Hello, I am wondering there is so much literature about carrier and timing recovery in communications*. However as far as I understand, OFDM uses a different approach in synchronization. My question is since OFDM is becoming more and more popular does this mean that these algorithms will cease to be used in future? * synchronization in software radios (chris dick, fred harris, michael rice) * high speed all digital symbol timing recovery based on FPGA (Zhang Jian, Wu Nan) * Interpolation in digital modems - Part 1 (F.M Gardner) * Interpolation in digital modems - Part 2 (F.M Gardner) * A BPSK/QPSK timing error detector for sampled receivers (F.M Gardner) the list goes on...
synchronization in OFDM and others...
Started by ●November 27, 2007
Reply by ●November 27, 20072007-11-27
On Nov 27, 3:47 am, "rifo" <rtur...@yahoo.de> wrote:> Hello, > > I am wondering there is so much literature about carrier and timing > recovery in communications*. However as far as I understand, OFDM uses > a different approach in synchronization. My question is since OFDM is > becoming more and more popular does this mean that these algorithms will > cease to be used in future? > > * synchronization in software radios (chris dick, fred harris, michael > rice) > > * high speed all digital symbol timing recovery based on FPGA (Zhang Jian, > Wu Nan) > > * Interpolation in digital modems - Part 1 (F.M Gardner) > * Interpolation in digital modems - Part 2 (F.M Gardner) > * A BPSK/QPSK timing error detector for sampled receivers (F.M Gardner) > > the list goes on...This supposes two things: 1) That OFDM techniques will completely supplant single-carrier communications methods at some point. 2) That the sync methods used prior to OFDM have no application in OFDM. I don't think either of these are valid. There are situations where single-carrier methods are better suited than OFDM, so SC will continue to be used. Interpolation is still essential in OFDM synchronization, even if the timing error detectors don't look the same. EB
Reply by ●November 28, 20072007-11-28
Hello, thanks for the answer. I would be happy if you can tell me these fields where a single carrier is more advantageous than OFDM. Also I have came across a paper, namely " Design and Implementation of ALL Digital I-Q Modulator and Demodulator for High Speed WLAN in FPGA" in this paper in both modulator and demodulator blocks, there is the I and Q pulse shaping and Quadrature de/modulator parts. So at this point I am a little bit confused. If I say that, "AFter the IFFT block in the transmitter, pulse shaping and Quadrature modulator blocks take place and therefore, synchronization blocks that I have told in the first post can be used" Will I be wrong?
Reply by ●November 28, 20072007-11-28
Okey I think I am wrong, since all these techniques assume a constant constellation scheme and build on the algorithm based on this assumption however OFDM uses many constellation schemes so this will not work. However I would still be happy to hear about fields where single carrier is better than OFDM approach.> Hello, thanks for the answer. >I would be happy if you can tell me these fields where a single carrier >is more advantageous than OFDM. >Also I have came across a paper, namely " Design and Implementation ofALL>Digital I-Q Modulator and Demodulator for High Speed WLAN in FPGA" > > in this paper in both modulator and demodulator blocks, there is the I >and Q pulse shaping and Quadrature de/modulator parts. So at this pointI>am a little bit confused. > If I say that, "AFter the IFFT block in the transmitter, pulse shaping >and Quadrature modulator blocks take place and therefore,synchronization>blocks that I have told in the first post can be used" Will I bewrong?> > >
Reply by ●November 28, 20072007-11-28
>>these fields where a single carrier is more advantageous than OFDMIt may be preferrable in the uplink from a mobile device, because the signal is easier to amplify with a power amplifier. The amplitude probability density is more "compact" towards highest amplitudes. There is a general trend: The more efficiently I use the spectrum, the more Gaussian the amplitude density becomes (which is difficult from the power amplifier point-of-view). For example, LTE (3G cellular evolution) will use SC-FDMA for that reason. -mn
Reply by ●November 28, 20072007-11-28
BTW, careful with the term "pulse shaping filter" in systems that involve frequency domain processing: In "conventional" time domain systems, we used to get the symbol stream - discrete samples, periodic over all frequencies by definition- and cut out a piece of spectrum for transmission. Since the required filter can't be ideally steep, it is instead designed as symmetric around the symbol rate according to Nyquist's pulse shaping rule. This avoids intersymbol interference, but the actual transmitted spectrum exceeds the symbol rate. For example, in WCDMA the design target for this factor is 1.22 (3.84 Mchips/second, but 4.096 MHz before the RRC filter closes completely) In OFDM-like systems, our IFFT in the transmitter is a perfect filter, as long as I observe a full cycle length (the observation interval may move within the cyclic prefix). Therefore, the "pulse shaping filter" in my book is an ideal lowpass filter. If I take one symbol and take the spectrum via FFT, there really isn't any signal energy in band-edge FFT bins that were empty during transmission, no big surprise. BUT there is a discontinuity between adjacent symbols that is not affected by the "ideal" frequency domain filter. I need filtering after the IFFT to keep the spectral emissions from the "step" between symbols under control. That's why I need the filter, but it can be misleading to think of it in terms of "pulse shaping filter", as used in time-domain systems. Strictly speaking, the filter also needs to suppress spectral replicas. But that is a different situations than in TD systems, because the spectrum outside from the IFFT has already some "empty" frequency range, which the lowpass filter may use as transition region. This is not to claim that one or the other word is correct, But it may be a good idea to think through what is actually the purpose of the filter. I have noticed that the different role occasionally causes quite a bit of confusion. Cheers Markus
Reply by ●November 28, 20072007-11-28
On Nov 28, 3:56 am, "mnentwig" <mnent...@elisanet.fi> wrote:> >>these fields where a single carrier is more advantageous than OFDM > > It may be preferrable in the uplink from a mobile device, because the > signal is easier to amplify with a power amplifier. The amplitude > probability density is more "compact" towards highest amplitudes. > > There is a general trend: The more efficiently I use the spectrum, the > more Gaussian the amplitude density becomes (which is difficult from the > power amplifier point-of-view). > > For example, LTE (3G cellular evolution) will use SC-FDMA for that reason. > > -mnLarge PAPR caused by high-efficiency spectral usage is certainly an issue that can skew the balance toward SC solutions, although there are ways to mitigate that which require additional processing. Essentially what you're looking at are engineering trade-offs that account for suitability to the purpose at hand. OFDM is a good way to deal with complex multipath fading environments, but if your application doesn't have that problem then you may save power or complexity by using a single-carrier approach. EB
Reply by ●November 29, 20072007-11-29
i remember..similar type of discussion went in groups long ago.. plz check at this link http://groups.google.com/group/comp.dsp/browse_thread/thread/ceac6381c6fe422e/35aef71bddbe4b5d?hl=en&lnk=gst&q=particlereddy#35aef71bddbe4b5d and also..this paper may help you further.. http://www.i3s.unice.fr/~deneire/pap/tubbax01vtc.pdf thanks ravi chandra
Reply by ●November 29, 20072007-11-29
>>>these fields where a single carrier is more advantageous than OFDM > >It may be preferrable in the uplink from a mobile device, because the >signal is easier to amplify with a power amplifier. The amplitude >probability density is more "compact" towards highest amplitudes. > >There is a general trend: The more efficiently I use the spectrum, the >more Gaussian the amplitude density becomes (which is difficult from the >power amplifier point-of-view). > >For example, LTE (3G cellular evolution) will use SC-FDMA for thatreason.> > >-mn >Hello Markus, Thanks for your response. But there is something that is not clear to me. I have read briefly about this SC-FDMA and it seems that there is still subcarrier mapping. http://hgmyung.googlepages.com/SCFDMA.pdf (slide 7) Do you think that these synchronization methods can be applied? Most probably there is some detail that I am missing. I would be happy if you can guide me. Thanks Rifat Do you think