GPU-Based Lossless Compression of Aurora Spectral Data using Online DPCM

Jiaojiao Li,Jiaji Wu,Gwanggil Jeon

doi:10.3390/rs11141635

Abstract

It is well known that aurorae have very high research value, but the data volume of aurora spectral data is very large, which brings great challenges to storage and transmission. To alleviate this problem, compression of aurora spectral data is indispensable. This paper presents a parallel Compute Unified Device Architecture (CUDA) implementation of the prediction-based online Differential Pulse Code Modulation (DPCM) method for the lossless compression of the aurora spectral data. Two improvements are proposed to improve the compression performance of the online DPCM method. One is on the computing of the prediction coefficients, and the other is on the encoding of the residual. In the CUDA implementation, we proposed a decomposition method for the matrix multiplication to avoid redundant data accesses and calculations. In addition, the CUDA implementation is optimized with a multi-stream technique and multi-graphics processing unit (GPU) technique, respectively. Finally, the average compression time of an aurora spectral image reaches about 0.06 s, which is much less than the 15 s aurora spectral data acquisition time interval and can save a lot of time for transmission and other subsequent tasks.

Highlights

Aurorae are considered some of the most beautiful wonders in nature, which are colorful and constantly changing
There are two main reasons for this: one is that aurora spectral images have very high scientific value and their acquisition is quite expensive, so aurora spectral images have long-term preservation value; the other is that minor information loss may cause large errors in some applications [2], no information loss is allowed in aurora spectral image compression
We proposed two improvements to improve the compression performance of the original online Differential Pulse Code Modulation (DPCM) method [9]: one is on the establishment of the linear system of equations when computing the prediction coefficients, and the other is on the encoding of the residual

Summary

Introduction

Aurorae are considered some of the most beautiful wonders in nature, which are colorful and constantly changing. We proposed two improvements to improve the compression performance of the original online DPCM method [9]: one is on the establishment of the linear system of equations when computing the prediction coefficients, and the other is on the encoding of the residual. In our online DPCM aurora spectral data compression scheme, the computing of the predicted value of each pixel is independent, so that they can be calculated in parallel using GPUs. the prediction coefficients for each pixel are calculated using the least square method (LSM), and the main operations of LSM are matrix multiplication and matrix inversion, both of which are well-suited for acceleration with GPUs. it is reasonable and suitable to accelerate the online DPCM algorithm in parallel with GPUs. Many researchers have utilized GPUs to speed up their algorithms. The enTchordoeugmheethxopderims tehnetsrawnegfeocuonddetrh[a8t].thSeinccoemwpereasrseiounspinegrfothrme aonclienoefDthPeCoMnlimneetDhPoCd,Mthaelgoonrliythm candbaetaimthpartonveeeddbsytotwboe eimncpordoevdemanendttsr.aOnsnme itsteodn tohetheestdabecliosdhemreins tthoef trheesildinueaal.r system of equations when computing the prediction coefficients, and the other is on the encoding of the residual

Our Improvement to the Original Online DPCM Method

The Improvement on the Encoding of the Residual

Some Basic Concepts of GPU Programming