged for their
hard work on the implementation and insights that led to improvements to
the spec.Discussions on the IETF 6man and atn mailing lists during the fall of
2020 suggested additional points to consider. The authors gratefully
acknowledge the list members who contributed valuable insights through
those discussions. Eric Vyncke and Erik Kline were the intarea ADs,
while Bob Hinden and Ole Troan were the 6man WG chairs at the time the
document was developed; they are all gratefully acknowledged for their
many helpful insights.This work is aligned with the NASA Safe Autonomous Systems Operation
(SASO) program under NASA contract number NNA16BD84C.This work is aligned with the FAA as per the SE2025 contract number
DTFAWA-15-D-00030.Error Characteristics of Fiber Distributed Data Interface
(FDDI), IEEE Transactions on CommunicationsThe OMNI Interface - An IPv6 Air/Ground Interface for Civil
Aviation, IETF Liaison Statement #1676,
https://datatracker.ietf.org/liaison/1676/ICAO Document 9896 (Manual on the Aeronautical
Telecommunication Network (ATN) using Internet Protocol Suite (IPS)
Standards and Protocol), Draft Edition 3 (work-in-progress)IPv4 Address Space Registry,
https://www.iana.org/assignments/ipv4-address-space/ipv4-address-space.xhtmlIPv6 Global Unicast Address Assignments,
https://www.iana.org/assignments/ipv6-unicast-address-assignments/ipv6-unicast-address-assignments.xhtmlAdaptation of the OMNI option Interface Attributes Preferences Bitmap
encoding to specific Internetworks such as the Aeronautical
Telecommunications Network with Internet Protocol Services (ATN/IPS) may
include link selection preferences based on other traffic classifiers
(e.g., transport port numbers, etc.) in addition to the existing
DSCP-based preferences. Nodes on specific Internetworks maintain a map
of traffic classifiers to additional P[*] preference fields beyond the
first 64. For example, TCP port 22 maps to P[67], TCP port 443 maps to
P[70], UDP port 8060 maps to P[76], etc.Implementations use Simplex or Indexed encoding formats for P[*]
encoding in order to encode a given set of traffic classifiers in the
most efficient way. Some use cases may be more efficiently coded using
Simplex form, while others may be more efficient using Indexed. Once a
format is selected for preparation of a single Interface Attribute the
same format must be used for the entire Interface Attribute sub-option.
Different sub-options may use different formats.The following figures show coding examples for various Simplex and
Indexed formats:ICAO Doc 9776 is the "Technical Manual for VHF Data Link Mode 2"
(VDLM2) that specifies an essential radio frequency data link service
for aircraft and ground stations in worldwide civil aviation air traffic
management. The VDLM2 link type is "multicast capable" , but with considerable differences from common
multicast links such as Ethernet and IEEE 802.11.First, the VDLM2 link data rate is only 31.5Kbps - multiple orders of
magnitude less than most modern wireless networking gear. Second, due to
the low available link bandwidth only VDLM2 ground stations (i.e., and
not aircraft) are permitted to send broadcasts, and even so only as
compact layer 2 "beacons". Third, aircraft employ the services of ground
stations by performing unicast RS/RA exchanges upon receipt of beacons
instead of listening for multicast RA messages and/or sending multicast
RS messages.This beacon-oriented unicast RS/RA approach is necessary to conserve
the already-scarce available link bandwidth. Moreover, since the numbers
of beaconing ground stations operating within a given spatial range must
be kept as sparse as possible, it would not be feasible to have
different classes of ground stations within the same region observing
different protocols. It is therefore highly desirable that all ground
stations observe a common language of RS/RA as specified in this
document.Note that links of this nature may benefit from compression
techniques that reduce the bandwidth necessary for conveying the same
amount of data. The IETF lpwan working group is considering possible
alternatives: [https://datatracker.ietf.org/wg/lpwan/documents].Per , IPv6 ND messages may be sent to either
a multicast or unicast link-scoped IPv6 destination address. However,
IPv6 ND messaging should be coordinated between the MN and AR only
without invoking other nodes on the *NET. This implies that MN / AR
control messaging should be isolated and not overheard by other nodes on
the link.To support MN / AR isolation on some *NET links, ARs can maintain an
OMNI-specific unicast L2 address ("MSADDR"). For Ethernet-compatible
*NETs, this specification reserves one Ethernet unicast address TBD2
(see: ). For non-Ethernet statically-addressed
*NETs, MSADDR is reserved per the assigned numbers authority for the
*NET addressing space. For still other *NETs, MSADDR may be dynamically
discovered through other means, e.g., L2 beacons.MNs map the L3 addresses of all IPv6 ND messages they send (i.e.,
both multicast and unicast) to MSADDR instead of to an ordinary unicast
or multicast L2 address. In this way, all of the MN's IPv6 ND messages
will be received by ARs that are configured to accept packets destined
to MSADDR. Note that multiple ARs on the link could be configured to
accept packets destined to MSADDR, e.g., as a basis for supporting
redundancy.Therefore, ARs must accept and process packets destined to MSADDR,
while all other devices must not process packets destined to MSADDR.
This model has well-established operational experience in Proxy Mobile
IPv6 (PMIP) .<< RFC Editor - remove prior to publication >>Differences from draft-templin-6man-omni-interface-35 to
draft-templin-6man-omni-interface-36:Major clarifications on aspects such as "hard/soft" PTB error
messagesMade generic so that either IP protocol version (IPv4 or IPv6)
can be used in the data plane.Differences from draft-templin-6man-omni-interface-31 to
draft-templin-6man-omni-interface-32:MTUSupport for multi-hop ANETS such as ISATAP.Differences from draft-templin-6man-omni-interface-29 to
draft-templin-6man-omni-interface-30:Moved link-layer addressing information into the OMNI option on a
per-ifIndex basisRenamed "ifIndex-tuple" to "Interface Attributes"Differences from draft-templin-6man-omni-interface-27 to
draft-templin-6man-omni-interface-28:Updates based on implementation experience.Differences from draft-templin-6man-omni-interface-25 to
draft-templin-6man-omni-interface-26:Further clarification on "aggregate" RA messages.Expanded Security Considerations to discuss expectations for
security in the Mobility Service.Differences from draft-templin-6man-omni-interface-20 to
draft-templin-6man-omni-interface-21:Safety-Based Multilink (SBM) and Performance-Based Multilink
(PBM).Differences from draft-templin-6man-omni-interface-18 to
draft-templin-6man-omni-interface-19:SEND/CGA.Differences from draft-templin-6man-omni-interface-17 to
draft-templin-6man-omni-interface-18:TeredoDifferences from draft-templin-6man-omni-interface-14 to
draft-templin-6man-omni-interface-15:Prefix length discussions removed.Differences from draft-templin-6man-omni-interface-12 to
draft-templin-6man-omni-interface-13:TeredoDifferences from draft-templin-6man-omni-interface-11 to
draft-templin-6man-omni-interface-12:Major simplifications and clarifications on MTU and
fragmentation.Document now updates RFC4443 and RFC8201.Differences from draft-templin-6man-omni-interface-10 to
draft-templin-6man-omni-interface-11:Removed /64 assumption, resulting in new OMNI address format.Differences from draft-templin-6man-omni-interface-07 to
draft-templin-6man-omni-interface-08:OMNI MNs in the open InternetDifferences from draft-templin-6man-omni-interface-06 to
draft-templin-6man-omni-interface-07:Brought back L2 MSADDR mapping text for MN / AR isolation based
on L2 addressing.Expanded "Transition Considerations".Differences from draft-templin-6man-omni-interface-05 to
draft-templin-6man-omni-interface-06:Brought back OMNI option "R" flag, and discussed its use.Differences from draft-templin-6man-omni-interface-04 to
draft-templin-6man-omni-interface-05:Transition considerations, and overhaul of RS/RA addressing with
the inclusion of MSE addresses within the OMNI option instead of as
RS/RA addresses (developed under FAA SE2025 contract number
DTFAWA-15-D-00030).Differences from draft-templin-6man-omni-interface-02 to
draft-templin-6man-omni-interface-03:Added "advisory PTB messages" under FAA SE2025 contract number
DTFAWA-15-D-00030.Differences from draft-templin-6man-omni-interface-01 to
draft-templin-6man-omni-interface-02:Removed "Primary" flag and supporting text.Clarified that "Router Lifetime" applies to each ANET interface
independently, and that the union of all ANET interface Router
Lifetimes determines MSE lifetime.Differences from draft-templin-6man-omni-interface-00 to
draft-templin-6man-omni-interface-01:"All-MSEs" OMNI LLA defined. Also reserved fe80::ff00:0000/104
for future use (most likely as "pseudo-multicast").Non-normative discussion of alternate OMNI LLA construction form
made possible if the 64-bit assumption were relaxed.First draft version (draft-templin-atn-aero-interface-00):Draft based on consensus decision of ICAO Working Group I
Mobility Subgroup March 22, 2019.