WARP Project Forums - Wireless Open-Access Research Platform

chunter wrote:

We do not use the greenfield format. The PHY will use the NONHT and HTMF preamble structures depending what PHY mode you are using. How are you calculating 70µs? What is that a measurement of exactly? The last sample of the data frame to the first sample of the ACK frame?

I do the experiment based on the dual-node log capture example in 802.11 reference design experiment framework. I changed the code to just let start a fully-backlogged locally generated traffic(LTG) flow from AP->STA. In this experiment, HTMF preamble structure is used. When I analysed the received data packets in STA from AP and ACK packets in AP. I found the time between the Data packet and corresponding ACK is 70us from the results of the experiment. I calculated the time difference between the data packets and the corresponding ACK packets through the following formula.

In the formula, lmac is the number of bytes of MAC packet, Nsubc is the number of data subcarriers, 52 in HTMF waveform (11n), Nbpsc is the number of bits per subcarrier.

The number of OFDM symbols of the entire physical layer packet is calculated based on the Non-HT PPDU.  The number of bytes of the MAC packet is 48 in which I set the number bytes of MAC payload as 20. MCS is chosen as 3 which means the modulation rate is 16-QAM(4 bits per subcarrier), the code rate is 1/2 and data rate is 26Mbps. However, after the calculation, the time duration of the data packet is 36us and the SIFS is 16us. The calculation result cannot match the experiment result.

Is there anything wrong to calculate the time difference of data packet and corresponding ACK packet?

murphpo wrote:
The formula you posted describes the duration of a NONHT waveform. The "5" term represents the preamble (STF(2) + LTF(2)) and SIGNAL field (1). A SISO HTMF waveform has 4 additional OFDM symbols (HTSIG(2) + HTSTF(1) + HTLTF(1)). Our calc_tx_time() Python method implements this calculation for both waveforms. Note that calc_tx_time() includes the 6 usec extension Chris described above.

I add the 4 additional OFDM symbols to the formula and calculate again, then I changed the payload length to test whether it matches other conditions. The calculation results and experiment result is shown in the following picture. However, according to the results, the experiment results usually 2 to 6 higher than the calculation for each length of the payloads. Why did this happen?  Is that because of the instrument accuracy?