[CUWiN-Dev] G and A and realworld throughput

[David Young]

On Tue, Jul 19, 2005 at 11:22:45AM -0500, Stelios Valavanis wrote:
> i couldn't listen to the metrofi "radio" piece (what the hell is a 'wax' 
> file). from their site i see thye are using proprietary wireless with some 
> kind of CPE. i wish they had more on the backhaul (my issue here) which i bet 
> was in the "radio" piece.
> 
> how does roofnet compare with cuwin?

We have different emphases.  Somebody do correct me if I'm wrong,
but I believe Roofnet emphasizes research.  It's not about the code,
it's about writing the research articles, and graduating.  The Roofnet
of today only has to be reproducible for the duration of the current
Ph.D. students' schooling.

In contrast, CUWiN is development-oriented.  We develop robust
implementation code from the available research in order to help our
"clients" meet their goals cost-effectively, especially where their goals
coincide with ours: enable end-users to build autonomous networks that
meet their community's data communication needs at a low cost.

CUWiN has emphasized scalability, autonomicity, and ease of use.
You could say that CUWiN is all about "plug & play."  A CUWiN node is an
appliance: you plug it in, and it provides a network.  You don't have
to know "how to work it."  Virtually all aspects of configuration, we
have automated.  A node even figures out whether to act as an Internet
gateway or not.

We have future-proofed the system in a lot of the ways.  Portability from
architecture to architecture is an important aspect of this.  The system
is also miniaturizable to fit on cheap, mass-produced routers.  And we
have the only integrated wireless mesh solution that I'm aware of which
routes both IPv4 and IPv6---I don't think that any of the commercial
mesh offerings do that, yet.  For peer-to-peer networking, we offer the
most scalable routing.

We reuse the "Routing Information Base" in Quagga, so there is the
potential to interoperate with all of the other routing protocols in
the Quagga suite, including OSPF, RIP, and BGP, and build a seamless
wireless/copper/fiber network that is routed from end to end.  Our system
is unique in this regard.

CUWiN's system can offer a lot of advantages to developers.  We have the
best build environment.  You can cross-build the whole system from sources
on Linux, FreeBSD, or NetBSD (any POSIX system will do) for all important
target architectures (x86, PowerPC, MIPS, ARM) and even some unimportant
ones (Sparc64, anyone?).  The code is thoughtfully structured and written,
and it has a lot of instrumentation to aid diagnostics and debugging.
Our system would be a fine basis for developing a commercial product.

In my estimation, the CUWiN developers are the best of anyone working
in the ad hoc wireless area, *dollar-for-dollar and pound-for-pound*
(excepting underpaid graduate students), at diagnosing and fixing bugs,
and producing quality designs and code.  You have to understand, though,
that less than $250k has been invested in this project, compared with $10M
or more invested in the wireless firms.  Even as low as our overhead
is, and even if we are twice as effective, we are at a significant
disadvantage, development-wise, compared to the Tropos and BelAirs and
Navinis of the world.

Dave

-- 
David Young             OJC Technologies
dyoung at ojctech.com      Urbana, IL * (217) 278-3933