I want to check the effect of this channel model on my signal transmitted with "impulse response-Ultra Wide Band" transmission (use very short pulses ~0.5 ns each), but i don't know where to find this channel impulse response or how can I generate it? if there is only one line of sight (LOS) path and one non-line of sight (NLOS) path, if I assume the channel impulse response something like h=[1 0 0 0 .4 0 0] and then filter(h,tx signal), it would be correct? best regards
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
Your "h" is a valid multipath channel impulse response. It is a special case, though, in that it represents a multipath delay that is an exact multiple of your sample period. Usually the delay will not be an exact multiple of the sample period.
I am not an expert on multipath channels, so hopefully someone else can chime in, but I think that the easiest way to generalize your multipath channel is to make the gain of the multipath complex rather than real. The phase of the gain would alter the phase of your signal, simulating a non-integer delay.
If you really wanted to be hardcore about it you could use this answer on how to make fractional delay filters to create a filter that delays the signal by as much or as little as you want, and then add it to [1 0 0 0 0...] (the LOS channel) to get the complete multipath channel.
$\begingroup$
$\endgroup$
I'm not sure Jim's answer is quite correct here: My understanding is that "fixed" refers to a situation where the transfer function is time-invariant, i.e. transmitter and receiver are stationary and the environment doesn't change. An example for that would be a radio station and a home radio receiver. The opposite of fixed would be variable, such as a car radio receiver or a cell phone. These are moving about so the transfer function and associated multi-path changes all the time.
Modelling of the channel paths as a simple delay (be it integer or fractional) is a gross simplification which only works if the signal itself is very narrow band. For an ultra wide band system that's probably not accurate and you will the frequency dependencies of the paths either in the frequency domain through properly capturing the impulse responses
