Dear all, I want to model multipath channel. Why multipath channel can modelled as Finite Impulse Response (FIR) filters (what is the relationship between FIR filters and multipath channel)? and how to model it as FIR filters? Thank you very much. Best wishes, estdev
FIR + multipath channel
Started by ●May 15, 2005
Reply by ●May 15, 20052005-05-15
"estdev" <nmaedewi@yahoo.com> wrote in message news:1116140459.141107.189560@g43g2000cwa.googlegroups.com...> Dear all, > > > I want to model multipath channel. Why multipath channel can modelled > as Finite Impulse Response (FIR) filters (what is the relationship > between FIR filters and multipath channel)? and how to model it as FIR > filters? > > Thank you very much.If the receiver in the multipath channel receives signals from a single source, generally each path in the multipath is delayed one from another and generally each path has different attenuation. A FIR filter has as input a single sequence and the output is a set of signals that have been delayed and weighted (attenuated if you will). I hope the resemblance is clear enough now. For more than one source you would probably use one FIR filter per source. So, a FIR filter can be defined (perhaps a very sparse filter - i.e. one with many zero coefficients) that has an output that is the same as a (perhaps simplified) multiple-path environment. This implies that the simulating filter have a unit delay that is adequate to resolve the paths as well as may be desired. In general, delay is not known for a path to within the propagation time of a single wavelength - so resolution down to sub-wavelength times is most often unecessary. If antennas or arrays are involved then the output of the filter needs to take into account the attenuation at each angle of arrival due to the receive beam pattern. At short range, the transmitter beam pattern may also be important. Or, maybe I should really say that at long range maybe the transmitter beam pattern could be neglected. It depends on the geometry. Things get more complicated and depart from the FIR filter model when the paths are dynamic and introduce Doppler shifts, etc. Fred
Reply by ●May 15, 20052005-05-15
estdev wrote:> Dear all, > > > I want to model multipath channel. Why multipath channel can modelled > as Finite Impulse Response (FIR) filters (what is the relationship > between FIR filters and multipath channel)? and how to model it as FIR > filters? > > Thank you very much.The coefficients if an FIR represent the impulse response of the filter, a room. a channel, or anything one chooses to model. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●May 15, 20052005-05-15
Fred Marshall wrote:> If the receiver in the multipath channel receives signals from a > single source, generally each path in the multipath is delayed > one from another and generally each path has different attenuation. > > A FIR filter has as input a single sequence and the output is a > set of signals that have been delayed and weighted (attenuated > if you will). > > I hope the resemblance is clear enough now. > > For more than one source you would probably use one FIR filter > per source.Na. Just model the whole response as a single FIR, with possibly gaps between the active taps for direct-path signals and active taps for indirect-path (more delay) signals.> So, a FIR filter can be defined (perhaps a very sparse filter - i.e. > one with many zero coefficients) that has an output that is the same > as a (perhaps simplified) multiple-path environment.Yup. <Other good stuff snipped> Ciao, Peter K.
Reply by ●May 15, 20052005-05-15
Peter K. wrote:> Fred Marshall wrote: > > >>If the receiver in the multipath channel receives signals from a >>single source, generally each path in the multipath is delayed >>one from another and generally each path has different attenuation. >> >>A FIR filter has as input a single sequence and the output is a >>set of signals that have been delayed and weighted (attenuated >>if you will). >> >>I hope the resemblance is clear enough now. >> >>For more than one source you would probably use one FIR filter >>per source. > > > Na. Just model the whole response as a single FIR, with possibly gaps > between the active taps for direct-path signals and active taps for > indirect-path (more delay) signals.Hunh? With two sources, each with it's own reflection pattern (i.e. path impulse response) and original data (i.e. filter input), how do you do that? Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●May 15, 20052005-05-15
"Jerry Avins" <jya@ieee.org> wrote in message news:OuKdnWQdMaiXWRrfRVn-gQ@rcn.net...> Peter K. wrote: >> Fred Marshall wrote: >> >> >>>If the receiver in the multipath channel receives signals from a >>>single source, generally each path in the multipath is delayed >>>one from another and generally each path has different attenuation. >>> >>>A FIR filter has as input a single sequence and the output is a >>>set of signals that have been delayed and weighted (attenuated >>>if you will). >>> >>>I hope the resemblance is clear enough now. >>> >>>For more than one source you would probably use one FIR filter >>>per source. >> >> >> Na. Just model the whole response as a single FIR, with possibly gaps >> between the active taps for direct-path signals and active taps for >> indirect-path (more delay) signals. > > Hunh? With two sources, each with it's own reflection pattern (i.e. path > impulse response) and original data (i.e. filter input), how do you do > that? > > JerryExactly Jerry...
Reply by ●May 15, 20052005-05-15
D'oh. I was thinking separate paths, not sources. I should read more carefully. :-) Thanks, guys. Ciao, Peter K.
Reply by ●May 16, 20052005-05-16
On Sun, 15 May 2005 17:50:01 -0400, Jerry Avins <jya@ieee.org> wrote:>Peter K. wrote: >> Fred Marshall wrote: >> >> >>>If the receiver in the multipath channel receives signals from a >>>single source, generally each path in the multipath is delayed >>>one from another and generally each path has different attenuation. >>> >>>A FIR filter has as input a single sequence and the output is a >>>set of signals that have been delayed and weighted (attenuated >>>if you will). >>> >>>I hope the resemblance is clear enough now. >>> >>>For more than one source you would probably use one FIR filter >>>per source. >> >> >> Na. Just model the whole response as a single FIR, with possibly gaps >> between the active taps for direct-path signals and active taps for >> indirect-path (more delay) signals. > >Hunh? With two sources, each with it's own reflection pattern (i.e. path >impulse response) and original data (i.e. filter input), how do you do that? > >JerryAre the two sources transmitting the same thing? Are they interferers? Do you listen to them at different times? Actually I don't understand which was meant here. If the propagation environment is static and the receiver has a single antenna then even two sources will have static FIR channel representations. They'll likely be different (i.e., different coefficients), but they'll still be static. If both sources are transmitting the same thing (like a single-frequency network) then the receiver can still sort it all out as a single FIR representation. Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
Reply by ●May 16, 20052005-05-16
Eric Jacobsen wrote: ...> If the propagation environment is static and the receiver has a single > antenna then even two sources will have static FIR channel > representations. They'll likely be different (i.e., different > coefficients), but they'll still be static. If both sources are > transmitting the same thing (like a single-frequency network) then the > receiver can still sort it all out as a single FIR representation.I don't believe that one FIR can model two sources, each whistling its own tune. Peter seemed to imply that it could, then corrected his remark. We're probably digging too deep*. Jerry ____________________________________ * "Deep" is an adjective that modifies the unnamed excavation. "Deeply", the adverb that ignorant purists might prescribe, would modify the process of digging. I once told one such knee-jerk grammarian that If I hadn't such high regard for her, I would suggest that she drop deadly. -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●May 16, 20052005-05-16
"Eric Jacobsen" <eric.jacobsen@ieee.org> wrote in message news:4288e609.48544625@news.west.cox.net...> On Sun, 15 May 2005 17:50:01 -0400, Jerry Avins <jya@ieee.org> wrote: > >>Peter K. wrote: >>> Fred Marshall wrote: >>> >>> >>>>If the receiver in the multipath channel receives signals from a >>>>single source, generally each path in the multipath is delayed >>>>one from another and generally each path has different attenuation. >>>> >>>>A FIR filter has as input a single sequence and the output is a >>>>set of signals that have been delayed and weighted (attenuated >>>>if you will). >>>> >>>>I hope the resemblance is clear enough now. >>>> >>>>For more than one source you would probably use one FIR filter >>>>per source. >>> >>> >>> Na. Just model the whole response as a single FIR, with possibly gaps >>> between the active taps for direct-path signals and active taps for >>> indirect-path (more delay) signals. >> >>Hunh? With two sources, each with it's own reflection pattern (i.e. path >>impulse response) and original data (i.e. filter input), how do you do >>that? >> >>Jerry > > Are the two sources transmitting the same thing? Are they > interferers? Do you listen to them at different times? Actually I > don't understand which was meant here. > > If the propagation environment is static and the receiver has a single > antenna then even two sources will have static FIR channel > representations. They'll likely be different (i.e., different > coefficients), but they'll still be static. If both sources are > transmitting the same thing (like a single-frequency network) then the > receiver can still sort it all out as a single FIR representation. > > Eric Jacobsen > Minister of Algorithms, Intel Corp. > My opinions may not be Intel's opinions. > http://www.ericjacobsen.orgEric, There's no reason to believe that independent sources would be transmitting the same things. If there were two sources transmitting the same things then timing and geometry still matter. If the model admits to a *single* input that represents *all* sources then a single FIR can simulate all paths as you've suggested. However, if there is any reason to require two inputs then a single FIR can't simulate the source/path combination because one can't apply common geometry to two sources unless they are coincident in space. Fred
