we have done several postings about our LTE testbed using the Zynq Xilinx and the FMCOMMS5 boards.
We are now in the process of testing our LTE baseband on a no-OS example, i.e. our C code implementation is running on the processor of the Zynq board. The fact is that we have done all steps correctly including the CFO (carrier frequency offset) estimation in order to compensate the RF board impairments. For the CFO in particular, we used the same process found in your LTE Matlab example here: MathWorks_tools/LTEReceiver.m at master · analogdevicesinc/MathWorks_tools · GitHub and the lines 86-92, providing first frequency offset and second time offset estimation. All baseband modules are correct and been tested.
However, we encounter the problem of not taking a small BER, expecting zero actually, using the loopback communication on the FMCOMMS5 board, i.e. using one Zynq board and one FMCOMMS5 board. It is in fact 0.5 always that means something goes wrong.
In order to find out what is going wrong, we have implemented an LTE baseband processing for your LTE Matlab example in order to calculate the BER of the LTE transmission. We made it and the BER is always found zero!
In sequel, and in order to find out where is the wrong part in our implementation, we tested the power levels of the transmitted and the received LTE waveforms for both C code and your Matlab implementation.
Let me summarise the outcomes of this power testing:
- the transmitted LTE waveform using our C code implementation is 0 dBm and your Matlab's case is -`12.5 dBm.
- the received LTE waveform using our C code implementation is 0 dBm and your Matlab's case is 55 dBm.
Please note that we use the dsp.SpectrumAnalyzer function in order to plot them before the RF Transmission and after the RF reception.
Now, our main question is why the received power of your Matlab LTE implementation is at this highest power level and how this is achieved? Note that we use the same RF configuration, for our no-OS example, i.e. slow attack etc. Using manual gain control, a maximum of 10 dBm is achieved in our no-OS example.
An additional question is whether the level of the received power (i.e. our 0 dBm) of the LTE waveform impacts the BER calculation or the 0 dBm are sufficient to have a normal (e.g. 0.001) BER.