Time series matrix factorization prediction of internet traffic matrices

Yunlong Song; Min Liu; Shaojie Tang; Xufei Mao

doi:10.1109/LCN.2012.6423629

Time series matrix factorization prediction of internet traffic matrices

Yunlong Song^*, Min Liu, Shaojie Tang, Xufei Mao

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

9 Citations (Scopus)

Abstract

Traffic matrices (TMs) are very important for traffic engineering and if they can be predicted, the network operations can be made beforehand. However, existing prediction methods are neither accurate nor efficient in practice. In this paper, we utilize the spatio-temporal property and low rank nature to directly predict the total TMs. The problem is that conventional matrix interpolation only works well when elements are missing uniformly and randomly. But in the case of TMs prediction, an entire part of the matrix is unknown. To solve this problem, we utilize some essential properties of TMs and add the time series forecasting into the matrix interpolation. We analyze our algorithm and evaluate its performance. The experiment result shows that our method can predict TMs under an NMAE of 30% in most cases, even predicting all the elements of next 3 weeks.

Original language	English
Title of host publication	Proceedings of the 37th Annual IEEE Conference on Local Computer Networks
Subtitle of host publication	LCN 2012
Editors	Tom Pfeifer, Anura Jayasumana, Damla Turgut
Place of Publication	Piscataway, NJ
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	284-287
Number of pages	4
ISBN (Print)	9781467315647
DOIs	https://doi.org/10.1109/LCN.2012.6423629
Publication status	Published - 2012
Externally published	Yes
Event	37th Annual IEEE Conference on Local Computer Networks, LCN 2012 - Clearwater, FL, United States Duration: 22 Oct 2012 → 25 Oct 2012

Other

Other	37th Annual IEEE Conference on Local Computer Networks, LCN 2012
Country/Territory	United States
City	Clearwater, FL
Period	22/10/12 → 25/10/12

Keywords

traffic matrices prediction
time series forecasting
matrix interpolation

Access to Document

10.1109/LCN.2012.6423629

Cite this

@inproceedings{507b08be0d114c2f9d0dcd522e99b15b,

title = "Time series matrix factorization prediction of internet traffic matrices",

abstract = "Traffic matrices (TMs) are very important for traffic engineering and if they can be predicted, the network operations can be made beforehand. However, existing prediction methods are neither accurate nor efficient in practice. In this paper, we utilize the spatio-temporal property and low rank nature to directly predict the total TMs. The problem is that conventional matrix interpolation only works well when elements are missing uniformly and randomly. But in the case of TMs prediction, an entire part of the matrix is unknown. To solve this problem, we utilize some essential properties of TMs and add the time series forecasting into the matrix interpolation. We analyze our algorithm and evaluate its performance. The experiment result shows that our method can predict TMs under an NMAE of 30% in most cases, even predicting all the elements of next 3 weeks.",

keywords = "traffic matrices prediction, time series forecasting, matrix interpolation",

author = "Yunlong Song and Min Liu and Shaojie Tang and Xufei Mao",

year = "2012",

doi = "10.1109/LCN.2012.6423629",

language = "English",

isbn = "9781467315647",

pages = "284--287",

editor = "Pfeifer, {Tom } and Jayasumana, {Anura } and Damla Turgut",

booktitle = "Proceedings of the 37th Annual IEEE Conference on Local Computer Networks",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

address = "United States",

note = "37th Annual IEEE Conference on Local Computer Networks, LCN 2012 ; Conference date: 22-10-2012 Through 25-10-2012",

}

Song, Y, Liu, M, Tang, S & Mao, X 2012, Time series matrix factorization prediction of internet traffic matrices. in T Pfeifer, A Jayasumana & D Turgut (eds), Proceedings of the 37th Annual IEEE Conference on Local Computer Networks: LCN 2012. Institute of Electrical and Electronics Engineers (IEEE), Piscataway, NJ, pp. 284-287, 37th Annual IEEE Conference on Local Computer Networks, LCN 2012, Clearwater, FL, United States, 22/10/12. https://doi.org/10.1109/LCN.2012.6423629

Time series matrix factorization prediction of internet traffic matrices. / Song, Yunlong; Liu, Min; Tang, Shaojie et al.
Proceedings of the 37th Annual IEEE Conference on Local Computer Networks: LCN 2012. ed. / Tom Pfeifer; Anura Jayasumana; Damla Turgut. Piscataway, NJ: Institute of Electrical and Electronics Engineers (IEEE), 2012. p. 284-287.

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

TY - GEN

T1 - Time series matrix factorization prediction of internet traffic matrices

AU - Song, Yunlong

AU - Liu, Min

AU - Tang, Shaojie

AU - Mao, Xufei

PY - 2012

Y1 - 2012

N2 - Traffic matrices (TMs) are very important for traffic engineering and if they can be predicted, the network operations can be made beforehand. However, existing prediction methods are neither accurate nor efficient in practice. In this paper, we utilize the spatio-temporal property and low rank nature to directly predict the total TMs. The problem is that conventional matrix interpolation only works well when elements are missing uniformly and randomly. But in the case of TMs prediction, an entire part of the matrix is unknown. To solve this problem, we utilize some essential properties of TMs and add the time series forecasting into the matrix interpolation. We analyze our algorithm and evaluate its performance. The experiment result shows that our method can predict TMs under an NMAE of 30% in most cases, even predicting all the elements of next 3 weeks.

AB - Traffic matrices (TMs) are very important for traffic engineering and if they can be predicted, the network operations can be made beforehand. However, existing prediction methods are neither accurate nor efficient in practice. In this paper, we utilize the spatio-temporal property and low rank nature to directly predict the total TMs. The problem is that conventional matrix interpolation only works well when elements are missing uniformly and randomly. But in the case of TMs prediction, an entire part of the matrix is unknown. To solve this problem, we utilize some essential properties of TMs and add the time series forecasting into the matrix interpolation. We analyze our algorithm and evaluate its performance. The experiment result shows that our method can predict TMs under an NMAE of 30% in most cases, even predicting all the elements of next 3 weeks.

KW - traffic matrices prediction

KW - time series forecasting

KW - matrix interpolation

UR - http://www.scopus.com/inward/record.url?scp=84874291974&partnerID=8YFLogxK

U2 - 10.1109/LCN.2012.6423629

DO - 10.1109/LCN.2012.6423629

M3 - Conference proceeding contribution

AN - SCOPUS:84874291974

SN - 9781467315647

SP - 284

EP - 287

BT - Proceedings of the 37th Annual IEEE Conference on Local Computer Networks

A2 - Pfeifer, Tom

A2 - Jayasumana, Anura

A2 - Turgut, Damla

PB - Institute of Electrical and Electronics Engineers (IEEE)

CY - Piscataway, NJ

T2 - 37th Annual IEEE Conference on Local Computer Networks, LCN 2012

Y2 - 22 October 2012 through 25 October 2012

ER -

Time series matrix factorization prediction of internet traffic matrices

Abstract

Other

Keywords

Access to Document

Other files and links

Fingerprint

Cite this