Forums

Single carrier with frequency domain equalization (SCFDE)

Started by philgo July 9, 2007
I have seen some paper demonstrating the benefit of SCFDE over OFDM,
basically low PAPR performance, comparable complexity, improved
performance especially when convolutional code rate is high. 

But why we see many standards still using OFDM, but not SCFDE?



philgo wrote:

> I have seen some paper demonstrating the benefit of SCFDE over OFDM, > basically low PAPR performance, comparable complexity, improved > performance especially when convolutional code rate is high. > > But why we see many standards still using OFDM, but not SCFDE? >
The one and only advantage of OFDM is the small computing burden. This allows for the simple and cheap receivers. With the single carrier, the bandwidth is determined by symbol rate, hence the symbol rate has to be high. Equalization has to be done for every symbol. Thus the computing burden is much higher. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
On Jul 9, 8:22 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> philgo wrote: > > I have seen some paper demonstrating the benefit of SCFDE over OFDM, > > basically low PAPR performance, comparable complexity, improved > > performance especially when convolutional code rate is high. > > > But why we see many standards still using OFDM, but not SCFDE? > > The one and only advantage of OFDM is the small computing burden. This > allows for the simple and cheap receivers. With the single carrier, the > bandwidth is determined by symbol rate, hence the symbol rate has to be > high. Equalization has to be done for every symbol. Thus the computing > burden is much higher. > > Vladimir Vassilevsky > > DSP and Mixed Signal Design Consultant > > http://www.abvolt.com
What about narrowband interference? Intuitively, it would seem that OFDM is better. John
Yes, OFDM has low receiver complexity. However, SC with freq. domain
equalization (SC-FDE) has similar complexity. Basically, OFDM has IFFT at
transmitter side and FFT at receiver side with linear equalizer. SC has
both IFFT and FFT and receiver side with also linear equalizer in the
freq. domain. 

In my mind, this SC-FDE structure is particularly useful for uplink, since
transmitter is free of high PAPR problem and receiver can afford the
IFFT/FFT complexity. But why we do not see SC-FDE in standard?



> > >philgo wrote: > >> I have seen some paper demonstrating the benefit of SCFDE over OFDM, >> basically low PAPR performance, comparable complexity, improved >> performance especially when convolutional code rate is high. >> >> But why we see many standards still using OFDM, but not SCFDE? >> > >The one and only advantage of OFDM is the small computing burden. This >allows for the simple and cheap receivers. With the single carrier, the >bandwidth is determined by symbol rate, hence the symbol rate has to be >high. Equalization has to be done for every symbol. Thus the computing >burden is much higher. > >Vladimir Vassilevsky > >DSP and Mixed Signal Design Consultant > >http://www.abvolt.com > > > > >
Aloha,

philgo wrote:

> Yes, OFDM has low receiver complexity. However, SC with freq. domain > equalization (SC-FDE) has similar complexity. Basically, OFDM has IFFT at > transmitter side and FFT at receiver side with linear equalizer. SC has > both IFFT and FFT and receiver side with also linear equalizer in the > freq. domain. > > In my mind, this SC-FDE structure is particularly useful for uplink, since > transmitter is free of high PAPR problem and receiver can afford the > IFFT/FFT complexity. But why we do not see SC-FDE in standard?
In 3gpp-LTE there will be (is, already in lab-systems) SC-FDE in uplink. To change modulation technique in fielded systems (standardized or proprietary) is a little bit complicated... Wishing a happy day, LOBI
On Tue, 10 Jul 2007 13:58:15 +0200, Andreas Lobinger
<newsreturns@biszumknie.de> wrote:

>Aloha, > >philgo wrote: > >> Yes, OFDM has low receiver complexity. However, SC with freq. domain >> equalization (SC-FDE) has similar complexity. Basically, OFDM has IFFT at >> transmitter side and FFT at receiver side with linear equalizer. SC has >> both IFFT and FFT and receiver side with also linear equalizer in the >> freq. domain. >> >> In my mind, this SC-FDE structure is particularly useful for uplink, since >> transmitter is free of high PAPR problem and receiver can afford the >> IFFT/FFT complexity. But why we do not see SC-FDE in standard? > >In 3gpp-LTE there will be (is, already in lab-systems) SC-FDE in uplink. >To change modulation technique in fielded systems (standardized or >proprietary) is a little bit complicated... > >Wishing a happy day, > LOBI
That's not quite the same thing. LTE adds an additional transform to what is already an OFDM system in order to get some of the low PAPR of an SC system. I thinkt the SC-OFDM being discussed is more like what was contained in the early 802.16 specs (yes, it's in there, IIRC in 802.16a). The idea was a native SC system with a training sequence built into the framing for the FDEQ. That spec had three different PHY options, though, and I think the SC-OFDM was taken the least seriously...i.e., got the least attention from anyone looking at actually building hardware. Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org

philgo wrote:

> Yes, OFDM has low receiver complexity. However, SC with freq. domain > equalization (SC-FDE) has similar complexity. Basically, OFDM has IFFT at > transmitter side and FFT at receiver side with linear equalizer. SC has > both IFFT and FFT and receiver side with also linear equalizer in the > freq. domain.
It is possible to equalize the incoming signal either in frequency or in the time domain. This will produce sequence of symbols in the case of SC and the one symbol in the case of OFDM. One OFDM symbol can be decoded directly. However the direct decoding of the equalized SC sequence seem to have lower performance compared to the direct decoding of the OFDM. The MLSE should be used; that complicates the decoder significantly. However I didn't the numbers and this can be a wrong assumption. VLV