In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at the receiver is very wide even though the datarate may be relatively small. Therefore in order for demodulation and detection to be performed in the digital domain, very fast ADCs and front-end DSP must be utilized. In some modern IR-UWB systems, this may not be feasible due to the extreme bandwidth (>1 GHz) of the signals. Since the underlying data rate is relatively low, much of this bandwidth is "redundant" in some sense. My question is this: do techniques exist in which some of the initial demodulation is performed in the analog domain, the output of which is a much smaller bandwidth signal that can then be more tractably (and economically!) converted and processed in the digital domain? I'd love to hear ideas on this. -- % Randy Yates % "So now it's getting late, %% Fuquay-Varina, NC % and those who hesitate %%% 919-577-9882 % got no one..." %%%% <yates@ieee.org> % 'Waterfall', *Face The Music*, ELO http://www.digitalsignallabs.com
DSP of Wideband Signals
Started by ●April 4, 2008
Reply by ●April 4, 20082008-04-04
On Apr 4, 8:56 am, Randy Yates <ya...@ieee.org> wrote:> In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at > the receiver is very wide even though the datarate may be relatively > small. Therefore in order for demodulation and detection to be performed > in the digital domain, very fast ADCs and front-end DSP must be > utilized. In some modern IR-UWB systems, this may not be feasible due to > the extreme bandwidth (>1 GHz) of the signals. > > Since the underlying data rate is relatively low, much of this bandwidth > is "redundant" in some sense. My question is this: do techniques exist > in which some of the initial demodulation is performed in the analog > domain, the output of which is a much smaller bandwidth signal that can > then be more tractably (and economically!) converted and processed in > the digital domain? > > I'd love to hear ideas on this. > -- > % Randy Yates % "So now it's getting late, > %% Fuquay-Varina, NC % and those who hesitate > %%% 919-577-9882 % got no one..." > %%%% <ya...@ieee.org> % 'Waterfall', *Face The Music*, ELOhttp://www.digitalsignallabs.comRandy, While not in the communications fields, Radar systems often LFM / Chirp waveforms which can have a wide bandwith. To ease the ADC requirements Stretch processing is often performed in the analog domain. Instead of mixing the received signal with a constant frequency sinusoid it is mixed with a signal whose frequency changes linearly with time - usually the chirp rate matches that of the transmitted signal. For radar signals a target at a specific range appears as a constant frequency (different ranges appear at different frequencies). To complete the pulse compression / matched filter, you simply perform an FFT, and you can relate the frequency bins to particular ranges. Not sure if that helps or not - or if it was what you were looking for. Cheers, David
Reply by ●April 4, 20082008-04-04
On Apr 4, 7:56 am, Randy Yates <ya...@ieee.org> wrote:> In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at > the receiver is very wide even though the datarate may be relatively > small. Therefore in order for demodulation and detection to be performed > in the digital domain, very fast ADCs and front-end DSP must be > utilized. In some modern IR-UWB systems, this may not be feasible due to > the extreme bandwidth (>1 GHz) of the signals. > > Since the underlying data rate is relatively low, much of this bandwidth > is "redundant" in some sense. My question is this: do techniques exist > in which some of the initial demodulation is performed in the analog > domain, the output of which is a much smaller bandwidth signal that can > then be more tractably (and economically!) converted and processed in > the digital domain? >Ahem, *cough cough*. Economic Sampling of Parametric Signals Julius Kusuma, Ph.D. thesis Department of Electrical Engineering and Computer Science, MIT, August 2006. http://www.rle.mit.edu/stir/documents/KusumaJ_PhDThesis2006.pdf Also: Sampling in Wideband Communication Systems An especially appealing application of our sampling scheme is in certain classes of nonlinear estimation problems encountered in wideband communication systems, such as Ultra-Wideband (UWB) and wideband code division multiple access (W-CDMA) systems, where the bandwidth of a transmitted signal is much larger than the bandwidth or rate of information being sent. In such systems, synchronization is a crucial task that imposes serious restrictions on system performance. There is a vast literature that has appeared recently, addressing both algorithmic and implementation issues of various synchronization techniques, with a clear trend toward eliminating the necessity for analog components and performing all processing digitally. Even though many high-performance synchronization schemes have already been proposed, their application in real time systems is often not feasible due to their high computational complexity. Furthermore, almost all of them use the Nyquist sampling rate, which requires very fast and expensive A/D converters and therefore high power consumption. This problem becomes critical in ultra-wideband systems, where in digital- oriented solutions A/D converters must operate in the gigahertz range. We have proposed several frequency domain methods for channel estimation and synchronization, which use low-rate uniform sampling and well-developed algorithmic solutions. Our approach takes advantage of the low-rank property and the algebraic structure of the data matrix under noise-free conditions. It leads to reduced computational requirements and faster acquisition compared to proposed digital techniques, thus allowing for a practical hardware implementation and low power consumption. It is particularly suitable in applications such as precise position location or ranging. References: J. Kusuma, A. Ridolfi and M. Vetterli, Sampling of communications systems with bandwidth expansion, Proc. IEEE ICC, Vol. 3, pp. 1601-1605, 2002. [detailed record] [bibtex] I. Maravic and M. Vetterli, Digital DS-CDMA receivers working below the chip rate, Proc. 36th Asilomar Conference on Signals, Systems and Computers, Vol. 2, pp. 1463-1467, 2002. [detailed record] [bibtex] I. Maravic and M. Vetterli, Digital DS-CDMA receiver working below the chip rate: theory and design, Technical Report, 2002. [detailed record] [bibtex] I. Maravic and M. Vetterli, Low-Complexity Subspace Methods for Channel Estimation and Synchronization in Ultra-Wideband Systems, Proc. International Workshop on Ultra-Wideband Systems (IWUWB), 2003. [detailed record] [bibtex] J. Kusuma, I. Maravic and M. Vetterli, Sampling With Finite Rate of Innovation: Channel and Time Estimation for UWB and GPS, New Frontiers in Telecommunications, Vol. 5, pp. 3540-3544, 2003. [detailed record] [bibtex] I. Maravic, M. Vetterli and K. Ramchandran, High-Resolution Acquisition Methods for Wideband Communication Systems, IEEE Conference on Acoustics, Speech and Signal Processing, Vol. 4, pp. 133-136, 2003. [detailed record] [bibtex] I. Maravic, M. Vetterli and K. Ramchandran, Channel Estimation and Synchronization with Sub-Nyquist Sampling and Application to Ultra- Wideband Systems, Proceedings IEEE International Symposium on Circuits and Systems, Vol. 5, pp. 381-384, 2004. [detailed record] [bibtex] http://lcavwww.epfl.ch/research/topics/sampling_FRI.html
Reply by ●April 4, 20082008-04-04
julius <juliusk@gmail.com> writes:> On Apr 4, 7:56 am, Randy Yates <ya...@ieee.org> wrote: >> In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at >> the receiver is very wide even though the datarate may be relatively >> small. Therefore in order for demodulation and detection to be performed >> in the digital domain, very fast ADCs and front-end DSP must be >> utilized. In some modern IR-UWB systems, this may not be feasible due to >> the extreme bandwidth (>1 GHz) of the signals. >> >> Since the underlying data rate is relatively low, much of this bandwidth >> is "redundant" in some sense. My question is this: do techniques exist >> in which some of the initial demodulation is performed in the analog >> domain, the output of which is a much smaller bandwidth signal that can >> then be more tractably (and economically!) converted and processed in >> the digital domain? >> > > Ahem, *cough cough*. > > Economic Sampling of Parametric Signals > Julius Kusuma, Ph.D. thesis > Department of Electrical Engineering and Computer Science, MIT, August > 2006. > > http://www.rle.mit.edu/stir/documents/KusumaJ_PhDThesis2006.pdf > > Also:> [...]Wow! Thanks Julius! Hey what's your phone number? -- % Randy Yates % "Though you ride on the wheels of tomorrow, %% Fuquay-Varina, NC % you still wander the fields of your %%% 919-577-9882 % sorrow." %%%% <yates@ieee.org> % '21st Century Man', *Time*, ELO http://www.digitalsignallabs.com
Reply by ●April 4, 20082008-04-04
Dave <dspguy2@netscape.net> writes:> On Apr 4, 8:56 am, Randy Yates <ya...@ieee.org> wrote: >> In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at >> the receiver is very wide even though the datarate may be relatively >> small. Therefore in order for demodulation and detection to be performed >> in the digital domain, very fast ADCs and front-end DSP must be >> utilized. In some modern IR-UWB systems, this may not be feasible due to >> the extreme bandwidth (>1 GHz) of the signals. >> >> Since the underlying data rate is relatively low, much of this bandwidth >> is "redundant" in some sense. My question is this: do techniques exist >> in which some of the initial demodulation is performed in the analog >> domain, the output of which is a much smaller bandwidth signal that can >> then be more tractably (and economically!) converted and processed in >> the digital domain? >> >> I'd love to hear ideas on this. >> -- >> % Randy Yates % "So now it's getting late, >> %% Fuquay-Varina, NC % and those who hesitate >> %%% 919-577-9882 % got no one..." >> %%%% <ya...@ieee.org> % 'Waterfall', *Face The Music*, ELOhttp://www.digitalsignallabs.com > > Randy, > While not in the communications fields, Radar systems often LFM / > Chirp waveforms which can have a wide bandwith. To ease the ADC > requirements Stretch processing is often performed in the analog > domain. > > Instead of mixing the received signal with a constant frequency > sinusoid it is mixed with a signal whose frequency changes linearly > with time - usually the chirp rate matches that of the transmitted > signal. > > For radar signals a target at a specific range appears as a constant > frequency (different ranges appear at different frequencies). To > complete the pulse compression / matched filter, you simply perform an > FFT, and you can relate the frequency bins to particular ranges. > > Not sure if that helps or not - or if it was what you were looking > for.All suggestions welcome and potentially useful. Thanks much for your suggestion and your time to respond, David. -- % Randy Yates % "She has an IQ of 1001, she has a jumpsuit %% Fuquay-Varina, NC % on, and she's also a telephone." %%% 919-577-9882 % %%%% <yates@ieee.org> % 'Yours Truly, 2095', *Time*, ELO http://www.digitalsignallabs.com
Reply by ●April 4, 20082008-04-04
On Fri, 04 Apr 2008 08:56:06 -0400, Randy Yates <yates@ieee.org> wrote:>In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at >the receiver is very wide even though the datarate may be relatively >small. Therefore in order for demodulation and detection to be performed >in the digital domain, very fast ADCs and front-end DSP must be >utilized. In some modern IR-UWB systems, this may not be feasible due to >the extreme bandwidth (>1 GHz) of the signals. > >Since the underlying data rate is relatively low, much of this bandwidth >is "redundant" in some sense. My question is this: do techniques exist >in which some of the initial demodulation is performed in the analog >domain, the output of which is a much smaller bandwidth signal that can >then be more tractably (and economically!) converted and processed in >the digital domain? > >I'd love to hear ideas on this.I think the ease by which it could be done with analog circuits depends on the nature of the spreading. It's a bit harder to build and synchronize an analog spread-spectrum despreader than it is to use the sort of LFM pulse compression techniques that Dave described. There's also a dispersive filter construction that will do the same function. In the IR-UWB case finding an analog filter that matches the pulse is probably a good first step (and I have no idea how easy/hard that might be). FWIW, when Intel demonstrated IR-UWB at IDC in 2002 http://news.zdnet.co.uk/hardware/0,1000000091,2106347,00.htm (unfortunately it appears that they've taken the pictures down) http://www.ultrawidebandplanet.com/technology/article.php/1404791 a lot of the basic processing was analog. It's wasn't the same signal as the current standard, naturally, and it wasn't the sort of thing you'd want to sell on the market. Nevertheless, it was an IR-UWB system with a fair amount of analog processing. Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org
Reply by ●April 5, 20082008-04-05
Randy Yates wrote:> In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at > the receiver is very wide even though the datarate may be relatively > small. Therefore in order for demodulation and detection to be performed > in the digital domain, very fast ADCs and front-end DSP must be > utilized. In some modern IR-UWB systems, this may not be feasible due to > the extreme bandwidth (>1 GHz) of the signals. > > Since the underlying data rate is relatively low, much of this bandwidth > is "redundant" in some sense. My question is this: do techniques exist > in which some of the initial demodulation is performed in the analog > domain, the output of which is a much smaller bandwidth signal that can > then be more tractably (and economically!) converted and processed in > the digital domain? > > I'd love to hear ideas on this.If you think about history for a moment, you'll see there have to be approaches without high speed ADCs. Military radios used wide band spread spectrum techniques in the early 70s, when no amount of money could buy you a fast ADC of any quality. Certainly nothing that would fit into a portable radio. Steve
Reply by ●April 5, 20082008-04-05
Eric Jacobsen <eric.jacobsen@ieee.org> writes:> [...]Thanks for the feedback/references Eric.> It's wasn't the same signal as the current standard,What do you mean by "current standard?" -- % Randy Yates % "Remember the good old 1980's, when %% Fuquay-Varina, NC % things were so uncomplicated?" %%% 919-577-9882 % 'Ticket To The Moon' %%%% <yates@ieee.org> % *Time*, Electric Light Orchestra http://www.digitalsignallabs.com
Reply by ●April 5, 20082008-04-05
Steve Underwood <steveu@dis.org> writes:> Randy Yates wrote: >> In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at >> the receiver is very wide even though the datarate may be relatively >> small. Therefore in order for demodulation and detection to be performed >> in the digital domain, very fast ADCs and front-end DSP must be >> utilized. In some modern IR-UWB systems, this may not be feasible due to >> the extreme bandwidth (>1 GHz) of the signals. >> >> Since the underlying data rate is relatively low, much of this bandwidth >> is "redundant" in some sense. My question is this: do techniques exist >> in which some of the initial demodulation is performed in the analog >> domain, the output of which is a much smaller bandwidth signal that can >> then be more tractably (and economically!) converted and processed in >> the digital domain? >> >> I'd love to hear ideas on this. > > If you think about history for a moment, you'll see there have to be > approaches without high speed ADCs. Military radios used wide band > spread spectrum techniques in the early 70s, when no amount of money > could buy you a fast ADC of any quality. Certainly nothing that would > fit into a portable radio.Hi Steve, DSSS and IR-UWB differ, if I'm correct, in one very significant way in this regard: the bandwidth of a DSSS signal is tied to its data rate (it's data rate is multiplied by the spreading factor), but the bandwidth of a IR-UWB signal is constant independent of the data rate and is established by the pulse shape and width. So it's not clear to me whether early military spread-spectrum projects just used smaller data rates, or if they used some sort of technology like this. -- % Randy Yates % "She tells me that she likes me very much, %% Fuquay-Varina, NC % but when I try to touch, she makes it %%% 919-577-9882 % all too clear." %%%% <yates@ieee.org> % 'Yours Truly, 2095', *Time*, ELO http://www.digitalsignallabs.com
Reply by ●April 6, 20082008-04-06
Randy Yates wrote:> Steve Underwood <steveu@dis.org> writes: > >> Randy Yates wrote: >>> In wideband systems such as CDMA (DSSS) and IR-UWB, the analog signal at >>> the receiver is very wide even though the datarate may be relatively >>> small. Therefore in order for demodulation and detection to be performed >>> in the digital domain, very fast ADCs and front-end DSP must be >>> utilized. In some modern IR-UWB systems, this may not be feasible due to >>> the extreme bandwidth (>1 GHz) of the signals. >>> >>> Since the underlying data rate is relatively low, much of this bandwidth >>> is "redundant" in some sense. My question is this: do techniques exist >>> in which some of the initial demodulation is performed in the analog >>> domain, the output of which is a much smaller bandwidth signal that can >>> then be more tractably (and economically!) converted and processed in >>> the digital domain? >>> >>> I'd love to hear ideas on this. >> If you think about history for a moment, you'll see there have to be >> approaches without high speed ADCs. Military radios used wide band >> spread spectrum techniques in the early 70s, when no amount of money >> could buy you a fast ADC of any quality. Certainly nothing that would >> fit into a portable radio. > > Hi Steve, > > DSSS and IR-UWB differ, if I'm correct, in one very significant way in > this regard: the bandwidth of a DSSS signal is tied to its data > rate (it's data rate is multiplied by the spreading factor), but the > bandwidth of a IR-UWB signal is constant independent of the data rate > and is established by the pulse shape and width. > > So it's not clear to me whether early military spread-spectrum projects > just used smaller data rates, or if they used some sort of technology > like this.Well, the ones I know (only a little) about used rather low bit rates, and gave pretty basic voice quality. However, they were spreading for stealth and jam immunity, not basic comms performance. Therefore, they spread over a huge band. Steve
