Different transmitters simultaneously send their signals to a common receiver and as expected, the Rx receives a superposition of all the signals with different timing offsets (due to different delays) and different channel gains (flat fading assumed). Multiple timing offsets from the transmitters need to be found and time aligned at the Rx. Do you have any clue where to start? Any idea, reference, etc. will be deeply appreciated. Thank you.
Multiple timing offsets
Started by ●April 11, 2012
Reply by ●April 11, 20122012-04-11
Reply by ●April 11, 20122012-04-11
>Different transmitters simultaneously send their signals to a common >receiver and as expected, the Rx receives a superposition of all the >signals with different timing offsets (due to different delays) and >different channel gains (flat fading assumed). Multiple timing offsetsfrom>the transmitters need to be found and time aligned at the Rx. Do you have >any clue where to start? Any idea, reference, etc. will be deeply >appreciated. > >Thank you. >Spread spectrum?
Reply by ●April 11, 20122012-04-11
On Wednesday, April 11, 2012 4:04:07 PM UTC-4, commsignal wrote:> Different transmitters simultaneously send their signals to a common > receiver and as expected, the Rx receives a superposition of all the > signals with different timing offsets (due to different delays) and > different channel gains (flat fading assumed). Multiple timing offsets from > the transmitters need to be found and time aligned at the Rx. Do you have > any clue where to start? Any idea, reference, etc. will be deeply > appreciated. > > Thank you.The keyword you may want to search on is "multiuser detection" or MUD for short. If you have spread-spectrum signals separated in time by more than a few chips, the solution is relatively straightforward because the signals are already decorrelated.
Reply by ●April 11, 20122012-04-11
On Wed, 11 Apr 2012 16:00:40 -0500, "jacobfenton" <jacob.fenton@n_o_s_p_a_m.gmail.com> wrote:>>Different transmitters simultaneously send their signals to a common >>receiver and as expected, the Rx receives a superposition of all the >>signals with different timing offsets (due to different delays) and >>different channel gains (flat fading assumed). Multiple timing offsets >from >>the transmitters need to be found and time aligned at the Rx. Do you have >>any clue where to start? Any idea, reference, etc. will be deeply >>appreciated. >> >>Thank you. >> > >Spread spectrum?Or FDMA? If they're not orthogonal in frequency, i.e., they're spread-spectrum or have some other kind of orthogonality or separability, then as John mentioned the best approach is probably a MUD (multi-user detection) technique. Alternatively, if they're adequately separated spatially some sort of beamforming with a MIMO (or maybe MISO) array could be applied. I'd think MUD might be simpler, though. Eric Jacobsen Anchor Hill Communications www.anchorhill.com
Reply by ●April 12, 20122012-04-12
>On Wed, 11 Apr 2012 16:00:40 -0500, "jacobfenton" ><jacob.fenton@n_o_s_p_a_m.gmail.com> wrote: > >>>Different transmitters simultaneously send their signals to a common >>>receiver and as expected, the Rx receives a superposition of all the >>>signals with different timing offsets (due to different delays) and >>>different channel gains (flat fading assumed). Multiple timing offsets >>from >>>the transmitters need to be found and time aligned at the Rx. Do youhave>>>any clue where to start? Any idea, reference, etc. will be deeply >>>appreciated. >>> >>>Thank you. >>> >> >>Spread spectrum? > >Or FDMA? > >If they're not orthogonal in frequency, i.e., they're spread-spectrum >or have some other kind of orthogonality or separability, then as >John mentioned the best approach is probably a MUD (multi-user >detection) technique. > >Alternatively, if they're adequately separated spatially some sort of >beamforming with a MIMO (or maybe MISO) array could be applied. I'd >think MUD might be simpler, though. > > > >Eric Jacobsen >Anchor Hill Communications >www.anchorhill.com >If multiplexing of any kind is not used, and all the links are conventional narrowband single-carrier systems, then what can be done? Secondly, we need time alignment before applying MUD, is it correct?
Reply by ●April 12, 20122012-04-12
On Wed, 11 Apr 2012 22:17:33 -0500, "commsignal" <commsignal@n_o_s_p_a_m.yahoo.com> wrote:>>On Wed, 11 Apr 2012 16:00:40 -0500, "jacobfenton" >><jacob.fenton@n_o_s_p_a_m.gmail.com> wrote: >> >>>>Different transmitters simultaneously send their signals to a common >>>>receiver and as expected, the Rx receives a superposition of all the >>>>signals with different timing offsets (due to different delays) and >>>>different channel gains (flat fading assumed). Multiple timing offsets >>>from >>>>the transmitters need to be found and time aligned at the Rx. Do you >have >>>>any clue where to start? Any idea, reference, etc. will be deeply >>>>appreciated. >>>> >>>>Thank you. >>>> >>> >>>Spread spectrum? >> >>Or FDMA? >> >>If they're not orthogonal in frequency, i.e., they're spread-spectrum >>or have some other kind of orthogonality or separability, then as >>John mentioned the best approach is probably a MUD (multi-user >>detection) technique. >> >>Alternatively, if they're adequately separated spatially some sort of >>beamforming with a MIMO (or maybe MISO) array could be applied. I'd >>think MUD might be simpler, though. >> >> >> >>Eric Jacobsen >>Anchor Hill Communications >>www.anchorhill.com >> > >If multiplexing of any kind is not used, and all the links are conventional >narrowband single-carrier systems, then what can be done? Secondly, we need >time alignment before applying MUD, is it correct?You need synchronization with at least one signal at a time for MUD, if you do something like successive cancellation. It's not always possible. Eric Jacobsen Anchor Hill Communications www.anchorhill.com
Reply by ●April 12, 20122012-04-12
>On Wed, 11 Apr 2012 22:17:33 -0500, "commsignal" ><commsignal@n_o_s_p_a_m.yahoo.com> wrote: > >>>On Wed, 11 Apr 2012 16:00:40 -0500, "jacobfenton" >>><jacob.fenton@n_o_s_p_a_m.gmail.com> wrote: >>> >>>>>Different transmitters simultaneously send their signals to a common >>>>>receiver and as expected, the Rx receives a superposition of all the >>>>>signals with different timing offsets (due to different delays) and >>>>>different channel gains (flat fading assumed). Multiple timingoffsets>>>>from >>>>>the transmitters need to be found and time aligned at the Rx. Do you >>have >>>>>any clue where to start? Any idea, reference, etc. will be deeply >>>>>appreciated. >>>>> >>>>>Thank you. >>>>> >>>> >>>>Spread spectrum? >>> >>>Or FDMA? >>> >>>If they're not orthogonal in frequency, i.e., they're spread-spectrum >>>or have some other kind of orthogonality or separability, then as >>>John mentioned the best approach is probably a MUD (multi-user >>>detection) technique. >>> >>>Alternatively, if they're adequately separated spatially some sort of >>>beamforming with a MIMO (or maybe MISO) array could be applied. I'd >>>think MUD might be simpler, though. >>> >>> >>> >>>Eric Jacobsen >>>Anchor Hill Communications >>>www.anchorhill.com >>> >> >>If multiplexing of any kind is not used, and all the links areconventional>>narrowband single-carrier systems, then what can be done? Secondly, weneed>>time alignment before applying MUD, is it correct? > >You need synchronization with at least one signal at a time for MUD, >if you do something like successive cancellation. It's not always >possible. > > >Eric Jacobsen >Anchor Hill Communications >www.anchorhill.com >When is it possible? Any alternative idea from anyone will be appreciated. Thanks.
Reply by ●April 12, 20122012-04-12
On 4/12/2012 6:29 AM, commsignal wrote:>> On Wed, 11 Apr 2012 22:17:33 -0500, "commsignal" >> <commsignal@n_o_s_p_a_m.yahoo.com> wrote: >> >>>> On Wed, 11 Apr 2012 16:00:40 -0500, "jacobfenton" >>>> <jacob.fenton@n_o_s_p_a_m.gmail.com> wrote: >>>> >>>>>> Different transmitters simultaneously send their signals to a common >>>>>> receiver and as expected, the Rx receives a superposition of all the >>>>>> signals with different timing offsets (due to different delays) and >>>>>> different channel gains (flat fading assumed). Multiple timing > offsets >>>>> from >>>>>> the transmitters need to be found and time aligned at the Rx. Do you >>> have >>>>>> any clue where to start? Any idea, reference, etc. will be deeply >>>>>> appreciated. >>>>>> >>>>>> Thank you. >>>>>> >>>>> >>>>> Spread spectrum? >>>> >>>> Or FDMA? >>>> >>>> If they're not orthogonal in frequency, i.e., they're spread-spectrum >>>> or have some other kind of orthogonality or separability, then as >>>> John mentioned the best approach is probably a MUD (multi-user >>>> detection) technique. >>>> >>>> Alternatively, if they're adequately separated spatially some sort of >>>> beamforming with a MIMO (or maybe MISO) array could be applied. I'd >>>> think MUD might be simpler, though. >>>> >>>> >>>> >>>> Eric Jacobsen >>>> Anchor Hill Communications >>>> www.anchorhill.com >>>> >>> >>> If multiplexing of any kind is not used, and all the links are > conventional >>> narrowband single-carrier systems, then what can be done? Secondly, we > need >>> time alignment before applying MUD, is it correct? >> >> You need synchronization with at least one signal at a time for MUD, >> if you do something like successive cancellation. It's not always >> possible. >> >> >> Eric Jacobsen >> Anchor Hill Communications >> www.anchorhill.com >> > > When is it possible? > Any alternative idea from anyone will be appreciated. > Thanks.A lot usually depends on the signal design. For example, a RAKE receiver is based on a signal design that consists of sinusoidal "chips" that are at hopping frequencies. The order of the transmitted frequencies is known at the receiver. In this design the same signal is passed through different delay channels. The idea is to add up all the energy. So it's a little different than your problem. But it's likely instructive. All of the received energy is passed through a delay line structure .. like a FIR filter without the summing being done. The delays match the frequency change intervals in order to line up all the frequency components at the same time. Then, there's a bandpass filter for each. Finally an incoherent or post-dectection sum is computed. It's done this way in place of a matched filter which would be messed up due to the multipath. It's rather like "the next best thing". Anyway, my point is that the signal design is important. The key and direct question is: What is there about the signal that allows separation? What is there about the preamble that might help deal with this? What spectral distinction is there? What temporal distinction is there? What correlation distinction is there? Each thing I think of gets right back to signal design: If the transmitter intervals are fixed then you ought to be able to "gate" a single transmitter's received signal easily. That is, listen at only the right times. Then it's only a matter of synchronizing the receiver to the transmitter of interest - like tuning a radio. Not trivial perhaps but possible. If the transmitters outputs do not overlap in time then this method should work to separate all the Tx/Rx ensembles. But, if the signals overlap in time and in frequency and have no special correlation properties then I don't think there is a way in the world. Every signal is "noise" to every other signal and, if what I fear is the case, there is perhaps no distinction that filtering will help resolve? You haven't really told us much so far so it's really hard to suggest a solution. Fred
Reply by ●April 12, 20122012-04-12
On Thu, 12 Apr 2012 08:29:51 -0500, "commsignal" <commsignal@n_o_s_p_a_m.yahoo.com> wrote:>>On Wed, 11 Apr 2012 22:17:33 -0500, "commsignal" >><commsignal@n_o_s_p_a_m.yahoo.com> wrote: >> >>>>On Wed, 11 Apr 2012 16:00:40 -0500, "jacobfenton" >>>><jacob.fenton@n_o_s_p_a_m.gmail.com> wrote: >>>> >>>>>>Different transmitters simultaneously send their signals to a common >>>>>>receiver and as expected, the Rx receives a superposition of all the >>>>>>signals with different timing offsets (due to different delays) and >>>>>>different channel gains (flat fading assumed). Multiple timing >offsets >>>>>from >>>>>>the transmitters need to be found and time aligned at the Rx. Do you >>>have >>>>>>any clue where to start? Any idea, reference, etc. will be deeply >>>>>>appreciated. >>>>>> >>>>>>Thank you. >>>>>> >>>>> >>>>>Spread spectrum? >>>> >>>>Or FDMA? >>>> >>>>If they're not orthogonal in frequency, i.e., they're spread-spectrum >>>>or have some other kind of orthogonality or separability, then as >>>>John mentioned the best approach is probably a MUD (multi-user >>>>detection) technique. >>>> >>>>Alternatively, if they're adequately separated spatially some sort of >>>>beamforming with a MIMO (or maybe MISO) array could be applied. I'd >>>>think MUD might be simpler, though. >>>> >>>> >>>> >>>>Eric Jacobsen >>>>Anchor Hill Communications >>>>www.anchorhill.com >>>> >>> >>>If multiplexing of any kind is not used, and all the links are >conventional >>>narrowband single-carrier systems, then what can be done? Secondly, we >need >>>time alignment before applying MUD, is it correct? >> >>You need synchronization with at least one signal at a time for MUD, >>if you do something like successive cancellation. It's not always >>possible. >> >> >>Eric Jacobsen >>Anchor Hill Communications >>www.anchorhill.com >> > >When is it possible? >Any alternative idea from anyone will be appreciated. >Thanks.Successive cancellation proceeds by detecting and removing (by cancellation) the highest SNR signal first, then the next, etc., etc., until no more can be detected. If the SNR (SNIR, actually) is equal for all signals or is too low to sufficiently detect any signal for cancellation, then obviously the operation can't be done. Statistically, for most cellular systems, the link terminal distribution will usually be fairly random through the antenna pattern so that it will be expected that there will be a fair amount of spread in the SNRs of the received signals. It's a complicated process and not trivial to get working, even for the easiest cases. Eric Jacobsen Anchor Hill Communications www.anchorhill.com
