WARP Project Forums - Wireless Open-Access Research Platform

RSSI is an analog signal generated by a power detection circuit in the MAX2829 Rx path. The RSSI detector is fed before the Rx baseband amplifier. Thus the Rx baseband gain does not affect RSSI. The Rx RF gain (LNA gain) does affect the RSSI. There are 3 valid LNA gain settings. In WARPLab these are encoded as 1 (lowest gain), 2 and 3 (highest gain). The RSSI plots in the MAX2829 datasheet refer to these settings as low, medium and high.

The RSSI signal is sampled by a 10-bit 10MSps ADC on the WARP v3 board. The ADC digital values [0,1023] span approx 0.6 to 2.6v, covering nearly the full range of the MAX2829 RSSI levels.

Calibrating Rx power requires specification of the bandwidth over which you're integrating power. This implies knowing the characteristics of the waveform. The RSSI vs Rx power plots in the datasheet assume a 20MHz Wi-Fi preamble. The RSSI signal will be proportional (in dB) to your signal's Rx power, but if your signal is not a 20MHz Wi-Fi preamble there will be an offset relative to the plots in the datasheet. If you need to calculate accurate Rx power from RSSI you will need to calibrate with your waveform and a known Rx power or known Tx power and path loss.

RSSI is nearly linear-in-dB for most Rx powers. But RSSI will saturate for out-of-range powers. For example, if the LNA is set to "High" (3 in WARPLab) and a Wi-Fi signal arrives with Rx power of -20dBm, the RSSI signal will saturate at its maximum value. In this case the LNA must be adjusted to 2 or 1 before the RSSI signal will accurately reflect Rx power. In the 802.11 Ref Design the Rx PHY captures the RSSI after the AGC settles and during the preamble LTF. This provides the best chance of an accurate Rx power estimate.

murphpo wrote:

RSSI is nearly linear-in-dB for most Rx powers. But RSSI will saturate for out-of-range powers. For example, if the LNA is set to "High" (3 in WARPLab) and a Wi-Fi signal arrives with Rx power of -20dBm, the RSSI signal will saturate at its maximum value. In this case the LNA must be adjusted to 2 or 1 before the RSSI signal will accurately reflect Rx power. In the 802.11 Ref Design the Rx PHY captures the RSSI after the AGC settles and during the preamble LTF. This provides the best chance of an accurate Rx power estimate.

Hi, I have a follow-up question about measuring the input power. I noticed that in the OFDM examples such as "wl_example_siso_ofdm_txrx", the snr is computed based on EVM, which makes sense to me. However when we did the experiment in an anechoic chamber, we still observed the change of SNR as the distance between the transmitter and receiver changes. So my question is what causes the "interference" in the process of computing SNR based on EVM since the anechoic chamber is supposed to have no other interfering signal. Or to put it in another way, does the SNR computed based on EVM has the same "meaning" as the SNR computed from the signal power divided by the noise power? Thank you!

Last edited by Slime (2016-Apr-22 00:18:26)

chunter wrote:

Noise at a receiver is fundamentally unavoidable. The EVM-based SNR calculation in the "wl_example_siso_ofdm_txrx" would be better served by being named SINR (signal to interference and noise ratio). Moving into an anechoic chamber significantly attenuated the interference term that might be present in that calculation, but it didn't do anything to affect the noise floor.

I see. Thank you for the explanation. BTW during our experiment we noticed that the output power of RF port B seems to be greater than RF port A on both of our two WARP V3 boards. I was wondering if this is normal? Thank you.