Followup discussion on link identifier
Hi,
In the 802.21 ad-hoc meeting in Singapore, link identifier
contribution (21-06-0608-00-0000-Link-Identifier.doc) is accepted as
modified, with mandating link-layer address of PoA, as far as I
understand.
Let me point out additional implications of link identifier.
[1] Michael commented that PoA must be an entity that terminates the
MAC service and hence it must have a link-layer address. Michael is
right in terms of the current definition of PoA and link in the
current 802.21 draft:
"
Link - communication facility or medium over which L2 network nodes
may communicate for exchange of L2 messages. Each link is
associated with a minimum of two endpoints. Each link endpoint has
a unique link-layer identifier.
"
and
"
Network Point of Attachment (Network PoA) - Network-side endpoint
of a link that includes a MN as an endpoint. Note: The PoA
definition is associated with an interface instead of a node. Here,
the term "network" means subnet, or subnet + VLAN, or broadcast
domain, or routing domain, as opposed to media type or the entire
set of connections behind the PoA.
"
On the other hand, the current definition of PoA and link is not
useful for some link-layers for which there is no explicit *MAC-layer
signaling* for network attachment. For example, in Ethernet, a
link-up event would be generated when a UE is physically connected to
a hub or switch, even before sending/receiving any MAC frame to/from
any node. For such a link-layer, it is more practical to define a
link to allow an unnumbered link endpoint such as a hub.
Having said that, I'd suggest considering a revised definition of link
by removing the last sentence.
"
Link - communication facility or medium over which L2 network nodes
may communicate for exchange of L2 messages. Each link is
associated with a minimum of two endpoints.
"
In this case, the link identifier can still contain a link-layer
address of PoA field, but depending on link types, the field may be
processed just as a placeholder.
[2] Michael commented that if we can use 802.21 link identifier to
represent a sequence of links in order to deal with end-to-end paths.
We can certainly use the 802.21 link identifier for such purpose, but
it may also be possible to use IETF DNA link identifier or some other
identifiers for the same purpose. This might need more discussion and
I would like to move forward the current 802.21 link identifier
definition as it is for now, without changing the field names from
link-layer address of MN/PoA to something else.
Thank you for your attention.
Yoshihiro Ohba