Roy, Radhika R, ALTEC
rrroy at ATT.COM
Fri Aug 28 19:10:37 EDT 1998
> Hi Everyone:
> If we consider some examples how mobility problems have been solved in
> different situations, we can find some clues how H.323 mobility problems
> should be addressed.
> In this context, mobile IP and mobile MAC (mobile MAC is not a standard,
> but it is used in many corporate networks). Let us assume both use IP/MAC
> protocol over the LANs. (In addition, there can also be other kinds of
> mobility environment: Virtual LAN layer 2 and 3, wireless, ATM, etc. I am
> not considering those for the sake of simplicity).
> If we consider that both of them are addressing the device mobility, we
> can analyze why they are using two different approaches to solve the same
> mobility problem. By device mobility, it is meant that a use will move
> from one place to another with his or her device.
> Mobile IP:
> It considers that the IP address that is being used by a device should
> remain the same or fixed. That is, this device will always be reachable
> using the same IP address whether the device remains in its home network
> or moves to a different network.
> This problem has been solved based on the above requirement.
> The price that has been paid by this is as follows:
> 1. Needs to have home and foreign agent that will tunnel or de-tunnel
> the IP packet. Note that the IP address of a device changes as soon as the
> device is attached to new network. It is the agents that encapsulate or
> de-encapsulate the IP transparently to the end user. I am not going
> in-depth in discussing this. This is done within network, and the end use
> is not affected.
> 2. New applications have been added to the end use's devices:
> Registration and Authentication. Registration is needed purely for solving
> the mobility problem, while authentication is needed from security point
> of view. These new applications are needed by the end user device.
> Mobile MAC:
> It is much more robust. It assumes that as the device moves from one
> network to another, there is a server in each LAN that detects the
> movement of the MAC addresses through listening. There is a central server
> in the network that has the repository of all MAC addresses that are being
> stored since the service is provided through subscription only.
> Here registration is not needed to solve the mobility problems. However,
> authentication may be needed for security purposes.
> Comparison between Mobile IP and Mobile MAC:
> 1. Both approaches solve the same problem: End user's device mobility.
> 2. But the architecture and the functional entities that are required
> are quite different.
> How do we apply this above insights of our findings in the context of
> H.323 mobility?
> H.323 Mobility:
> In H.323 mobility, we are dealing primarily with the following situations:
> 1. Transport level addresses: RTP, RTCP, UDP, TCP port addresses.
> 2. GK in each Zone (a GK will also have a network level address).
> Now we can look back.
> A use may move from one network to another with the device. As a result,
> the GK in which the registration of transport addresses has been made will
> If appears that we may have to define the home and foreign GK something
> like that.
> Then, we may have to deal with the problems of registration of the user
> device when it moves to the new network.
> Finally, we have to deal with the transport addresses in mobility
> The question that I have: Can we use the "Mobile IP" transparently to
> solve H.323 mobility problems? Can we use other adopted "Mobility Scheme"
> transparently to solve H.323 mobility problems?
> If the answer is NO, then we have to design a new H.323 mobility
> architecture to solve H.323 mobility problems.
> That is what I have to say for now.
> Thanks and regards,
> Radhika R. Roy
> AT&T, USA
> Tel: +1 732 949 8657
> E-mail: rrroy at att.com
More information about the sg16-avd