Skip to content

monteiro-diogo/AST-Data-Compression

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

19 Commits
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

AST Data Compression: Lossless & Lossy Evaluation

This repository contains the source code, datasets, and final report for my Summer Internship project at Active Space Technologies.

The project focuses on evaluating and implementing both lossless and lossy data compression algorithms, targeting telemetry applications where reducing storage requirements and accelerating data transmission are critical.

πŸš€ Project Overview

The study evaluates the performance, compression ratios, and computational efficiency of various algorithms, comparing high-level (Python) and low-level (C++) implementations.

  • Lossless Compression: Evaluation of Gzip, Brotli, Zstd, BZ2, LZMA, and LZ4.
  • Lossy Compression: Evaluation of the Opus codec for audio/analog signal data.
  • Implementation Comparison: Benchmarking execution time and memory behavior between C++ and Python.

πŸ“ Repository Structure

  • /Lossless - Contains the Python and C++ source code for evaluating lossless algorithms.
  • /Lossy - Contains the C++ Opus encoder implementation and MATLAB scripts for objective signal analysis.
  • AST_Final_Report.pdf - The comprehensive final internship report detailing methodology, results, and conclusions.

πŸ“Š Key Findings

  1. Optimal Lossless Codec: Zstd (at low to moderate levels) provides the best practical trade-off between compression ratio, runtime, and resource consumption.
  2. Optimal Lossy Codec: Opus at 128 kbps delivers near-transparent perceptual quality for complex waveforms while yielding a ~91% file-size reduction.
  3. C++ vs. Python: C++ implementations reduced compression time by approximately 18.2% and exhibited highly stable, predictable memory behavior, making it the superior choice for resource-constrained telemetry systems.

About

This repository explores data compression techniques, both lossless and lossy.

Resources

License

Stars

Watchers

Forks

Contributors