I'm trying to analyze a over the air RF capture. I have a RAW IQ data samples but the sample rate is 30.72Msps. The bandwidth occupied by the signal is about 10MHz wide or 15.56MHz with guard carriers. Since the sample rate is just under Nyquist how is this going to affect me? Obviously the bandwidth of each carrier is much less than the nyquist frequency given the sample rate. The data was presented to me but I could acquire more data at a higher sample rate if needed. The goal would be to analyze and decode the OFDM waveform and data.
for LTE ofdm at 30.72 Msps the bandwidth is (1200*15KHz = 18MHz => 9MHz on either side of centre). That implies guard band of 1Mhz (from 20 MHz nominal bandwidth). Can you elaborate on your figures.
This is not an LTE application, but I'm also not sure if they are using an LTE chipset. Looks like the bins are 15KHz. I really don't know much other than what is written here: https://github.com/proto17/dji_droneid also I'm pretty ignorant when it comes to OFDM.
The earlier DJI drones used proprietary modulation, but the newer ones use an LTE chipset.
There is a company that you may want to look at, that is doing what you are interested in.
Nevertheless, all the words they say are from LTE and OFDM. Zadoff-Chu sequences are used for several purposes in LTE, particularly in random access preambles, frequency offset estimation and compensation, phase derotation, equalization, symbol decision, descrambling - this all is done in LTE. Moreover, Turbocodes are specifically used in LTE shared channel. And note people also clearly tell about OFDM. So whatever is end application is there, I suspect they use one or another LTE chipset with all consequences.
As to fractional bandwidth in OFDM, say 10 MHz of used b/w with respect to 20 MHz of maximum possible in that design, it could be achieved as with slower sampling rate, say 15.36 instead of 30.72, so by leaving those extra carriers as zeros.
Keep in mind, that beast is capturing complex signal, so Nyquist is not that far.
If your RF (air) signal is 10MHz and has further guard band carriers to 15.56MHz then your signal is 15.56MHz and so you need to sample it more than 30.72
Moreover, if it is centred on 7.5MHz as your link suggests then 7.5MHz + halfband = 7.5 + 15.56/2 on either side of centre. I have doubts about these figures.
If the RF Bandwidth is 10 MHz there is no need to sample at or more than 30.72 MHz assuming complex baseband samples —sampling at twice the FFT sampling rate (exactly 30.72 MHz) is convenient but not necessary to determine carrier and time offset and demodulate the 15.56 MHz FFT subcarriers. What sampling is required above the necessary 20 MHz depends on the anti-alias filters used ahead of the ADC.
When you say you have doubts about the figures, which ones specifically? Any examples and specific concerns?