I have a few general questions pertaining to OFDM ...

Let us assume an 1048 point IFFT and number of number of sub-carriers as 600.

1)  in this case, does it matter which sub-carriers in the IFFT are modulated and which are not?

for example, since 586 sub-carriers are not occupied, can these be any sub-carriers in the N point?

2) for the sub-carriers that are unused in the IFFT, are 0's (zeros) passed for all those sub-carriers?

3) for case 2 above, what would happen if data from an used sub-carrier is also passed to un-used sub-carrier?. Will this result in the overall bandwidth of the OFDM signal?

The subcarriers near the edges of the band are not used in order to provide "guard band" to facilitate the rolloff of channel selection filters and to prevent interference with adjacent bands.

Populating or not populating subcarriers within the occupied part of the bandwidth can be optional, and there are cases where you may not want to populate a portion of the spectrum due to either multiple access schemes or because you've detected an interferer in that part of the channel, or whatever other reason might make sense.  Sometimes you may want to leave a portion of the band unpopulated momentarily in order to assess noise level or whether there is interference energy there.

Not populating some subcarriers also provides some power concentration to the remaining subcarriers, which can also be useful.

To your third point, duplicating subcarriers in other subcarrier positions would be a form of repetition coding, which isn't very efficient, but does allow things like diversity combining to be used in the receiver for the duplicated subcarriers.   It would, however, probably be more efficient in most cases to just use a lower FEC code rate and populate all of the subcarrier positions.

I hope that helps a little bit.

The parameters you defined sound exactly like the 10 MHz LTE channel commonly used for cellular telephony.

1) It does matter which subcarriers are used.  So, for LTE, there are 300 on each side of the DC band.

2) Yes. The unused subcarriers are represented by 0s for the real and imaginary part of the data in the IFFT.

3) If some of the data from used sub-carriers is also placed into all the un-used bands, the bandwidth will be expanded to the full bandwidth of the 1024 IFFT.  This would mean that for LTE, the channel requirements would be violated.

For LTE, the widest BW is 20 MHz, and the middle 1200 (600 each side of DC) of the 2048 subcarriers are occupied.  Since each subcarrier is 15 kHz wide, this represents an active bandwidth of 18.015 MHz, with an empty subcarrier in the middle that would be the DC band at baseband.  See http://niviuk.free.fr/lte_resource_grid.html

So, for the 1024 point IFFT, the equivalent would be an active BW of 9.015 MHz in a 10 MHz band.  If all the subcarriers are occupied, then the BW would be 15.36 MHz, or significantly wider than the designated BW of the channel.

One important thing to know is that there are actually additional subcarriers at the edges of the 10 and 20 MHz bands that are sometimes used for Narrow Band IoT (NB-IoT) channels for low bit-rate data.

In LTE, the sub-carriers are aggregated into sets of 12 known as resource blocks.  In the 20 MHz channel there are 100 PRBs for regular data and telephony.  However, there are 5 additional resource blocks at each end of the channel that can be accessed for things like NB-IoT.

Thanks a lot. A few follow-up questions ...

On question 1, so the LTE standard defines exactly which sub-carriers to be used then?

On the NB-IoT, in case of a 2048 pt FFT, which sub-carriers would carry NB-IoT data?

From the IFFT processing point of view, I believe there is no real frequency we are referring to here. I mean, in a 1024 pt FFT, the N point could mean sub-carrier spacing of 15 KHz or 30 KHz. It is just not matter at the IFFT stage. Is this correct?

Probably a very silly question. If my actual data is somewhat dominated by 0s then does that mean I am actually occupying less than available bandwidth?

As suggested by SlartibartFast, the usage of the subcarriers and symbols within a frame are a shared resource between the user equipment (UE) and the base station downlink (DL) logical channels.  This means that often, there are subcarriers that are un-occupied for various periods of time.

Because the uplink (UL) from the UEs to the basestation, is a shared resource, and because the signals need to arrive at the basestation in a coherent fashion, there are issue to account for in terms of different levels and different delays from each UE, so some of what we are talking about get fairly complex.

Some resource blocks are dedicated to the control of the whole process and UEs are told which subcarriers and symbols (resource elements) to use, both for UL and DL traffic.

I don't have in my head the details of the NB-IoT allocations, but there are at least three.

1) Use of in-band subcarriers for data
2) Use of guard-band subcarriers
3) Standalone very narrow band (15kHz) channels that are dedicated for the NB-IoT data traffic.

Section 10 of the 3GPP TS 36.211 standard cover the NB-IoT details.  Good luck with deciphering the standard at first read.  It takes some serious commitment to wade through it.

Other sections of the same standard discuss the very fine details of all the rest of the discussion.

https://www.etsi.org/deliver/etsi_ts/136200_136299/136211/14.02.00_60/ts_136211v140200p.pdf

In the context of the width of each subcarrier, there is an absolute definition.  For the 2048 IFFT case, that sampling rate is defined to be 30.72 MHz, which in turn defines the exact width of each subcarrier.

And your last question is not silly at all.  However, remember that bandwidth is usually defined as the range of frequency over which the average signals spread.  In the case of this concept of shared resources, there may be times when some subcarriers at the edges of the band may or may not be occupied.  So, for some (very) short periods of time the bandwidth may narrow a little, but that would not be in the spirit of the definition of bandwidth.  Equally possible is the case when the outer subcarriers are occupied, but the inner ones are not, so the spectrum can change dramatically and very dynamically depending on the loading of the user space.