Hi, Everyone:
We will have the conference call tomorrow (3/23) related to H.323 mobility. We discussed the mobility related contributions in the last SG16 meeting held on Feb'00. I am attempting to clarify some issues so that we can save our valuable time in the conference call making a rapid progress.
I believe that we have made an agreement in the last Feb'00 meeting as follows:
1. We will address the mobility management in the H.323 layer that will need location updates due to the changes in the H.323 point of attachments (e.g., network addresses, zone/domain boundaries). 2. We will NOT address handoff in the H.323 layer assuming the fact the handover is taken care-of transparently in the lower layer. That is, we will not open or close connections/media channels (e.g., Q.931, H.245 OLCs, etc) in the H.323 layer for the purpose of handover.
The communication environments that we may have to deal with where H.323 mobile phones/terminals are being used over the IP network infrastructure are as follows:
1. The cell-based wireless networking environment with 100s, 1000s of users 2. The wireless LAN environment 3. Mobility in the wire-line network
Couple of basic problems that we might have to address in H.323 mobility:
* A method for the home and foreign GK discovery by which a mobile entity can perform the following dynamically :
1. Determines whether it is currently connected to its home network/GK or to a foreign network/GK and 2. Detects when it has moved from one network/zone[GK]/domain to another (network/zone[GK]/domain) 3. Location updates among the home, visiting, and visited GK (and possible use of the VLFs/HLFs) dynamically
The "discovery" mechanism is a classical problem whether it is a wireless or wire-line network. The solution to this problem has been to send the "advertise" message using multicast or limited broadcast. There has not been any unicast solution for this problem.
For example, people can trace the origin of this "discovery" problem in the case of finding the default router. IETF has provided the ICMP message that uses "multicast" or "limited broadcast."
IETF's mobile IP folks have also used the extended the ICMP message for their agent discovery using multicast or limited broadcast.
The basic conclusion seems to be that we have to use the GK discovery mechanisms using the "multicast" since no unicast solution is available.
Some mobility support that already exists in H.323 (or can be combined with the lower layer mobility solution for which no new H.323 mobility standard work is needed):
1. Location updates with re-registration: This mechanism already exists in H.323 where it is statically determined (that is, either by manual or by other means) that there is a change in network address (in a given zone), the re-registration of the H.323 entity is made. 2. Communications with the home GK while the H.323 terminal has moved to a different zone/domain: If it is known statically (that is, either by manual or by other means) that the H.323 entity is in a different zone/domain, it can attain a new network address (e.g., IP) using DHCP and then the communication with the home GK can be established directly by the H.323 mobile entity. Alternatively, an mobile IP solution can also be used in the IP layer to perform this job using the IP layer home and foreign agent and then can communicate directly with the H.323 home GK.
I would appreciate if all members provide comments on this as well.
Best regards,
Radhika R. Roy AT&T H.323 Ad Hoc Mobility Group + 1 732 420 1580 rrroy@att.com