Real time aquisition of EEGdata through TCP server

Post Reply
gangadhar
Posts: 2
Joined: Mon Jul 23, 2007 5:08 pm
Location: Switzerland

Real time aquisition of EEGdata through TCP server

Post by gangadhar »

Hello,

We have tried to acquire EEG data with TCP server of ActiveVIEW. We have observed a huge delays more than 100ms (at least 10% of the time) in reading a data of the size of the TCP array. Is it normal?

We have done the above experiment with following conditions.
1) Sampling rate set to 512 Hz.
2) Number of electrodes considered 32+1 or 64+1 (both the cases tested)
3) Reading of the TCP IP data is done with Instrument control toolbox of MATLAB on another computer
3) We tried increasing the input buffer size in the client side
4) The delay in reading is computed using psychtool box(MATLAB)
5) The Laptop connected to BioSemi system is with P4, 2Ghz. And the client is P4, 512RAM, 2Gz.
6) We are using ActiveVIEW Laptop 6.03 version
7) Both the computers are directly connected and isolated from the rest of the network.

Will all the above settings, We have observed a consistent delay in reading the packets.

Apart from the delays, we have observed a flicker in the raster scan-line once as soon as we establish the TCPIP connection, and continues to flicker until the connection is closed. Is it normal?

Can we reduce delays

Thanks in advance,

regards,
Gangadhar, IDIAP Research Institute.

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Post by Coen »

It is quite normal to see delays in the order of 100 ms between the events in the real world and the display on the client PC. Actually, this is what you expect: data acquisition via USB on a Windows PC causes delays of a few tens of msec, while the TCP/IP file transfer between two Windows PCs add another couple of tens or even hundreds of ms (depending on PC and network speed). Your negative qualification "huge" for the delays seems a bit premature, I would recommend to be more hesitant in using such strong qualifications for matters outside your expertise.

Best regards, Coen (BioSemi).

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Post by Coen »

For feedback applications, I recommend to eliminate the TCP/IP transfer from your loop. To keep delays to a minimum, is best to work in LabVIEW directly on the acquisition PC. Learning the LabVIEW language and working on the basis of the existing ActiView code is a quicker route to success than developing a new acquisition program form scratch.

In your case I would recommend to start by purchasing LabVIEW 8.2, either the regular "full development" version (http://www.ni.com/labview/), or the (low-cost) student edition (http://www.ni.com/labviewse/select.htm). Next step is to familiarize yourself with the LabVIEW principles (I recommend the book http://www.amazon.com/LabVIEW-Everyone- ... 478&sr=8-1)
). Continue with studying the stripped down version of ActiView available from http://www.biosemi.com/download/ActiView603-Light.zip
and then you can continue with adding your own code to the full ActiView version. Several customers have followed this route successfully, Please don't hesitate to contact me on any LabVIEW programming issues.

A LabVIEW course is available from http://cnx.org/content/col10241/latest/ . Further help may be
available on http://forums.lavag.org/ and http://forums.ni.com/ni/board?board.id=170, and of course on this forum.

Best regards, Coen (BioSemi)

gangadhar
Posts: 2
Joined: Mon Jul 23, 2007 5:08 pm
Location: Switzerland

thanks for the clarifications

Post by gangadhar »

Thanks for the clarifications... I will contact the form and you for the help in programing.

Post Reply