Update from UPV #49
Update from UPV #49
Conversation
The recevier doesn't search every cell ID when loses the synchrononization, tries to connect to the same already known cell.
…peat the reading of a sample file. The computation of the BLER was erronious, it has been fixed. Also the columns and structure on how the data is saved in the CSV has been change, now every time there's a new measurment a new row is saved in the file indicating the status of the receiver (SYNCING, SEARCHING, o printing the data if it's processing).
… and processors within the rt-wui
…the samples to a disk. With this secondary buffer, the chunks to be saved are bigger, and this way minimizes the blocking of writting when saving by small chunks.
…ty for the rt-wui Only pendind things are the displayed of cir and correlation.
|
@alessandro-lucco: Fyi, this PR includes the changes presented by @jsroldan in the last call. If you find the time, it would be nice if you can run some tests on your side as well. We are currently testing against sample recordings: https://5g-mag.github.io/Getting-Started/pages/lte-based-5g-broadcast/additional/sample-files.html |
Thanks Daniel and Jaime, I did some tests in the last day. |
Thanks for testing @alessandro-lucco . I only tested a sample file so far and that worked fine. Where does the live signal come from? Is it the 5G-MAG transmitter, or are you using a commercial solution like R&S? Do you see the same problem with the version that does not include the changes by @jsroldan ? |
At the moment i can only test with R&S transmitter. No, with the other versions I have always done the same steps (with same input signal) and I did not encounter segmentation faults, except in very few cases. I think it's a buffer issue related to the SDR management. |
|
The next week I'll debug it to try to figure out what is wrong, I've never tested with a SDR so thanks @alessandro-lucco for the finding! |
|
Comment from the call by @alessandro-lucco : No 5GBC signal needed, but SDR needs to be connected to reproduce the issue that @alessandro-lucco reported above. |
|
Hi all, I finally found the error, the modem stored the samples in the writing buffer although writing the I/Q samples to a file was disable, so basically the buffer was been fulfilled but never emptied (read), no checks if writing is enable to do that. |
The samples were saved to the writing buffer even when the writing was disable, causing an overflow.
|
We wait for confirmation for merging this PR after telco between @jsroldan and @alessandro-lucco |
|
Last week, Alexandro and I had a meeting to check the issues Alexandro found. We can confirm that are a few things that I need to have a look at. The sample rate is not choose correctly (which happens also in previous versions), and the writing of the samples to a file present some problems. |
Changelog: