# Over-the-air (OTA) BER measurement

I'm in the process of evaluating BER performance for a GNU Radio BPSK receiver. The setup for the experiment is illustrated below. Using a spectrum analyzer just before the USRP, the average channel power (in the bandwidth of interest) without a BPSK signal is measured in order to determine the noise level (N). After this, a BPSK transmitter is turned on and the average channel power (C + N) is recorded. Defining $$\Delta$$ as the difference between the two readings, the $$E_{b}/N_{o}$$ value can be calculated by the equation:

$$\mathsf{\frac{E_{b}}{N_{o}} = 10\log\left(10^{\frac{\Delta}{10}} - 1\right) - 10\log\left(\frac{W}{Rb}\right)}$$

The spectrum analyzer, shown below, is capable of showing the instantaneous average channel power level. The problem I'm facing is the power level variation of up to ~1dBm. Given that $$\Delta$$ is a difference between two readings, its variation is ~2dB.

This is a pretty big variation for BER evaluation. I was wondering what is usually done in practice when BERT involve physical hardware. All pieces of advice are highly welcomed.