🚀 go-pugleaf

From article <1992Oct27.135715.18417@m.cs.uiuc.edu>,
by sidell@swine.cs.uiuc.edu (Jeff Sidell):
> I'm wondering how incoming satellite data is usually
> represented?

Until recently, data compression was not widely used in space-to-ground
data transmission.  Part of the problem is that, until recently, the
on board processors used in space were extremely limited in computing
power.  Remember that microprocessors used in space must be radiation
hard, launch certified, and they must usually conform to stringent limits
on power consumption.  Looking at what's up in orbit today, you're
lucky if you find a processor as powerful as the 8088.

With that 8088 I just described, it is common for the CPU to be the
bottleneck on the path from instrument to the ground.  The downlink
may run at 10000 baud, the the CPU can't compress data that quickly.

The exception is for deep space probes, where downlink data rates are
very low.  The current mission to Jupiter, with its disabled antenna
for the primary downlink, has pushed NASA into the use of compression
techniques as never before, all the work being done after launch and
uploaded into the spacecraft computers.

One of the University of Iowa's newer instruments (on the ISTP Polar
mission) has just enough added hardware that it can manage to use vecor
quantization for the images it sends down.  I investigated the
possibility of lossless compression on those images using my splay
tree based codes, and it was just barely feasible also.  The images
are inherently noisy at the source (noise like snow on a TV screen),
so the compression choice is one between poor but accurate compression
or good compression that smooths out the noise but may also obscure
the features being studied.
				Doug Jones
				jones@cs.uiowa.edu