Cell Relay Archive

[albert.e.manfredi@boeing.com]

In article <199802251912.LAA15745@thranduil.trillium.com.trillium.com>,
  rajeev@trillium.com (Rajeev Gupta) wrote:

[ ... ]

Thanks for the thorough description, Rajeev.

> In WATM, the handover issue is still far from being settled. The basic
> architecture indicates that when the mobile terminal moves (it
> determines this based on measured radio signal strength, etc.), it
> requests the old switch to initiate a handover, giving to it a list of
> target new radio ports. The old switch will then query the targets and
> based on resource availability, select one of them to be the new port
> (note that the new port can be on the same or different switch).
>
> At this point, there are a number of ways in which the connection can be
> rerouted to the new switch. One possibility is to extend the connections
> from the old switch to the new switch (quick to execute but may waste
> network resources). Another is to reroute connections from a fixed
> anchor switch in the network (similar to anchor MSC routing in cellular
> networks). Another possibiity is to build a new path all the way from
> the calling party to the new switch (latency may be too high). Finally,
> a partial reroute may be done by finding an intermediate cross-over
> switch that can switch the connection from the old switch to the new
> switch, such that the new path is optimal.
>
> There is considerable debate in the WATM group on which method(s) to
> choose or whether to standardize anything at all. Another area of
> discussion is how the mobile terminal communicates with the switch to
> conduct handover signalling e.g. directly with the switch on the UNI or
> with the radio access point or some combination.

Yes, I see that the solutions being considered are in fact variations upon
mobile IP solutions, as you said. Seems to me the more "final" solution would
be different, though. More of a built-in PNNI-like solution, but it's
probably prohibitively demanding in terms of table sizes.

If each mobile station has a globally-unique MAC-like address (i.e. flat
space), then in principle, whenever such a station comes on line, it can
retrieve its temporary prefix (autoconfiguration) from the nearest available
WATM switch.

As one moves about, the prefix assigned to your MAC-like address changes. So
the trick is to have a sort of DNS-like network that keeps track, in real
time, of the mapping between all of these MAC-like unique addresses and their
temporary prefix.

If one turns off one's mobile unit, obviously the network will not know what
prefix to assign, so it assigns none. And the DNS-like table shows no
mapping, so any call to the mobile unit gets a "no connection" signal.

If the mobile unit is operating and in range of a switch, it gets assigned a
prefix.

I kind of won't be satisified until something like this shows up. Of course,
the DNS-like network of servers will have to have an entry for every single
solitary mobile unit on earth, but perhaps we're not too far away from making
that feasible?

Bert
manfredi@arl.bna.boeing.com

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/   Now offering spam-free web-based newsreading