[Tinyos-devel] more bugging about current consumption on MicaZ in tinyos-2.x

[henri dubois-ferriere]

On 19/07/06, Eric Weddington <eweddington at cso.atmel.com> wrote:

> Being new to the area of wireless sensor networks, I've been wondering
> about the idea of "off the shelf" platforms. How hard is it to introduce
> a new design? And now for the obvious bias: Specifically, how hard is it
> to introduce a new design with the new Atmel AT86RF230 radio? and
> possibly an ATmega1281? It would be interesting to stack it up against
> the current "off the shelf" platforms.

What do you mean by "how hard is it to introduce a new design", do you
mean porting tinyos to this new design, getting people to adopt it,
or doing the design itself?
If you mean the former, then here is some high-level feedback:
Writing a driver for a new radio is  clearly  a good chunk of work,
but there are many existing drivers for radios with diverse
interfaces, so you have lots of models to go by.


Supporting a new MCU will also be some work, and possibly less smooth
than for a radio, because tinyos has been made to run on more radios
than MCUs, and so things are probably more mature on that side. But
then again, the MCU abstractions are now quite well defined. The other
practical issue is whether gcc supports the MCU. If it doesn't then
that's another chunk of work (david gay has explained somewhere on
this list or on tinyos-help what needs to be done to use a non-gcc
compiler with nesc, don't have the direct link at hand unfortunately).

[Of course in either case things are much easier with intimate
knowledge of the radio/mcu, as will obviously be the case at Atmel.]

And in general I think people are glad to see new platforms emerge
that have proper tinyos support.

Henri