Hi Kaan,
it depends on the range of your I/Q samples. In the example the data type is converted from double to fixed-point. Here you must know the range of your data. In the example you can check:
>> max(rxData)
ans =
-5.1297 - 0.0465i
>> min(rxData)
ans =
-1.7579e-04 - 1.3235e-04i
With that you choose the signedness, word length, and fraction length. With fixdt(1,18,14) you will have a range between -8 .. 8-2^-14. When you reduce the word length you also have to reduce the fraction length by the same amount of bits to keep this range. For 8 bit word length: fixdt(1,8,4) which gives a range of -8 .. 8-2^-4. You will have much less precision.
Your data could be complete integter with a range between -128 .. 127. You can interpret the values differently but it would not work using it directly as rxData without changing settings (range would be to high -> overflow). You can do this with the same Data Type Conversion block by changing the "Input and output to have equal:" to stored integer.
It could be necessary to also make fixed-point changes within the "Acquisition and Tracking" subsystem.
Best regards,
Tom
