Spacecraft data compression, e.g. New Horizons Pluto mission

[SolidComp]

Hi all -- Because of free-space path loss over the distance between Earth and Pluto, the New Horizons exploration spacecraft was only able to send data at a rate of 2 kbps.

Does anyone know what kind of compression NASA uses for New Horizons and other spacecraft?

It seems to mostly be imaging data, like the images of Pluto and its moon Charon. I'm curious if they're using something new and state of the art, or something old, and how it might be improved.

It sounds like they can upgrade their data compression codec in place:

Another exciting possibility is that we can dramatically augment New Horizons' capabilities by uploading new observing and onboard data-reduction software once the spacecraft's flyby software is no longer needed.

(From one of the mission webpages at John Hopkins University.)

Here's some data if you're interested, though it's post-decompression I think:

https://pdssbn.astro.umd.edu/holding...0/dataset.html

https://pds.jpl.nasa.gov/

[Shelwien]

There's a book like this: https://www.amazon.com/Satellite-Dat.../dp/B007EM0CJ6
Their tech doesn't look that advanced or otherwise useful really, but its easy to understand, since they have to save energy first, they can't run cmix there.

Contents
1 Development of On-Board Data Compression Technology at Canadian Space Agency
2 CNES Studies for On-Board Compression of High-Resolution Satellite Images
3 Low-Complexity Approaches for Lossless and Near-Lossless Hyperspectral Image Compression
4 FPGA Design of Listless SPIHT for Onboard Image Compression
5 Outlier-Resilient Entropy Coding
6 Quality Issues for Compression of Hyperspectral Imagery Through Spectrally Adaptive DPCM 
7 Ultraspectral Sounder Data Compression by the Prediction-Based Lower Triangular Transform 
8 Lookup-Table Based Hyperspectral Data Compression
9 Multiplierless Reversible Integer TDLT/KLT for Lossy-to-Lossless Hyperspectral Image Compression
10 Divide-and-Conquer Decorrelation for Hyperspectral Data Compression
11 Hyperspectral Image Compression Using Segmented Principal Component Analysis
12 Fast Precomputed Vector Quantization with Optimal Bit Allocation for Lossless Compression of Ultraspectral Sounder Data
13 Effects of Lossy Compression on Hyperspectral Classification
14 Projection Pursuit-Based Dimensionality Reduction for Hyperspectral Analysis

[stbrumme]

According to Wikipedia, the main computer is a 12 MHz Mongoose V CPU (that's a MIPS chip, see https://en.wikipedia.org/wiki/New_Ho...light_computer ).
Image processing quite likely runs on dedicated hardware - at least partially - but chances are that it's slow as well.

I think efficient error correction poses a much bigger problem than compression ratio.