Capturing CSI Data using the AraMIMO Deployment

Platform: Skylark Wireless

Resources needed: Central Unit (CU), Distributed Unit (DU), Radio Unit (RU), and a Customer Premises Equipment (CPE)/User Equipment (UE).

Short Description: The experiment is designed to capture raw CSI data using AraMIMO deployment.

Detailed Description: AraMIMO consists of multiple CPEs connected to the Base Station (BS), a Skylark Wireless Faros V2 equipment. Using this experiment, we launch a container equipped with APIs and AraMIMO Repo. For the network diagram, refer to Monitoring AraMIMO Wireless Link Behavior.

Detailed Steps

Follow the steps of Monitoring AraMIMO Wireless Link Behavior up to Step 8 until the container is launched and the user has SSH access to the container.
Go to the root home directory.
```
# cd
```
Go to the aramimo directory.
```
# cd aramimo
```
The aramimo directory has multiple scripts to collect the measurement data. Run the csi script to start capturing the CSI data.
```
# ./csi_capture_script.py <time in seconds>
```
NOTE

The time can not be greater than 100 seconds. Each frame is 10ms, so 100s would capture CSI data of 10,000 frames.
As an example, enter the time as 10 seconds and execute the script. The script connects to the Skylark CU and start collecting the data.
Once the CSI capture is complete, the data will be saved to the /home/root/sklk_data/csi_data directory. The files can be saved locally to user’s computer via jumpbox for analysis. This script captures 3 files, i.e., csi.h5, csi_info.json, and hardware.json in a zip file. The file csi.h5 contains the actual CSI data while the other two files contain system-related data such as data rates, SNR, and system parameters.
To save the files to local PC, first copy it over to the jumpbox server using scp. On the jumpbox server run the command below.
```
# scp -r root@<floating_ip_of_container>:/home/root/sklk_data/csi_data ~/
```
The data from jumbox server can be copied to you local computer using the user’s ssh-key (assuming the data is saved user’s home directory on jumpbox server). On your local machine run the command below.
```
# scp -i <path_to_user_private_key> -r <username>@jbox.arawireless.org:~/csi_data .
```
The whole csi_data folder will be copied to the current directory of your host computer where the above command is run.
To explore the CSI data, we need to unzip the zip file. Go to the csi_data folder and run the following to untar the files.
```
cd csi_data
unzip csi_563000000.0_24000000.0_03_24_2024_02_40_52.zip
```

Here is an example script to read and explore the contents of the h5 file.

import h5py
csi_read = h5py.File('csi.h5', 'r')
for key in csi_read.keys():
   print(key)
   for subkey in csi_read[key].keys():
      print(csi_read[key][subkey])

Keys are ref_uplink, ref_downlink and ue_uplink. Subkeys are bin, ch, data, frame, hdl.
The hdl (or handle) is the user number. ch (or channel) is the channel of the hdl (each user has two channels, 0 and 1). frame is the frame number, bin is the frequency offset from the center frequency. data contains the actual IQ values for each hdl, ch and frame.
As an example, following line will print the IQ data for the uplink pilots in first frame, first hdl, and first ch.
```
print(list(csi_read['ue_uplink']['data'][0:512]))
```
The BS has a total of 42 antennas. The uplink pilot from each UE is received on all the antennas. The first 4 values in the 512 values above are for the reference antennas. There are two reference antennas, which also receive these pilots and the first 4 values (I and Q) in the above 512 values correspond to these reference antennas. The next 84 values are the I and Q values on the 42 BS antennas. The remaining values has no data becuase we only has 43 antennas, however the application support upto 256 antennas.

The following will print the first 100 hdl, ch, frame and bin respectively.

print(list(csi_read['ue_uplink']['hdl'][0:100]))
print(list(csi_read['ue_uplink']['ch'][0:100]))
print(list(csi_read['ue_uplink']['frame'][0:100]))
print(list(csi_read['ue_uplink']['bin'][0:100]))

To check which hdl correspond to which UE, run the following command in ara-sklk-cli.
```
get_connected_ues
```

Tip

If you want to perform weather measurements along with the RAN experiments, refer to ARA Weather APIs.