[ns] sensor networks

[Sagnik Bhattacharya <sagnik@cs.virginia.edu>]

I have to conduct some simulations as part of my research on sensor
netowrk applications. I'll be very glad if anyone could give me their
opinions on the subject of viability of ns-2 for my simulations. I'll
briefly descirbe our scenario and requirements to you.

   A typical scenario shall have in the order of 10,000 nodes spread on a
rectangular grid, communicating between themselves using a single
frequency wireless channel, but using multihops. Each of these nodes is
severely constricted in battery power and and can do a very limited amount
of processing as the processing and memory capacities are very low. That
means that we can't have a heavyweight protocol stack. Also, the MAC
protocols need to be simplified. We want to test our applications on top
of a lightweight communication protocol layer.

   Also we want to introduce a new notion of addressing. Instead of nodes
having unique addresses, we want to address each node by its location. And
we want to obtain our simulation results over large periods of time.

   So, could anyone please give me an assesment of how ns
will scale to these large scenarios and how much effort is involved to
implement our ideas in ns?

Thanks
Sagnik