I am studying how to establish a 802.11ac multipath channel for my WiFi simulation.
As we know, the impact of channel is equivalent to that the transmit signal go through a "channel filter". Therefore, the specification provides the coefficients of channel taps to us so that we can build the "channel filter".
However, It is based on that the sampling rate of baseband signal is equal to that of the "channel filter".
The problem I found that my baseband signal has the sampling rate of 80MHz, but the channel tap spacing is 5nsec (100MHz). There are different sampling rates!
Then, how can I do for the mismatch of sampling rates, and get a correct simulation?
Kind Regards
Jiajun Zhu
Hi Jiajun
802.11ac channel model are specified in delay domain with e.g. 5 ns delay resolution. You can simply transform the generated channel model realization (i.e. channel impulse responses) with the discrete Fourier transform to channel frequency responses on your simulation bandwidth. The update rate of radio channel is much lower compared to you system bandwidth, because in the model the Doppler shifts are in order of Hertz, not kHz or MHz.
Regards
-Pekka
Rate matching is accomplished by resampling the baseband signal by an interpolation of 5 divided by four, that is for every four samples are substituted by five samples..You can assume that the fifth sample i an average value of the other four and inserted in the middle of then, that i two samples before and two samples after.
Wish you success
This paper might be useful for your question. The simplest way would be to create a channel model that is an approximation to the original one by shifting each channel tap to the nearest sample point...
Conference Paper Conversion of reference tapped delay line channel models to ...
Thanks to Pekka Kyösti, Abdelhalim Zekry, Youssef El Hajj Shehadeh, Tarek Helaly and Lei Shi, firstly.
Your ideas are very useful to me.
Particularly, I am curious to the method provided by Prof. Abdelhalim Zekry. Why the fifth sample is equal to the average of other four samples? Further, why the fifth sample is inserted in the middle of the four samples?
I think that the resampling is the best way. On my way, I would upsample the 80MHz baseband signal to 400MHz by interpolation, then downsample the 400MHz baseband signal to 100MHz by decimation.
However, the computation is very high! In addition, the filters used by interpolation and decimation cannot perfectly filter out the sampling noise. Furthermore, the unperfect filter can decay the useful signal bandwidth.
The method provided by Prof. Abdelhalim Zekry is very simple and efficient, but i dont understand the principle of this method. Thus, I am curious to this method.
Many thanks
Jiajun Zhu