TY - GEN
T1 - SIRCS: Slope-intercept-residual Compression by Correlation Sequencing for Multi-stream High Variation Data
AU - Ye, Zixin
AU - Hua, Wen
AU - Wang, Liwei
AU - Zhou, Xiaofang
N1 - Funding Information:
This work was partially carried out within the project ClustOverlap supported by Reunion Island Region-grant DIRED 20140704. The authors are also very grateful to the 3 anonymous reviewers for their valuable comments.
Publisher Copyright:
© Springer Nature Switzerland AG 2019.
PY - 2019
Y1 - 2019
N2 - Multi-stream data with high variation is ubiquitous in the modern network systems. With the development of telecommunication technologies, robust data compression techniques are urged to be developed. In this paper, we humbly introduce a novel technique specifically for high variation signal data: SIRCS, which applies linear regression model for slope, intercept and residual decomposition of the multi data stream and combines the advanced tree mapping techniques. SIRCS inherits the advantages from the existing grouping compression algorithms, like GAMPS. With the newly invented correlation sorting techniques: the correlation tree mapping, SIRCS can practically improve the compression ratio by 13% from the traditional clustering mapping scheme. The application of the linear model decomposition can further facilitate the improvement of the algorithm performance from the state-of-art algorithms, with the RMSE decrease 4% and the compression time dramatically drop compared to the GAMPS. With the wide range of the error tolerance from 1% to 27%, SIRCS performs consistently better than all evaluated state-of-art algorithms regarding compression efficiency and accuracy.
AB - Multi-stream data with high variation is ubiquitous in the modern network systems. With the development of telecommunication technologies, robust data compression techniques are urged to be developed. In this paper, we humbly introduce a novel technique specifically for high variation signal data: SIRCS, which applies linear regression model for slope, intercept and residual decomposition of the multi data stream and combines the advanced tree mapping techniques. SIRCS inherits the advantages from the existing grouping compression algorithms, like GAMPS. With the newly invented correlation sorting techniques: the correlation tree mapping, SIRCS can practically improve the compression ratio by 13% from the traditional clustering mapping scheme. The application of the linear model decomposition can further facilitate the improvement of the algorithm performance from the state-of-art algorithms, with the RMSE decrease 4% and the compression time dramatically drop compared to the GAMPS. With the wide range of the error tolerance from 1% to 27%, SIRCS performs consistently better than all evaluated state-of-art algorithms regarding compression efficiency and accuracy.
KW - Correlation mapping
KW - Error detection
KW - High variation data
KW - Linear regression model
KW - Multi-signal compression
UR - http://www.scopus.com/inward/record.url?scp=85065525515&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-18576-3_12
DO - 10.1007/978-3-030-18576-3_12
M3 - Conference article published in proceeding or book
AN - SCOPUS:85065525515
SN - 9783030185756
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 191
EP - 206
BT - Database Systems for Advanced Applications - 24th International Conference, DASFAA 2019, Proceedings
A2 - Natwichai, Juggapong
A2 - Li, Guoliang
A2 - Yang, Jun
A2 - Gama, Joao
A2 - Tong, Yongxin
PB - Springer Verlag
T2 - 24th International Conference on Database Systems for Advanced Applications, DASFAA 2019
Y2 - 22 April 2019 through 25 April 2019
ER -