Getting high values of BER when employinh a RCC shaping filter in a 16 QAM system

Hi guys,


I have been struggling for while with this problem and I am not able to solve it. I want to consider a single carrier system with 16 QAM with pulse shaping, until this point the system works just fine. I use a Square Raised Cosine Filter both in TX and RX and I do not get any interference, and thus 0 value of BER.

However whenI add a pseudo random sequence as preamble for channel estimation purposes I do get ISI. I believe the problem is with using different modulation for data and preamble and when I filtered data + preamble the filtering process introduces ISI.

There is anyone that could tell how can I solve this issue? The following picture shows the simulink implementation:



Many thanks,

Rodolfo

ISI is from non flat group delay within the channel BW for preamble and data.
Measure it and fix that.

Hi SunnySkyguy,

Many thanks for your helpful information. I am measuring the group delay between the TX and RX inputs and it is fixed to 40 ( filter span in symbols*oversampling factor) regardless the use of preamble.


Am I measuring the delay in the right position? I would appreciate if you could give some more information about this issue, for example: How does the preamble length changes the delay of the system? I could not find any information about it on the literature.

Many thanks,

Rodolfo

- - - Updated - - -

r_gomes said:

Hi SunnySkyguy,

Many thanks for your helpful information. I am measuring the group delay between the TX and RX inputs and it is fixed to 40 ( filter span in symbols*oversampling factor) regardless the use of preamble.


Am I measuring the delay in the right position? I would appreciate if you could give some more information about this issue, for example: How does the preamble length changes the delay of the system? I could not find any information about it on the literature.

Many thanks,

Rodolfo

Click to expand...

I also noticed that if I discard the first 11 samples of the received symbol sequence, I do not get ISI on the constellation, as you can see in the attached pictures. However, discarding the same samples on the TX sequences and considering the same group delay (40) I still get high BER.