Illustrative example

In this example we will create an acquisition server on a Raspberry Pi and stream EEG data in real-time from them through Kafka using the OpenBCI-Stream library.

7ea3fe1fe8e64fa280a2bccb265d1016

Devices used:

  • Raspberry Pi 4 Model B 4GB RAM

  • Cyton Biosensing Board (8-channels)

  • OpenBCI WiFi Shield

  • Computer (with WiFi)

Conventions used:

  • The red window frames indicate that this window is “owned” by Raspberry through remote connection using SSH.

  • The IP for the Raspberry is 192.168.1.1

  • The IP for the OpenBCI WiFi Shield is 192.168.1.113

Acquisition server

The guide to create an acquisition server explain the process to set up the server over a Raspberry Pi, after finish and reboot the system the Raspberry will be an Acces Point, we must connect the OpenBCI WiFi Shield to this network as well as the main computer (where BCI Framework will be executed).

We must verify that the respective daemons are running correctly on Raspberry:

$ sudo systemctl status kafka zookeeper@kafka

b2844bfdedee49eeb6d48985f33a4b2b

$ sudo systemctl status stream_eeg stream_rpyc

31af3ed9ba3146d1baa4fa06b3e2e4b1

The system uses the NTP to synchronize clocks, so the Rasberry must have a wired connection to the internet to synchronize their own clock, to ensure this we can verify the connection and restart the daemon:

$ sudo systemctl restart ntpd

$ nptq -pn

80bfc39cac42490a91f9599f80135c02

After a while, the clock will be synchronized (notice the * in the server 186.30.58.181)

038780de3ae146cfa978f84737114c8e

We can verify the status of the WiFi Shield with the command:

$ curl -s http:/192.168.1.113/board | jq

f1190883137b42828c2588afe49f2253

The above commands being executed through a SSH connection, also can be done by connecting a monitor and a keyboard to the Raspberry.

Configure montage

A simple montage 8 channels:

0bf5e20961014fb69c55f659ed3770ed

Configuration and connection with OpenBCI and then start the stream

Connect with the Raspberry that is running under the IP 192.168.1.1 and the WiFi Shield on 192.168.1.113. The sample frequency of 1000 samples per second with a transmission of packages of 100 samples.

2d704d4fe45c44b3a0595778c20b5a2c

Impedances

Once the streamming is started the impedances can be displayed from the Montages tab

6d8a9ef90ee1497b9db0e55e4f8fcabd

Raw EEG and topoplot

672d532318034f5aa6d02d8284d78bc1

P300 speller

7c8af367304c45f2a1e0042b7ba03da9

EEG records reading

[1]:
from openbci_stream.utils.hdf5 import HDF5Reader

filename = "record-04_20_21-13_58_25.h5"
file = HDF5Reader(filename)
print(file)
file.close()
==================================================
record-04_20_21-13_58_25.h5
2021-04-20 13:58:25.717151
==================================================
MARKERS: ['NO-TARGET', 'TARGET']
SAMPLE_RATE: 1000
STREAMING_SAMPLE_RATE: 100
DATETIME: 1618945105.717151
MONTAGE: standard_1020
CHANNELS: {1: 'Fp1', 2: 'Fp2', 3: 'T3', 4: 'C3', 5: 'C4', 6: 'T4', 7: 'O1', 8: 'O2'}
START-OFFSET: 346.3931083679199
SHAPE: [8, 68900]
END-OFFSET: 377.5792121887207
==================================================
[2]:
with HDF5Reader(filename) as file:
    data, classes = file.get_data(tmin=0, duration=0.125)

data.shape, classes.shape
[2]:
((65, 8, 125), (65,))