Multichannel Waveform Compression Algorithm on sbRIO? Any idea?

[jiangliang]

Dear All:

     I found I have to deal with 128 channel data comes at about 50Hz or so, I already know the signal itself changes relatively slow, I would like to use 9651 to capture and store as much data as possible, the SDCARD or USB is not connected at current product, so I have to think about algorithm, which is not my area...

     So I try to find information about waveform compression, but doesn't find much of useful information about it, does any one have any idea about it?

     There a many question in my mind, like:

     1. Say that there is a good algorithm around, can it easily port to RT linux?

     2. Is Zynq powerful enough to deal that amount of data realtime?

     3. What to expect on the compression rate? would it reach 10:1? or maybe only about 1.2~1.5:1?

     4. Is it wise to consider audio compression algorithm like mp3,ogg, or the lossless ones, like flac?

     5. I even considered the ADPCM, seems it can save some amount of space and can easily use FPGA to coding. are there any downside about ADPCM?

     Thanks~

[Deagle]

Jiangliang,

It looks like there's a compression-related community example for LabVIEW FPGA which may be a good starting point. I recommend trying to implement this on FPGA since you'll be sampling 128 channels. If you do find an open-source software compression library that you'd like to try out on NI Linux RT, there's some good getting-started points on compiling for NI Linux RT as well as an NI Linux RT community that can help with issues specific to Linux RT.

Regarding the compression rate, it's difficult to speculate on what it will be since it depends highly on the characteristics of signal. I believe the existing audio compression schemes are specifically optimized for audio waveforms and may not work well for your signal. Is adding a USB or SD storage device possible?

Bryan

[jiangliang]

Thank you very much, I'll check the link you provide.