About the cc2530 RSSI and the correlation value

The correlation you are refering to is a correlation of the preamble to expected preamble (or possibly the sync word or some other known data pattern) which is not defined in the 802.15.4 specification.  LQI is an ambiguous term and can mean different things in different contexts.  This radio adheres to the 802.15.4 spec so LQI is with respect to that definition.

The reason you are getting inverted results from what you expect is most likely due to multipath issues.  In effect, signals are bouncing off of wall, benches, you, etc. in your test environment and constructively and destructively combining throughout the space.  This has the effect of creating a three dimensional standing wave with local minima and maxima throughout the space.  Thus, if node A is sitting in a local minima with respect to node B's 3D standing wave, and it is at a local maxima with respect to node C's 3D standing wave, you will measure a higher RSSI from node C.

Depending on the amount of reflection and attenuation from the reflective elements in your testing space, you can easily see as much as 40dB of variation in signal strength by moving any radio by only 1/4 wave length which at 2.4GHz is only around 3cm!

One way to visualize how significant this issue can be is to visualize a single cycle of your carrier signal.  In this single cycle, you will have at least two minima and two maxima (remember we are measuring power density so it is cos(omega)^2).  At 2.4GHz, a wavelength is approximately 12.5cm so you will have to local minima in 12.5cm of linear length.  Since you can fit around 8 of these waves in a linear meter, you now have the possibility of 16 local minima in a linear meter.  Now extend the linear meter into a square meter and you now have 16x16=256 possible local minima in a square meter.  Extend once more for a cubic meter and you have 16x16x16=4096 possible local minima in a cubic meter of space.

So the answer is that you must move the local minima around to get the actual highest RSSI reading if you are going to measure distance by this method.  You can achieve this by moving a radio or you can achieve this by moving objects in the local space of the radios as both methods will move the 3D standing wave pattern.  This is usually not an acceptable answer so the only other option is to change frequencies.  This option has possibilities.  Thus the suggested method for doing this is to make measurements on multiple frequencies (seperated by 20MHz or so) and use the highes RSSI reading from all frequency measurements.

The idea is that you move the local minima around by changing the frequency (which also moves the local maxima) and you should be able to get at least one local maxima measurement over the set of frequencies.  It would be possible to have a pathological case where you are always at a local minima but the probability of this is now much lower by makeing multiple measurements over multiple frequencies.

Take a look at this design note for more information, particularly the section on frequency diversity.

http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=swra317

Jim

hi,

        first or all,Thank you for patient answer.i think i will have a try the "4 Single Radio Amplitude-Based Antenna Diversity",but i can't find the board "Antenna Diversity EVM, Rev B" in ti.com

        if i use more than 2 frequencies in 2530,Whether can cause child node lose father node?i know it will rejoin the net,but some time it won't,or after long time ,why?

       thanks

This is most likely due to some change in timing on one of the child nodes.  For example, if both try to send data at about the same time, then one will fail the CCA (Clear Channel Assessment) test.  In that case, it will back off for a random amount of time and then retry the transmission.  If during the back off period the frequency should change, then the code waiting to do the retry doesn't know about it so it doesn't know to change frequencies too.  In effect, the backoff period has generated a delay in your timing.  Now it is lost and on a different channel.  Now there will never be any conflicts because is always on the wrong channel.

You can setup a timer to know when to change frequencies but there is no guarantee.  You don't want to change the frequency in and interrupt thread as you cannot know if you are in the middle of a transmit for receive (actually you can figure this out but you still have to decide when to change channels and you don't want to sit in the ISR until it's ok to do so).

I'm not sure it this is really the problem, I'm just guessing but in my experience, managing the timing of change of frequency has historically been a huge hurdle to overcome.

As far as the antenna diversity EVM board, those were never actually put on the web for normal purchasing channels.  I can get you a couple if you like but they were designed for the 900MHz band so the antenna's are spaced incorrectly for 2.4GHz systems.  Also, the RF path will need to be tuned for the frequency change too.

Jim