It's amazing just how much of a loaded word "telemetry" has become in the past few years --- 10 years ago it'd just remind me of space missions and such, like in the article, but now I associate the word more with pervasive surveillance and privacy invasion.
Interplanetary telemetry compression
11–16 of 16 posts
Re: Interplanetary telemetry compression
#12Have you considered existing scientific columnar data storage formats, like HDF5 or Parquet?
Their main advantage is that they have good, mature implementations in a variety of languages, which would be handy if you ever find Javascript to be too slow.
Re: Interplanetary telemetry compression
#13Sounds like a use case for zstd with custom dictionary.
Don't forget the paq family, especially because of their tunability. You could probably replace the column representation with a context model that just classifies based on the json value or so, and context mixing, at least in the pre-zpaq days, does the rest there. Also check delta encoding.
If you are able to spare the JSON format the way it is, or at least don't have to provide bit-exact representation, just data-exact ones, you might want to check out if you can make a proper json transform into a binary format suitable for this encoding, possibly checking if you can code a float represented inside a string with a printf-style encoder for the decoding, and then just store this as a datatype "single prevision, standard posix format", or "single prevision, this format, followed by a length-prefixed format string". You should make sure that some context models know where you are right now, e.g., if it's a format string, a content string, a length-prefix, etc. Context Mixing really is amazing.
Re: Interplanetary telemetry compression
#14Have you considered existing scientific columnar data storage formats, like HDF5 or Parquet? Their main advantage is that they have good, mature implementations in a variety of languages, which would be handy if you ever find Javascript to be too slow.
Sure, I looked at parquet and HDF5. The compression is much better with iptm for the kind of data I am interested in, and the algorithm is also much simpler. It's less than 500 LOC in total.
The reason I wrote this in typescript is that it is nice for prototyping when working with JSON data. If it ever turns out too slow, I would write a rust version. Should not take more than a few days.
I have not profiled it, but I would guess that the most time is spent in the zlib deflate implementation, which is of course already a C library, so I would not expect miracles from a rewrite in rust.
Re: Interplanetary telemetry compression
#15Have you considered existing scientific columnar data storage formats, like HDF5 or Parquet? Their main advantage is that they have good, mature implementations in a variety of languages, which would be handy if you ever find Javascript to be too slow.
I developed an archive system for our spacecraft TM/TC using HDF5 with blosc-lz compression for speed reasons (and I plan to move to Zstd in the future). One of the main issue I have seen in the industry is the difficulty to upgrade the hardware so we have to design something working great on regular hard drives and with minimal RAM requirements.
While HDF5 is good when you manage the whole system, it is always tricky for sharing data with other people so we are also using SQLite. We lose the compression but it is very easy to share and people are more familiar with it.
Re: Interplanetary telemetry compression
#16Have you considered existing scientific columnar data storage formats, like HDF5 or Parquet? Their main advantage is that they have good, mature implementations in a variety of languages, which would be handy if you ever find Javascript to be too slow.
I developed an archive system for our spacecraft TM/TC using HDF5 with blosc-lz compression for speed reasons (and I plan to move to Zstd in the future). One of the main issue I have seen in the industry is the difficulty to upgrade the hardware so we have to design something working great on regular hard drives and with minimal RAM requirements. While HDF5 is good when you manage the whole system, it is always trick…
For the export at GSOC we had a streaming REST interface based on akka-streams. It is used to get the data into other systems based on spark for more complex analysis.
Of course you also lose the compression, but the target system is typically only interested in a small subset of the data.