Fractal Resampling Time Series Compression

Ref-Nr: TDO0170

Technology abstract

ESA is offering a compression method called Fractal Resampling, which can be used to compress time series data. This algorithm has been designed to be run on spacecraft and planetary probes, and thus a special effort has been made to devise a solution with low computational complexity. 

 

Technology Description

The fractal resampling takes the original time series data samples and produces a set of fewer samples, not necessarily at regular intervals, that resembles the original time series while offering a configurable maximum error guarantee. With the proposed fractal resampling technique, it is possible to sample data at higher rates on-board and downlink only the data samples necessary to reconstruct the original signal with the desired level of accuracy, with significant data volume reductions.

Fractal Resampling was developed in 2012 by members of the ESOC Advanced Mission Concepts and Technologies team, Josツ-Antonio Martᄀnez-Heras, Tiago Francisco and Alessandro Donati.

Innovations & Advantages

  • Compressing time series, while keeping all the features and their characteristics
  • Inspired by 2D mid-point displacement
  • Offers mathematically proven guarantees
  • Dramatically reduces data volume, enables better user experience and reduces data intensive processing time

Further Information

The fractal resampling takes the original time series data samples and produces a set of fewer samples, not necessarily at regular intervals, that resembles the original time series while offering a maximum error guarantee. This means that all interesting information (e.g. peaks, short lived events) can be retained.
The fractal resampling technique works by discarding data that are not interesting. The first step is to define what ?interesting? means. Our definition of interesting in this context is the data that are needed to take decisions. In this sense, if there is a need to plot a parameter in order to take an informed decision and we could take exactly the same decision using much less data, the data that is missing is not interesting.
 
 
 
 

Current and Potential Domains of Application

Currently the technique is being used as to compress data transmitted from the Rosetta spacecraft now orbiting Comet 67P/Churyumov-Gerasimenko.

The same approach can be used in any field where telemetry data are present, and thus there is a broad range of potential applications including automotive, aerospace, factories and all theirs subsystems, autonomous devices/vehicles and medical monitoring.