|
Prev: FFT on dsPIC33F
Next: SHARC, cache and sdram
From: Dani Camps on 7 May 2008 10:08 Dear all, I am trying to model a multipath in matlab but I do not have the communications toolbox and so neither the raileighchan function. I would appreciate your help on assessing if my approach to generate the FIR discrete channel is correct. I have two vectors containing the complex gain and the delay of each path in my channel, taps_gain and taps_delay and a variable, Ts, containing the sample period in the receiver. Now in order to generate the FIR components of the channel I do: -------------------------------------------------------------------------------- num_taps = floor(1.5*max(taps_delay)/Ts); h_bb=[]; for n = 0:num_taps; g_n = sum(taps_gain .* sinc((n- taps_delay/Ts))); h_bb = [h_bb g_n]; end --------------------------------------------------------------------------------- Where I heuristically approximate the number of taps in the channel as 1.5*Delay_Spread/Sampling_Time, since the contribution of the different paths to taps after Delay_Spread/Sampling_Time is going to be small. Is this approximation correct ? To compute the contribution of each path to each FIR component of the channel, I use the standard procedure of weighting the gain of each tap with a delayed sinc function. Another question is whether the range that I use, n=0:num_taps, is correct or should I use n=-num_taps/2:num_taps ? I think Matlab does it in the other way. Is there any other thing that I am missing in order to create the FIR response of the channel ? One last question once I have the FIR taps of the channel, h_bb, then in order to obtain the frequency response I should simply do: F = fft(h_bb, N); And to transform the discrete frequency from the FFT into a real frequency: f = 1/Ts *(0:N-1)/N; Is the previous correct ? Thanks for clarifying if my understanding. Best Regards Daniel
From: cpshah99 on 7 May 2008 13:10 >Dear all, > >I am trying to model a multipath in matlab but I do not have the >communications toolbox and so neither the raileighchan function. I >would appreciate your help on assessing if my approach to generate the >FIR discrete channel is correct. > >I have two vectors containing the complex gain and the delay of each >path in my channel, taps_gain and taps_delay and a variable, Ts, >containing the sample period in the receiver. > >Now in order to generate the FIR components of the channel I do: > >-------------------------------------------------------------------------------- >num_taps = floor(1.5*max(taps_delay)/Ts); > >h_bb=[]; >for n = 0:num_taps; > g_n = sum(taps_gain .* sinc((n- taps_delay/Ts))); > h_bb = [h_bb g_n]; >end >--------------------------------------------------------------------------------- > >Where I heuristically approximate the number of taps in the channel as >1.5*Delay_Spread/Sampling_Time, since the contribution of the >different paths to taps after Delay_Spread/Sampling_Time is going to >be small. Is this approximation correct ? > >To compute the contribution of each path to each FIR component of the >channel, I use the standard procedure of weighting the gain of each >tap with a delayed sinc function. > >Another question is whether the range that I use, n=0:num_taps, is >correct or should I use n=-num_taps/2:num_taps ? I think Matlab does >it in the other way. > >Is there any other thing that I am missing in order to create the FIR >response of the channel ? > >One last question once I have the FIR taps of the channel, h_bb, then >in order to obtain the frequency response I should simply do: > >F = fft(h_bb, N); > >And to transform the discrete frequency from the FFT into a real >frequency: > >f = 1/Ts *(0:N-1)/N; > >Is the previous correct ? > >Thanks for clarifying if my understanding. > >Best Regards > >Daniel > > > > > > > > > %%%%%%% Hi I have been working on multipath channels but I have not modelled yet. Instead I am using standard channel from Digital Communications by John G. Proakis for my simulations. He gives three channels. One good, one bad and one worst. Just use this channel and use function 'conv' to introduce channel effect. Hope this helps. Chintan P Shah
|
Pages: 1 Prev: FFT on dsPIC33F Next: SHARC, cache and sdram |