[MLB-WIRELESS] Re: Node x is over this way -was- Applications

[Ben Anderson]

> > darn those roaming nodes :-)
> > i have a hunch there's a way around that,
>
> Well, I have wondered is it's possible to have an addressing scheme that's
> split into two components - the "network location" and the "station ID".
>
> The scheme works like this:
>
> For routing purposes, the network ID is used to decide which node to send
> packets to.
>
> For applications, they are only interested in communicating with the
device,
> so they only talk in terms of the station ID.  However, the network sees
the
> packet address of having an address of network ID + station ID.
>
> For fixed stations, they are configured with a fixed, non zero network ID
> (or can be assigned one by a transform from things like GPS coordinates).
> Each AP/routing node would have its own network ID, and the mojo system
annd
> the rest would still work.

A transform of GPS, to be useful in propegating data in the correct
direction without having to do whole-network broadcast discovery, is no more
secure than just giving people the gps co-ordinates, it has to be a
reversable algorithm.

> Mobile stations would have a network ID of zero, which means "Get my
network
> ID from the nearest AP").  Their node ID would be non zero and unique,
just
> as in the case of routing nodes.
>
> There might be a bit of broadcasting required when nodes move ("Is Station
> xxxxxx out there"?), but once a connection is established, and/or nodes
have
> cached the location information, the routers would automatically attach
the
> mobile node's temporary network ID to the station ID for routing
purposes...
>
> Make sense?

Your language makes sense...  I have a couple of problems with the technical
detail...  First, can you define "networkID" and "stationID" more
accuratly...  I'm guessing networkID is the GPS style address, and stationID
is just a uniquie identifier, like an ip address.
For a node with no physical location address, for a node to be able to
arbitarily say "i want to send a packet to 10.87.33.55" that packet is then
going to be have to sent to the whole network, as there's no way to figure
out which way to start propegating the packet.  (fixed-home node cache is
what I proposed to fix this).  For fixed nodes, it doesn't sound any
different to me than what I proposed.  (gps address at the data-link layer,
ip address at the network layer)  Or possibly squeezing another complete
layer in there ;)

> Clear as mud?
>
> :)

Good enough, how'd I go on the interpretation?

Ben.


--
To unsubscribe, send mail to minordomo at wireless.org.au with a subject of 'unsubscribe melbwireless'  
Archive at: http://www.wireless.org.au/cgi-bin/minorweb.pl?A=LIST&L=melbwireless
IRC at: au.austnet.org #melb-wireless