On May 15, 1:22 am, donnacha.d...@gmail.com wrote:
> I have a couple of questions if you don't mind. One of the
> advantages of multicarrier (at least for wireline channels) is
> the ability to notch out tones which have high interference, or
> which are forbidden for use (not so common in wireless because
> the channel changes so fast). I presume that CE-OFDM is
> no-longer "notchable" am I right?
For wireline multicarrier (like DSL), the channel is really
stable and a variety of things may be done to optimize
transmission, including bit/power loading, and potentially
notching out interference. For wireless communications, there
are many interference mitigation techniques out there. From what
I've seen, beamforming technique are more powerful than notch
filtering ones.
I would say that the primary reason people use OFDM in wireless
is to avoid conventional equalization techniques. Some exciting
developments that have sprouted out of OFDM include single carrier
with frequency-domain equalization (SC-FDE).
The motivating factors behind CE-OFDM was obviously battery-life
improvement for a wireless handset using OFDM. CE-OFDM certainly
accomplishes this and the complexity is low. Depending on the
modulation index, there may or may not be excess bandwidth.
There is a threshold effect with CE-OFDM and reducing it is an
area of active research. Depending on the target Eb/N0, however,
the threshold effect may be a mute point.
> Also, from your work, there seems to be excess bandwidth with
> CE-OFDM compared to regular OFDM (I'm looking at your PSD
> results). Do you have a figure for this excess bandwidth?
Nice Work Steve,
I have a couple of questions if you don't mind. One of the advantages
of multicarrier (at least for wireline channels) is the ability to
notch out tones which have high interference, or which are forbidden
for use (not so common in wireless because the channel changes so
fast). I presume that CE-OFDM is no-longer "notchable" am I right?
Also, from your work, there seems to be excess bandwidth with CE-OFDM
compared to regular OFDM (I'm looking at your PSD results). Do you
have a figure for this excess bandwidth? Given that you have removed
the ability to notch, and that you have excess bandwidth, all in the
name of constant envelope, I would suggest that a filterbank precoder
is a much better solution e.g. :
New precoding for intersymbol interference cancellation using
nonmaximally decimated multirate filterbanks with ideal FIR equalizers
Xiang-Gen Xia
IEEE Transactions on Signal Processing
Publication Date: Oct. 1997
Volume: 45 , Issue: 10
On page(s): 2431 - 2441
I dont know a lot about filterbank precoders, but they seem like a
much more controllable mechanism for constant envelope modulation with
multipath resistance.
cheers
DD
http://www.xerenet.com
Reply by ●May 15, 20072007-05-15
On May 14, 6:50 pm, Steve Thompson wrote:
> Many of the issues discussed in this discussion have been
> addressed in my PhD thesis:
>
> http://76.222.14.118/cv.pdf
Dear Group,
Apologies in advance for the coming shameless plugs, but how can
I resist?
Many of the issues discussed in this discussion have been
addressed in my PhD thesis:
http://76.222.14.118/cv.pdf
CE-OFDM is a fun idea since it is an easy way to get a wideband
constant envelope signal with easy equalization. The
equalization hinges on the cyclic prefix, which guarantees the
theorem of circular convolution. This way, linear equalization
may be performed with FFTs. Synchronization and channel
estimation are performed just as they are in OFDM.
The shameless plugging continues:
http://76.222.14.118/pres/o/o18.pdf
That presentations shows some pictures of our hardware prototype.
Regards,
Steve Thompson
http://76.222.14.118/work/
PS: Why the cryptic IP address above? Because I just moved, and
my server moved with me. As the DNS servers remap
elsteve.com to my new IP address, the 76.222.14.118 must be
used. After that, the above links should be equivalent to
http://elsteve.com/cv.pdf,http://elsteve/pres/o/o18.pdf, and
http://elsteve.com/work/ respectively.
Reply by ●May 14, 20072007-05-14
On May 14, 9:53 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> sampson...@gmail.com wrote:
>
> > CE-OFDM is OFDM with a Phase Modulator on the output. Google finds
> > some references.
>
> OFDM riding on FM is ridiculous; it is a way to combine the
> disadvantages of both modulations. This is the low energy and spectral
> efficiency, nonlinear effects, a complex equalization in the time domain
> and a sensitivity to selective fading. Why anybody needs this?
>
> Vladimir Vassilevsky
>
> DSP and Mixed Signal Design Consultant
>
> http://www.abvolt.com
I believe that CE-OFDM was originally proposed as a way to modify OFDM
such that an efficient nonlinear transmitter could be used, hopefully
preserving the benefits of OFDM in a multipath channel. The nonlinear
transmitter will distort the OFDM waveform but not the CE-OFDM one. I
think the conclusion is that with a more complicated equalizer, CE-
OFDM can do well compared to OFDM.
A wider channel must also be required for CE-OFDM because of the
frequency spreading in the modulator. The phase modulation also
introduces a threshold effect in the receiver that isn't an issue for
OFDM.
John
John
Reply by Steve Underwood●May 14, 20072007-05-14
Vladimir Vassilevsky wrote:
>
>
> julius wrote:
>
>>
>>> OFDM riding on FM is ridiculous; it is a way to combine the
>>> disadvantages of both modulations. This is the low energy and spectral
>>> efficiency, nonlinear effects, a complex equalization in the time domain
>>> and a sensitivity to selective fading. Why anybody needs this?
>>>
>>
>> The thought of coming up with a synchronizer for this gives me
>> shivers, and it's only Monday morning.
>>
>
> They don't synchronize the FM demodulator. The whole thing operates like
> the OFDM modem working via the analog FM channel.
>
>> To be fair, John Proakis knows heck of a lot about synchronization and
>> he builds impressive systems for all sorts of challenging
>> environments,
>> so I am going to wait until he builds a receiver system for this.
>
> Well, I'd say the synchronization is one of the weakest spots of the
> book of Proakis on the Digital Communication. Mostly he cites Viterbi,
> Van Trees and other classics; there is no treatment on the acquisition
> of the sync, only the performance when the sync is in lock already, no
> description of the algorithms, etc.
I'd say that Proakis book simply skips the whole subject of
synchronisation, with a few words and a little hand waving. A very poor
show from a smart guy.
Steve
Reply by Vladimir Vassilevsky●May 14, 20072007-05-14
julius wrote:
>
>>OFDM riding on FM is ridiculous; it is a way to combine the
>>disadvantages of both modulations. This is the low energy and spectral
>>efficiency, nonlinear effects, a complex equalization in the time domain
>>and a sensitivity to selective fading. Why anybody needs this?
>>
>
> The thought of coming up with a synchronizer for this gives me
> shivers, and it's only Monday morning.
>
They don't synchronize the FM demodulator. The whole thing operates like
the OFDM modem working via the analog FM channel.
> To be fair, John Proakis knows heck of a lot about synchronization and
> he builds impressive systems for all sorts of challenging
> environments,
> so I am going to wait until he builds a receiver system for this.
Well, I'd say the synchronization is one of the weakest spots of the
book of Proakis on the Digital Communication. Mostly he cites Viterbi,
Van Trees and other classics; there is no treatment on the acquisition
of the sync, only the performance when the sync is in lock already, no
description of the algorithms, etc.
Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com
Reply by julius●May 14, 20072007-05-14
On May 14, 8:53 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> OFDM riding on FM is ridiculous; it is a way to combine the
> disadvantages of both modulations. This is the low energy and spectral
> efficiency, nonlinear effects, a complex equalization in the time domain
> and a sensitivity to selective fading. Why anybody needs this?
>
> Vladimir Vassilevsky
>
> DSP and Mixed Signal Design Consultant
>
> http://www.abvolt.com
The thought of coming up with a synchronizer for this gives me
shivers,
and it's only Monday morning.
To be fair, John Proakis knows heck of a lot about synchronization and
he builds impressive systems for all sorts of challenging
environments,
so I am going to wait until he builds a receiver system for this.
Julius
Reply by Vladimir Vassilevsky●May 14, 20072007-05-14
sampson164@gmail.com wrote:
>
> CE-OFDM is OFDM with a Phase Modulator on the output. Google finds
> some references.
OFDM riding on FM is ridiculous; it is a way to combine the
disadvantages of both modulations. This is the low energy and spectral
efficiency, nonlinear effects, a complex equalization in the time domain
and a sensitivity to selective fading. Why anybody needs this?
Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com
Reply by dvo●May 14, 20072007-05-14
>On May 13, 10:12 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
>wrote:
>> julius wrote:
>> >>Not quite :-) Imagine an OFDM symbol as a member of the class of
n-FSK
>> >>sequences designed for demodulation in the frequency domain...
>>
>> > Fine, I suppose one can apply PPM-type coding, except
>> > that it's in the frequency domain :-).
>>
>> The engineer's approach will be take the standard OFDM and clip it so
>> the envelope will have to be a constant :-)
>>
>> VLV
>
>CE-OFDM is OFDM with a Phase Modulator on the output. Google finds
>some references.
>
>John
>
>