WARP Project Forums - Wireless Open-Access Research Platform

The RAM contains the phase values of the channel estimates. The subtraction computes the phase error of each received pilot tone, as the actual phase of the received pilot minus the phase of the pilot subcarrier's channel estimate.

B. About phase error calc block.

murphpo wrote:

The phase error calc block extracts the pilot tones from each OFDM symbol and estimates the phase error of each pilot tone (vs. the known value for the pilot). The phase error estimates are averaged to generate a single phase error value per OFDM symbol. This value is fed into the phase noise tracking block

Do you mean the ofdm symbol  have"one"phase error value?

We have check the output waveform.


In the waveform，we use the four pilots calculate the avg. value and the phase error has four values.
We think it seems segmented concept.
   
ex.One ofdm symbol has 64 sub-carriers，and pilot location are 8 、22、44、58，and 1 sub-carrier~8 sub-carrier has one phase error value... 9 sub-carrier~22 sub-carrier has phase error value ....etc

Is it correct?

dear murphpo:

about post#29,

In the waveform，we use the four pilots calculate the avg. value and the phase error has four values.
We think it seems segmented concept.
   
ex.One ofdm symbol has 64 sub-carriers，and pilot location are 8 、22、44、58，and 1 sub-carrier~8 sub-carrier has one phase error value... 9 sub-carrier~22 sub-carrier has phase error value ....etc

Is it correct?

In the waveform，we use the four pilots calculate the avg. value and the phase error has four values.
We think it seems segmented concept.
   
ex.One ofdm symbol has 64 sub-carriers，and pilot location are 8 、22、44、58，and 1 sub-carrier~8 sub-carrier has one phase error value... 9 sub-carrier~22 sub-carrier has phase error value ....etc

Is it correct?

The PHY uses the pilot tones to estimate and correct phase error due to residual CFO (CFO left after the time-domain coarse CFO estimation+correction has been applied). Residual CFO manifests in the frequency domain as a phase offset which is constant across subcarriers in each OFDM symbol and increases with each OFDM symbol. Because the phase error due to CFO is equal across subcarriers, the phases of the 4 pilot tones are averaged to compute a single phase error estimate. The resulting phase error correction applies a single phase value to all subcarriers in each OFDM symbol.

1.Use the pilot phase error calc block which it can get the phase error value.
Why this model can not use this value to compensate the phase error.

2.Why this model need to add the phase noise tracker block to compensate the phase error?

The phase error tracking system is not strictly required, but will improve performance in situations where noise may degrade the phase estimates extracted from the pilot tones. The "Real-time Phase Tracking" papers we discussed above describe this in detail.

That ResetRx signal will assert when the PHY detects an error in the received header (CRC fails) or if header indicates an invalid packet length (good CRC, but invalid LENGTH field). When the ResetRx signal asserts the PHY pipeline is flushed immediately - the PHY does not attempt to decode the payload, because it does not know what rate/length values to use. In this case there are no decoded bits for the BER counters to track. A bad header does not mean zero BER. It means no bits at all. BER is not a sufficient metric for characterizing a packet-based system. BER should be considered along with PER.

Does we need to normalized our equalizer algorithm output to original output range of  Zreo-Forcing equalizer ?

Yes- in the OFDM Reference Design Rx PHY the decoder black box also implements the de-mapping from modulated symbols to soft bits. The rx_i/rx_q inputs to the decoder black box should be equalized symbols with (ideally) the same values as symbols from the transmitted constellation. If your equalizer introduces a constant scaling vs. the transmitted symbols, you should remove this scaling from the signals before the decoder. Alternatively you could modify the decoder Verilog to shift the decision boundaries to match your scaled constellation.