RSSI switching values

RE: RSSI switching values

MartinBL — Fri, 29 Sep 2017 10:31:48 GMT

I don't remember and it is actually not possible to say exactly either. The X-axis in the plot is not time, but sample number. The time between each point in the plot can be anything dependent on how fast the peripheral is advertising and whether all advertising packets are picked up by the central.

But let us assume that it is ~10 samples per "level" in my plot. Then we can say that the Scan window is e.g. 100ms and the advertising interval is 10ms.

RE: RSSI switching values

Srikant — Thu, 28 Sep 2017 12:57:47 GMT

if you can, can you please tell me what is the time for the flat part of the graph (before increasing or decreasing; ignoring the small fluctuations). A approximate value would also help. Thank you.

RE: RSSI switching values

MartinBL — Thu, 28 Sep 2017 11:38:26 GMT

It is just the number of received advertising packets.

RE: RSSI switching values

Srikant — Thu, 28 Sep 2017 11:10:36 GMT

In the last image, what scale is your x-axis?

RE: RSSI switching values

MartinBL — Wed, 27 Sep 2017 09:41:48 GMT

I added some sentences and a picture in my answer that hopefully explains what I mean.

RE: RSSI switching values

Srikant — Wed, 27 Sep 2017 07:27:13 GMT

and What do you mean by different performance? does it mean slightly different powers in the three channels?

RE: RSSI switching values

MartinBL — Wed, 27 Sep 2017 06:58:32 GMT

It has been a while a go since I did it, but I'm pretty sure I:

Used a Central example from the SDK.
Scanned for packets from a peripheral with a certain UUID in the advertising packet.
Read the RSSI values (p_gap_evt->params.adv_report.rssi) on BLE_GAP_EVT_ADV_REPORT events.
Printed the values via UART to a computer and stored the values in a .CSV file and opened it in excel.

RE: RSSI switching values

Srikant — Wed, 27 Sep 2017 06:40:27 GMT

How did you read the advertisements? using a smartphone? What do you mean by different performance? does it mean slightly different powers in the three channels?

RE: RSSI switching values

MartinBL — Mon, 25 Sep 2017 14:15:14 GMT

Hi,

Could it simply be that you are scanning advertising packets with long scan windows and that the three distinct "levels" correspond to the three different advertising channels. I have seen a very similar pattern before where this was the case. Have a look at the picture below:

If you scan with relatively long scan windows compared to advertising intervals you will pick up several packets from channel 37, then from channel 38, and finally from channel 39. Now remember that even though they are numbered in sequence, the three advertising channels are spaced out in the entire spectrum:

Since it is impossible to design a radio with perfect frequency response across the entire BLE spectrum you will always have slightly different performance on the three different channels. In the image below the red line illustrates how the performance might differ over the 40 BLE channels (This is just an illustration. The curve might be very different). It shows how the antenna is able to output more power on advertising channel 38 and data channels 11-16 than e.g. on data channel 36 and advertising channel 39. More power on the air shows up in your plot as higher dBm.

The RSSI on each channel can also vary depending on your surroundings. A signal on a given frequency might e.g. travel more easily through a wall than a different frequency.

The bottom line: You have no guarantee that all three advertising channels will perform equally well. In fact, that is why there are three of them; To increase the chances of reaching your receiver. If one channel is useless due to interference or just poor antenna design, then hopefully you can still reach your receiver through one of the other two channels.

In the figure below I have reproduced a similar pattern just by playing around with scan windows and advertising interval:

RE: RSSI switching values

endnode — Fri, 22 Sep 2017 13:35:51 GMT

Btw. half a meter apart you are still almost in the near field, behavior of these things with wide spectrum of radiating things can be very far from our linear intuition...

RE: RSSI switching values

endnode — Fri, 22 Sep 2017 13:34:39 GMT

I can only say that this is what most of (cheap) Android phones do and I've seen such periodic fluctuations many times. It's very hard to say what is actual root cause (and even radio engineers might not be able to tell without spending weeks in the lab with particular electronics and isolating the things on different layers one by one) but my theory is that systems with volume of 150x150x8mm3 packed with several antennas + 4-band GSM/3G/LTE + Wifi (sometimes dual-band) + Bluetooth + GPS(Glonass...) + bunch of more or less unshielded chips and all that powered from battery, it's even mystery for me how all of that can work so well. Having +-10dBm noise on some of these radios when you are +30/50dBm from sensitivity thresholds is probably peanuts for handset makers.

RE: RSSI switching values

Srikant — Fri, 22 Sep 2017 13:11:08 GMT

In the lawn, no user presence here. The beacon and mobile device (mounted on tripod) were half a meter apart.

RE: RSSI switching values

Srikant — Fri, 22 Sep 2017 13:09:38 GMT

Hi @endnode,

Thanks for the lightning fast reply. The transmitter is nRF51822 chip but the receiver is a smartphone device. A noise would cause random fluctuations but this looks (very)periodic. Don't you think? I need to model this data for localization problem. Just wondering why this happens so that I get a better understanding about the measurement process.

RE: RSSI switching values

Matt — Fri, 22 Sep 2017 13:09:36 GMT

Outdoor as in a soccer field?

RE: RSSI switching values

endnode — Fri, 22 Sep 2017 13:02:49 GMT

What is the receiver and what is transmitter? Are both nRF5x chips with Nordic stack and custom FW? From the first look I would say: this is how typical "static" RSSI plot looks like. Even you think that both transmitter and receiver are static and that nothing is changing in the environment the signal strength and radio stages are so sensitive that even small changes and noise (e.g. from power stage on the board or inducted from any other source in the system) cause +-10dBm fluctuations. This is the main reason why most of these "location tracking" systems based on RSSI are failing (or take big effort to "smooth" the curve and cross check the data from many receivers and patterns).