Hi there, I have a question on OFDM. As we know, most OFDM systems nowadays do not use all the available tones (subcarriers) for data modulation. Let us say we have 64 tones in total and we might end up with 48 data tones, 4 pilot tones, 3 DC tones and 9 null tones, which is kind of an typical setup. My question is regarding the null tones, which are usually placed on the two boundaries of the total 64 subcarriers. My understanding of using null tones is that, we can ease the filter implementation complexity/cost in the following DAC. I heard that doing so, we may skip oversampling (or an oversampling factor of 1). Please correct me if I am wrong. But, can I choose not to force those boundary tones to be null tones and instead transmit data on those tones? My current understanding is that, I need to do oversampling (say factor 2) in order to make the filter implementation affordable. My ultimate question is, if I am willing to do oversampling instead of forcing null tones on the boundaries, is it a workable OFDM modulation? Do I need to do anything else if I do not use any null tones? Please help. Thanks
Do we have to use null tones in OFDM modulation?
Started by ●September 9, 2008
Reply by ●September 9, 20082008-09-09
Null tones provide the guard band. If you place the data into the guard band, then you will spill it into the adjacent channel (and/or pick the spillage from the adjacent channel). VLV phil wrote:> Hi there, > > I have a question on OFDM. As we know, most OFDM systems nowadays do > not use all the available tones (subcarriers) for data modulation. > Let us say we have 64 tones in total and we might end up with 48 data > tones, 4 pilot tones, 3 DC tones and 9 null tones, which is kind of an > typical setup. > > My question is regarding the null tones, which are usually placed on > the two boundaries of the total 64 subcarriers. My understanding of > using null tones is that, we can ease the filter implementation > complexity/cost in the following DAC. I heard that doing so, we may > skip oversampling (or an oversampling factor of 1). Please correct me > if I am wrong. > > But, can I choose not to force those boundary tones to be null tones > and instead transmit data on those tones? My current understanding is > that, I need to do oversampling (say factor 2) in order to make the > filter implementation affordable. > > My ultimate question is, if I am willing to do oversampling instead of > forcing null tones on the boundaries, is it a workable OFDM > modulation? Do I need to do anything else if I do not use any null > tones? > > Please help. Thanks
Reply by ●September 10, 20082008-09-10
On Sep 10, 2:22�am, phil <phi...@gmail.com> wrote:> Hi there, > > My question is regarding the null tones, which are usually placed on > the two boundaries of the total 64 subcarriers. My understanding of > using null tones is that, we can ease the filter implementation > complexity/cost in the following DAC. I heard that doing so, we may > skip oversampling (or an oversampling factor of 1). Please correct me > if I am wrong.[kp] Well, in typical designs even though there are null tones, there will be an oversampling factor of 2 (or even 3). The objective of oversampling is to remove the aliases created outside the desired band.> But, can I choose not to force those boundary tones to be null tones > and instead transmit data on those tones? My current understanding is > that, I need to do oversampling (say factor 2) in order to make the > filter implementation affordable. > > My ultimate question is, if I am willing to do oversampling instead of > forcing null tones on the boundaries, is it a workable OFDM > modulation? Do I need to do anything else if I do not use any null > tones? >[kp] It might create problems when the available spectrum is divided in multiple channels. The right most subcarrier of your channel might interfere with left most subcarrier from a similar transmission happening on the adjacent band.> Please help. ThanksHope this helps. Krishna ~blogs @ http://www.dsplog.com
Reply by ●September 11, 20082008-09-11
Thanks for the replies. Yes, I agree. We need to be protective from adjacent bands. But what if the following? Suppose each band is 20MHz but the total OFDM signal takes a bandwidth (counting the null tones and all others) of only 16MHz, while the 5MHz on the boundaries (2MHz each side) are kept unused for interference avoidance between adjacent bands. I think we can use those null tones inside the OFDM modulation. The bandwidth efficiency would be 16/20, around 80%. Of course, the numbers here are only for illustration purpose and the actual efficiency depends on actual implementation and design. Do you think there is any serious problem with the above? Please share your criticism/comment. Thanks. On Sep 9, 10:42�pm, Krishna <krishna.pil...@gmail.com> wrote:> On Sep 10, 2:22�am, phil <phi...@gmail.com> wrote:> Hi there, > > > My question is regarding the null tones, which are usually placed on > > the two boundaries of the total 64 subcarriers. My understanding of > > using null tones is that, we can ease the filter implementation > > complexity/cost in the following DAC. I heard that doing so, we may > > skip oversampling (or an oversampling factor of 1). Please correct me > > if I am wrong. > > [kp] Well, in typical designs even though there are null tones, there > will be an oversampling factor of 2 (or even 3). The objective of > oversampling is to remove the aliases created outside the desired > band. > > > But, can I choose not to force those boundary tones to be null tones > > and instead transmit data on those tones? My current understanding is > > that, I need to do oversampling (say factor 2) in order to make the > > filter implementation affordable. > > > My ultimate question is, if I am willing to do oversampling instead of > > forcing null tones on the boundaries, is it a workable OFDM > > modulation? Do I need to do anything else if I do not use any null > > tones? > > [kp] It might create problems when the available spectrum is divided > in multiple channels. > The right most subcarrier of your channel might interfere with left > most subcarrier from a similar > transmission happening on the adjacent band. > > > Please help. Thanks > > Hope this helps. > Krishna > ~blogs @http://www.dsplog.com
Reply by ●September 16, 20082008-09-16
On Sep 10, 8:37�pm, phil <phi...@gmail.com> wrote:> Thanks for the replies. Yes, I agree. We need to be protective from > adjacent bands. > > But what if the following? Suppose each band is 20MHz but the total > OFDM signal takes a bandwidth (counting the null tones and all others) > of only 16MHz, while the 5MHz on the boundaries (2MHz each side) are > kept unused for interference avoidance between adjacent bands. I think > we can use those null tones inside the OFDM modulation. > > The bandwidth efficiency would be 16/20, around 80%. Of course, the > numbers here are only for illustration purpose and the actual > efficiency depends on actual implementation and design. > > Do you think there is any serious problem with the above? Please share > your criticism/comment. Thanks. > > On Sep 9, 10:42�pm, Krishna <krishna.pil...@gmail.com> wrote: > > > On Sep 10, 2:22�am, phil <phi...@gmail.com> wrote:> Hi there, > > > > My question is regarding the null tones, which are usually placed on > > > the two boundaries of the total 64 subcarriers. My understanding of > > > using null tones is that, we can ease the filter implementation > > > complexity/cost in the following DAC. I heard that doing so, we may > > > skip oversampling (or an oversampling factor of 1). Please correct me > > > if I am wrong.can anybody please help to comment here? thanks> > > [kp] Well, in typical designs even though there are null tones, there > > will be an oversampling factor of 2 (or even 3). The objective of > > oversampling is to remove the aliases created outside the desired > > band. > > > > But, can I choose not to force those boundary tones to be null tones > > > and instead transmit data on those tones? My current understanding is > > > that, I need to do oversampling (say factor 2) in order to make the > > > filter implementation affordable. > > > > My ultimate question is, if I am willing to do oversampling instead of > > > forcing null tones on the boundaries, is it a workable OFDM > > > modulation? Do I need to do anything else if I do not use any null > > > tones? > > > [kp] It might create problems when the available spectrum is divided > > in multiple channels. > > The right most subcarrier of your channel might interfere with left > > most subcarrier from a similar > > transmission happening on the adjacent band. > > > > Please help. Thanks > > > Hope this helps. > > Krishna > > ~blogs @http://www.dsplog.com