1

I have the problem of calculating the Signal to Noise Ratio for a packeted transmission using FSK or BPSK modulation with convolutional coding on a micro controller unit which communicates with an RF chip (AX5043).

The only information that is available is the RSSI, background RSSI and the packet error rate which will be calculated on the MCU. The RSSI is known as being not so reliable for SNR calculation, so my question is if there is another way to estimate the SNR, perhaps using the PER.

Marcus Müller
  • 30,525
  • 4
  • 34
  • 58
Sleepwalkr
  • 11
  • 1
  • You could make a model assumption that SNR is the only thing that leads to errors, and estimate SNR based on that. But: That assumption is known to be wrong in practice, and so SNR is neither really the relevant number nor actually sensibly observable. What's your motivation to estimate SNR? For what purpose? Maybe we can find another way. – Marcus Müller Nov 05 '19 at 11:34
  • There's a number of reasons. First, i have calculated some link margins and i'd like to validate them through actual measurements. So i would have to calculate the SNR at the receiver and see over which threshold i can decode FSK, PSK, etc. This would also lead to a characterization of the RF chip that i'd like to do. Second i'd like to provide some Eb/N0 vs BER/PER curves. The idea is to estimate the SNR over the period of a burst transmission (or a number of them) – Sleepwalkr Nov 05 '19 at 11:55
  • So, but your threshold of decodability is defined by the actual hardware, so your SNR isn't actually as helpful as your PER for analyzing wether the goal of your link budget has been met. If you want to characterize your chip, I'd argue you'll need an external measurement device for SNR first – what you're trying here is estimate SNR based on a characterization of your device, to get a characterization of your device, which you would have needed for the SNR estimate. – Marcus Müller Nov 05 '19 at 12:10
  • Unfortunately no additional hardware can be added at this point, so i have to make due with what is already there. So you suggest that PER suffices for the link margin validation? And what about the BER curves? The goal of the experiment is to showcase the difference in performance between various modulations with the specific setup – Sleepwalkr Nov 05 '19 at 12:28

0 Answers0