[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Re: DSP software on Sunsat SO-35

Hi Chris,

 > Am I correct in thinking that:
 > 1.  a complete CCD image is stored as a large file on-board the
 > spacecraft 
 > 2.  The complete *.ccd file (approx 1 to 2 MB) is then downloaded
 > using 9600bps 
 > 3.  There is a spare 386 on-board

There is 64 MB of storage space on board Sunsat.  The 24 bit images 
(near infra-red, red, green) fills approximately 50 MB for a typical 
image.  Delta compression on the DSP reduce every pixel to 15 bits, 
resulting in about 30 MB to be downloaded for every image.  3.5 to 4 
MB of GPS data must be downloaded for JPL every week, detailed WOD 
files are collected at regular intervals, the transponder is 
scheduled for weekends, and some time is set aside for all kinds of 
experiments.  All of this happens at 9k6.  Not much of the week 
remains for downloading images!

We are still working on running the engineering model of the 386 
system in protected mode.  Code intended for the Sunsat works fine on 
an evaluation module for the 386EX processor.  However, many low 
level drivers must still be developed for the satellite.

 > Is it possible for the 386 to access the file, process it, and
 > then hand it back to the main processor for transmission at
 > 9600bps? 

Yes, both processors has access to all of the satellite's subsystems.

 > If so, I have an idea.  The raw *.ccd file could be compressed
 > using the ZIP algorithm (instead of an image-specific compression
 > algorithm).  The .zip files could then be downloaded instead of the
 > .ccd files. 
 > ZIP compression could conceivably achieve an 80% compression
 > ratio.  That is, a 1MB .ccd file compresses to 200kB, meaning the
 > same amount of data can be transferred 5 times faster, or the
 > equivalent of sending the raw data over a downlink of 48kbps! 

Experiments indicate that the ZIP algorithm achieves closer to a 40% 
compression ratio.  Compared to the low overhead delta compression 
running on the DSP it is a bit dissapointing.  The delta compression 
algorithm is of course tailored to image characteristics.

 > Presumably you have an engineering model that this can be tested
 > on?  I note that you're short of hands to do the work...  I could
 > look at creating the code to do the ZIP compression on the 386
 > processor, however I'll need assistance from you in terms of better
 > understanding the hardware environment, etc. 

Thank you for the proposition.  I guess in future (when we've got the 
386 running to its full potential) we could make use of such offers.

Johann, ZR1CBC

   JG Lochner  ESL, Universiteit van Stellenbosch
   e-pos:      lochner@ing.sun.ac.za
   webtuiste:  http://esl.ee.sun.ac.za/~lochner

Via the amsat-bb mailing list at AMSAT.ORG courtesy of AMSAT-NA.
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org