[net2-wg] link estimator interface

[Joe Polastre]

see previous message:

That's why you need a correlation factor, C, based on actual measured
packet error rate.  LQI is a quick start mechanism that supplements
the long term packet error rate.

This is documented in numerous places (cough, Zigbee specification,
cough), but for some reason no one has chosen to implement it.  Sigh.

-Joe

On 3/6/06, Philip Levis <pal at cs.stanford.edu> wrote:
> On Mon, 2006-03-06 at 13:35 -0800, Joe Polastre wrote:
> > Again, correlation value is the state-full function, as per the specification.
>
> I'm confused. Are you talking about this function:
>
>   uint16_t correlation(uint8_t v) {
>     uint16_t c = (80 - (v - 40));
>     c = (((c * c) >> 3) * c) >> 3;
>     return c;
>   }
>
> This is what all of the Moteiv code seems to use. It's stateless.
>
> It also does more than what ChipCon recommends in the CC2420 data sheet,
> which is
>
> LQI = (CORR - a) * b
>
> which is also a stateless, linear function. Neither the ZigBee nor
> 802.15.4 specification seem to say anything about a stateful link
> estimation function. Which specification are you thinking of?
>
> Phil
>