Author Archives: Wenhua Chen

R&S Cup On-line DPD Competition Winners

In the registration period, seven teams were registered to the competition. During a dozen days of debugging, each team used our remote lab and functions to debug its DPD algorithm. And Finally, six teams successfully participated in the competition. Here is a photo of the remote lab:
After fierce competition, the champion and runner-up were decided.
Thank you again for every team’s active participation and suggestions during this period, and also thank Rohde-Schwarz for its strong support for this competition.

1.Xiangjie Xia, Chenxing Li and Qiannan Zhang, University of Electronic Science and Technology of China, China, score 116.6, win a prize of $600.

2.Thomas Ackermann, Julian Potschka, Tim Maiwald, Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany, score 76.2, win a prize of $400.

We congratulate the winners and thank all of you for your participation! 

Detail rules of RS Cup DPD Competiton

Rohde & Schwarz Cup On-line DPD Competition

Rules in Detail

  1. The system will be closed at 8:00 PM (UTC +8:00), May 24.
  2. You should make video call with us 3 minutes before you start. We will open the remote lab for you, while other teams are not accessible.
  3. Each team has a 30-minute time slot, including not more than 25 minutes for training, while the remaining minutes are for testing.
  4. If the upload speed is slow (less than 100kbps), the test time can be extended appropriately. There will be no scores if you don’t finish in time.
  5. We will send you an encrypted document, which includes 3 groups of test data:

test_signal_1a.txt  —  The first test signal of channel 1.

test_signal_2a.txt  —  The first test signal of channel 2.

test_signal_1b.txt  —  The second test signal of channel 1.

test_signal_2b.txt  —  The second test signal of channel 2.

test_signal_1c.txt  —  The third test signal of channel 1.

test_signal_2c.txt  —  The third test signal of channel 2.

The format of the test data and the training data we sent is exactly the same.

  1. When you finish the training, we will tell you the password. Then you need to generate the DPD signals with the ‘test_signal_1a.txt’ and ‘test_signal_2a.txt’ and upload them via the sender function. We will save the results file and tell you the score, then you can do the same procedures with ‘b’ and ‘c’ by turns.
  2. During the testing period, no more training is allowed.
  3. The final score is the average of the three tests, which will be published on the RS cup page of the DPD site:

Time setting and other details of R&S Cup online DPD competition

The R&S Cup online DPD competition will be held at May 24 and 25. The whole competition time lasts for 36 hours, from 10:00 AM May 24 to 10:00 PM May 25 (UTC+8:00). Each team will have half an hour for the competition, and we need you to send us the available time slots before 0:00 May 21 (UTC+8:00). We will post the schedule of the competition on the website and by email.

During the 36 hours of competition time, the remote lab will be opened exclusively to the team in progress by the schedule. At that time, other teams will not be able to use the remote lab.

Notice that the signals of 2 channels should be a single-carrier signal ( for ch1) and a three-carrier signal ( for ch2) .The scoring rule requires each team to make better ACPR with higher power and coupling coefficient.

Intelligent Microwave and Circuit System Lab of Tsinghua University

Rohde & Schwarz Cup On-line DPD Competition

The R&S Cup on-line DPD competition is going to be held for world-wide participants!

  • WHAT: On-line DPD competition to linearize a 2-channel MIMO transmitter with crosstalk. The MATLAB scripts, instruments platform and evaluation rules are the same with IMS2019 SDC DPD competition. Please visit the DPD competition website ( for detailed rules.
  • Fig. 1 Block diagram of test bench.
  • WHEN: Between 23th and 25th of May 2019. The exact date will be determined according to the available time of participants.
  • WHO CAN PARTICIPANT: Anyone interested in the DPD linearization technique.
  • PRIZES: Sponsored by Rohde & Schwarz
    • 1st team: 600 USD
    • 2nd team: 400 USD
    • 3th team: 200 USD

The final scores will be posted on:

  1. Send an e-mail containing the university, name of the team members, their contact details (e-mail preferred) and a team photo to and to with the subject line “Registration: R&S Cup On-line DPD Competition” before 13th May, 2019 (Beijing is in time zone CST, UTC + 8). For the IMS2019 SDC DPD teams, only a simple reply to inform us whether participating in this competition is enough.
  2. The skype video call will be established with each team on the competition day in case of you need assistance.
  3. The exchange of the test data will be accomplished through FTP and MATLAB scripts ‘ims2019_THULabSender.p’ and ‘ims2019_THULabReceiver.p’, so make sure the FTP connection will not be blocked by the PC firewall. Visit for more details.
  4. Each team will have 30 minutes to train their DPD algorithm. In your exclusive time slot, other teams cannot have access to our server. After training, the TEST SIGNAL will be given to the participants to generate the predistorted signal through the linearization algorithm, and upload to our platform for evaluation. No more algorithm training is allowed at this moment.
  5. Each team will have 3 times to send the predistorted signal and calculate the SCOREs. The final score is the average of these three scores.Scoring rules:
    • ACPR values below -45 dBc improve the score.
    • NMSE values below -36 dB improve the score.
    • Average output powers in excess of 36 dBm improve the score while below 36 dBm decrease the score. The output powers are considered only when the ACPRs reach -45 dBc.
    • The value of the coupling strength is added in the formula. The coupling strength is considered only when the ACPRs reach -45 dBc.

    Score formula:






At the day of the competition at the IMS venue in Boston, each test will be performed 3 times. The average of these three realizations will be your final SCORE.

[UPDATED]About FTP connection

We are so sorry to delay the opening of the server, because of our campus gateway.

So we have built a server on public network. The IP of the server is different now. For security reasons, we decided to keep the server’s ID secret. So we will send you the two renewed functions by email.

If you are unable to connect to the FTP server, we recommend that you change the following settings:

Internet information service in control panel:  ‘FTP service’, ‘FTP extensions’ and ‘TFTP client’ is open;

Internet Explorer browser  is not off-line working;

Firewall allows ‘windows resource managers’ and FTP;

Internet Options Advanced Setting: cancel the ’Using passive FTP‘ option.

These settings should solve the problem of disconnection. Then the ‘FileZilla’ Client can be completely replaced by the two functions. Click here to view the Chinese version of the settings. If you have any problems about FTP connect, please contact Chason by



DETAILS about uploading the experimental file

When you propose to upload your signal after predistortion, you can use the ‘ims2019_THULabSender.p’ function. This function could check the size and type of your data, save it and upload it to the FTP server. You can call it like this:


such as:

ims2019_THULabSender(-20,[1 1],inputsignal,’ims2019_teamA’,’*****’);

The meanings of parameters are:

The first parameter, coupling factor(cf), is variable, between -30 to -10. Of course coupling factort will make your DPD process harder as you change it from -30 to -10. But in turns, you will get a higher score if you do well. If the paremeter you provide is out of range, it will be set to the boundary value.

The second, is the powers of your 2 signals, in dBm. It is normalized by a mapping tabel, so you can make it from 0 to 1. Here shows the mapping:







The third data, whose type is double, is the same size as the input data we give you before DPD.It’s a 2-channel signal,the 4 columns are:[I of channel 1,Q of channel 1,I of channel 2,Q of channel 2], where the ‘I’ means in-phase signal, ‘Q’ means quadrature signal.

You can also package your own data, and drop it to FTP server via FileZilla Client. In this case, please contact us for further information.


These are all the things you need to pay attention to when uploading files.

Once you upload your .mat file, we will start processing, which probably takes 1 minute. After 1 or 2 minutes, you could use the ‘ims2019_THULabReceiver.p’ function to get the file from your FTP folder.You can call it like this:

[score,pms,output]= ims2019_THULabReceiver(‘ims2019_teamA’,’******’,’Foldername’);

where the ‘Folername’ is the name of the folder where you want to store the result file under the current path. If the folder with that name does not exist, matlab automatically creates one.Also, the pms (parameters) is a vector:

[channel1 power,channel2 power,coupling factor,channel1 ACPR1, channel1 ACPR2, channel2 ACPR1, channel2 ACPR2]

You could also login the FileZilla client. The file, which we will send back in your team’s folder, is a .mat file, named [results_RX.mat], with 1*3 cell in it:

{score,[channel1 power,channel2 power,coupling factor,channel1 ACPR1, channel1 ACPR2, channel2 ACPR1, channel2 ACPR2],output}

The data obtained by the two methods are the same. But if you download it manually instead of using functions, you also need to load the data and extract the corresponding data from the cell.

These are all the notes during uploading. It is worth emphasizing again that you should download the results in time, otherwise, you may overwrite it after the next upload.


Any question please contact Chason by

Best regards.

The IMS2019 DPD SDC team

NOTICE: how to visit the remote lab

Hello, participants of IMS2019 DPD SDC. The remote experimental platform will be aviliable on 25th, April. The communication between your PC and our FTP server is set by FileZilla. So, please download and install ‘Filezllia client'(a kind of FTP software) on .

We will provide every team your ID and default password of FTP client. The platform mostly bases on matlab, you can upload it with the ‘ims2019_THULabSender.p‘ function we provide. 

The steps of the platform are: once you submit your data via ‘ims2019_THULabSender.p‘ function, we will recognize the data first and put it into our experimental platform to do the processing.Then we will send it back to matlab for subsequent processing and scoring, and return a .mat file with a cell in it, whose form is  {[score],[signal power1,signal power2,coupling factor],[outdata]}.

Or you can upload the .mat file manually:

Each team will have the permission to upload the data file into one same folder named “upload”, so you can drop it into the ‘upload’ folder via the FileZilla Client. Make sure you will download your results in time from your exclusive folder, such as ‘ims2019_teamA’. Otherwise, it could be covered by the result of next submission (to protect the server’s storage room).

Now you should download and configure FileZilla client and make it work. You can use its network configuration guide , it’s quite helping. (We met some problems in this process, make sure your firewall is open for FileZilla, and make sure your PC have a public net IP).

If you meet any problems during the configuration, please contact:

Best regards

The IMS2019 DPD SDC team