Scope of the MIH PoS definition
A point that came out during today's conference call on the MIH Communication
Model:
Should the definition of MIH PoS be associated with a "network node" or with an
"interface"?
In my opinion, the answer depends on the amount of information that we want to
convey with the "MIH PoS" denomination. The group should discuss the two options
and choose one.
Association with the network node (i.e., having MIH PoS as a node attribute that
applies on a per-UE basis) does not convey any information on the type of
transport that may or may not be used to support the MIH exchanges with the UE.
This is because a node-level attribute cannot qualify the MIH capabilities of
the individual interfaces.
Association with the interface (i.e., having MIH PoS as an interface attribute
that applies on a per-UE basis) conveys instead a clearer indication of the type
of transport to be used for MIH exchanges with the UE.
Please note that here "interface" is not meant as a physical port in the node,
but rather as the virtual entry point to that port at a given protocol layer.
The same physical port can be associated with one or more L2 interfaces (one L2
interface in the 802.3 case, one L2 interface per VLAN ID in the 802.1Q case,
one L2 interface per CID in the 802.16 case, ...), and one or more IP interfaces
(if the node has a routing table, it would be reasonable to associate a
different IP interface with each entry in the routing table that points to that
physical port).
Here is an example to illustrate the two options:
A Node has (among others) an 802.11 port and an 802.16 port. The 802.11 port is
not MIH capable. The 802.16 port is MIH capable.
Case 1: MIH PoS is an Interface attribute.
A UE that is attached to the Node through its 802.16 port has an MIH PoS on the
Node termination of the 802.16 connection (or link). Saying that the MIH PoS is
associated with the 802.16 link (i.e., the PoA and the MIH PoS of the UE are
co-located) implies that native L2 transport MAY be used to exchange MIH
information between the UE and the Node.
A UE that is attached to the Node through the 802.11 port does not have an MIH
PoS associated with the 802.11 interface on the node. Saying that the MIH PoS is
not associated with the 802.11 L2 interface implies that the native L2 transport
of the interface MAY NOT be used for transport of MIH information.
Case 2: MIH PoS is a Node attribute.
The Node is the MIH PoS for any UE that exchanges MIH information with it. Extra
information must be added to the "MIH PoS" tag if we want to determine the type
of transport to be used for MIH exchanges.
One interesting point from the discussion of Case 1, 802.11 interface: Ruling
out the native (i.e., 802.11) transport for MIH messages does not automatically
imply the use of L3 and the association of the MIH PoS with the L3 interface to
the UE. It is possible in fact that the node that includes the 802.11 PoA is
part of a bridged network (802.1D or 802.1Q) that forwards packets based on
their Ethernet headers. For MIH exchanges between the UE and the Node, a new
EtherType in the Ethernet Header could be used to identify MIH-related frames,
without involving IP. We are therefore in a situation where L2 transport is used
for MIH information, but not the same L2 as the link between UE and PoA.
Is this situation really possible? If yes, is the L2 vs. L3 transport
distinction still useful? Shouldn't it be replaced with the distinction between
native L2 (the L2 of the PoA) and everything else (where the "everything else"
is dictated by the general transport model of the network and not by the MIH
capabilities of the nodes/interfaces)?
Andrea