Cell Relay Archive

[rajeev@trillium.com (Rajeev Gupta)]

> From: albert.e.manfredi@boeing.com
> 
> The biggest problem with "home server" is that if I am traveling in
> Keflavik, I sure as hell don't want someone from Helsinki have to call
> me via a server in Johannesburg. I want to achieve a true mobile
> capability, where calls are placed as efficiently as stationary unit
> telephone calls.

This has more to do with the economics of charging and billing
(particulalry across international boundaries) than technology.
Certainly the network has the information to route the call efficiently,
which is fine in a private network. But if you have public networks that
charge for usage, and the call crosses multiple networks, then every
operator needs to be compensated.

A simple way to do this is to use triangular routing. The caller pays
the cost from his network to the mobile's home network (like any other
phone call), while the callee pays the cost from its home network to its
current network location (like any other phone call). This also
alleviates issues like informing the caller of call charges. Of course,
there can be endless debate about whether the caller or the callee
should pay for roaming charges (depending on who you think benefits from
the user's mobility).

It is technically possible to route the call without going through the
home network, but then all transit networks have to establish and
transact billing information with the home network - again, possible to
do but potentially expensive to implement.

Nevertheless, you cannot get around the fact that the home network has
to be contacted to get the subscriber service and billing data.  The
home location registers serve as a distributed database of all mobile
subscribers, and data is temporarily cached at visited locations when
needed. It would not be very scalable to require that all
networks/switches store the subscriber data for all mobile terminals -
not to mention the political-ecomonic infeasibility of this in
international public network scenarios.
 
> 1. Dial the unique, flat, address of a mobile unit, from anywhere on the
> planet (request for mobile location could be identified with an AFI),

Flat addresses are not scalable or routable. You have earlier suggested
having a permanent prefix -- that makes the address hierarchical and
assuming that the prefix is routable, no different from a regular
directory number.

> 2. Have that request for mobile location (full, hierarchical, temp address)
> routed to the closest earth-based uplink,
>
> 3. Have the request query a satellite-served mobile Directory Service,

Here you gloss over the fact that every terminal (mobile or stationary)
needs to be connected to a service provider. That service provider
maintains the subscriber data such as what services are requested,
current location, etc. Service providers will be very unwilling to
share their subscriber data with others, without specific bilateral
arrangements. 

In your model, who would own the satellites? If you divide the satellites
by ownership, then you end up with "home" satellites -- conceptually
no different than location registers (except that you put them up
in a satellite).
 
I think your proposal is not very different from what occurs now
anyway, except that you want to put location registers in satellites.
You still have to account for how public networks operate, how charging
and billing will occur, how subscriber data will be shared, how
to build large,scalable networks and how to build large, scalable
databases.

______________________________________________________________________
Rajeev Gupta                            Trillium Digital Systems, Inc.
mailto:rajeev@trillium.com                     http://www.trillium.com