This is Yu He from Nutaq. Welcome to Nutaq’s latest 5G Massive MIMO Testbed radio demonstration. In the previous video we had demonstrated a phase and gain calibration algorithm on our TitanMIMO platform. The phase and gain calibration video has demonstrated that the channel difference of TitanMIMO transceivers can be controlled within only 5 degrees in phase and 1% in gain. Today we’re going to demonstrate the uplink from 4 UEs to 16 antennas in our Massive MIMO Testbed. Let’s start with the UE.
As you can see, we have 2 Zynq SoC on the desk and each of them is equipped with Nutaq’s ADC9361 based FMC radio and able to playback an IQ stream. Each of those FMC radio cards can handle up to 2 UEs so we have 4 UEs here. As you can see there are many cables here. Some of them are used for 4 antennas of the 4 SISO UEs.
This is our 16 patch antenna for our basestation. For the moment it is on the uplink, so all the antennas are plugged into those TitanMIMOs 16×16 and are using all their receiver channels for reception. We have our clocking board here which provides reference clock for all the TitanMIMOs but also share two for UEs through the FMC radio on the Zynq SoC. The cables from the front panel are used to generate the trigger signals. Matlab itself work on the actual TitanMIMO sending a pulse to tell each of the UEs to transmit their own IQ stream at the same time. When the pulse goes on, the Zynq SoC is slaved to that pulse and transmit to those antennas here to the antenna array on the base station. At the same time, the same pulse is sent to each of those TitanMIMOs to start the recording process on their own on-board memories. Afterwards, the Matlab code will do the rest which means it will download the actual content of each of the on-board memories into Matlab and will perform the uplink to the transmission that was sent.
So that’s pretty much your own code. I will skip the actual FMC radio initialization of the basestation and instead will just simply start the call. You can see here the pulse is sent and right here you can see the recording. So the pulse was already sent and DDR3 has fueled in uplink UEs which is getting downloaded. Here you can see the channel for this example of UE #1 and UE#2 for antenna #1. And that the 4 constellations are recorded and received. This was done using a carrier frequency of 2.45 Ghz to escape the wifi interference but we have many meaningful over the air results at 2.4 and 5 GHz too. Actually the 5 GHz works better as there is less noise floor. As it’s scalable, you can see that I have one more UE right here. Since everything is easily changed through ethernet, which allows the Matlab code to extract all the data without any problems, in the next episode of this video series we are going to show you our validation of Argo and least square algorithm for reciprocity calibration over the air on the Nutaq Massive MIMO Testbed.
Thank you for watching this video. See you soon.
If you would like to get additional information on Nutaq’s 5G massive MIMO Testbed including documentation, specs or pricing, click here.