WARP Project Forums - Wireless Open-Access Research Platform

isky wrote:

1.  The first one is about the serial number problem of subcarrier use.
In the example, some unused subcarriers are placed in the middle, such as [1 28:38]. So the position of the data subcarrier is [2:7 9:21 23:27 39:43 45:57 59:64];
In the 802.11 protocol, some unused subcarriers are placed on both sides [1:6 60:64], and the location of the data subcarriers is [7:12 14:26 28:32 34:38 40:52 54 :59];
I tried to modify the serial number used by the subcarriers in the example, but the program will often report an error, 'No LTS Correlation Peaks Found!' And occasionally I will receive the data, so I am confused about the serial number problem of using subcarriers.

I think your confusion is coming from whether or not you consider the 1st subcarrier DC or the maximum negative frequency. The WARPLab example specifies the indices such that DC is the first index and the largest frequency indices are in the middle of the vector. This is the typical order you deal with when using FFTs since negative frequencies are really just larger positive frequencies that have aliased around the Nyquist frequency. See MATLAB's fftshift for details. Furthermore, MATLAB indexes the first element of an array with 1, not zero, which makes it even more confusing.

So, when the code says:

SC_IND_DATA = [2:7 9:21 23:27 39:43 45:57 59:64];

That's equivalent to:

[-26:-22 -20:-8 -6:-1 1:6 8:20 22:26]

if you want to think in terms of negative subcarriers. This is the standard 802.11a/g subcarrier layout, so you shouldn't need to change anything. The LTF (which is composed of LTS), doesn't depend on the above subcarrier mapping, so changing it shouldn't affect your ability to find LTS correlation peaks during packet detection and FFT timing alignment.

isky wrote:

2.  In OFDM design, does the DC subcarrier interfere with the surrounding subcarriers? I collected some data and found that some subcarriers on both sides of the DC subcarrier were subject to some interference.

You'll have to clarify what you mean by "subject to some interference." Actual over-the-air external interference can be wideband and span more than one subcarrier. The presence of an interferer near DC could easily affect many subcarriers.

isky wrote:

3.  In OFDM design, some subcarriers are not used and are idle. For example, [1 28:38] in the example, can we use these subcarriers to send data? What kind of impact will it bring us?

Not really. Those are disabled because they can't be carried well or at-all with a radio. "1" is DC, which is definitely not going to work. "28:38" are the high frequency subcarriers. You start to run into the low pass filters in the radio with these subcarriers. 802.11a/g disables them to abide by spectral mask requirements. Notably, you can use at least some of these subcarriers for transmitting data. That's one of the things 802.11n changed. That standard uses an extra 4 data-bearing subcarriers as compared to 802.11a/g.

For the third question, I did not make it clear.  I would like to ask, can I use the guard subcarriers to transmit some data? Because the textbook says, the 802.11a channel contains 64 subcarriers, of which there are 12 guard subcarriers, 4 pilot subcarriers, and 48 data subcarriers. In my guess, I think it works, but it may bring some unpredictable and bad problems. What do you think about this plan?

Have you tried implementing this and measuring performance? The WARPLab OFDM example scripts make it easy to customize the waveform and evaluate it in simulation (*always* test in simulation first) and hardware.

Moreover, I still have a question. Can I use warp to measure the actual SNR of the channel? I have not found a related example. What can you recommend? The signal-to-noise ratio of the channel plays an important role in analyzing other parameters such as bit error rate, and I have been confused.

Our example scripts do not estimate SNR. You would need to implement this yourself. There are many related discussions on these forums about estimating SNR based on Rx amplitude, Rx gains and noise power.