Presentation | 2018-07-20 Traffic Matrix Prediction based on Bidirectional Recurrent Neural Network and Long Short-Term Memory Van An Le, Phi Le Nguyen, Yusheng Ji, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Accurate prediction of the future network traffic plays an important role in various network problems (e.g. traffic engineering, capacity planning, quality of service provisioning, etc.). Measuring all the network traffic is impossible or impractical due to the monitoring resources constraints as well as the dynamic of temporal/spatial fluctuations of the traffic. To this end, a common approach is to solve the traffic matrix interpolation using compress sensing and matrix completion. Besides that, there are some studies exploiting Deep Learning techniques such as Restricted Boltzmann Machine or Recurrent Neural Network to estimate the traffic volume. However, their proposal reveals a poor performance regarding the traffic inference when the measurement data has a highly missing rate. In this paper, we propose a highly accurate traffic prediction algorithm by leveraging the advantages of Long Short-Term Memory (LSTM) in the time series estimation and modifying the Bidirectional Recurrent Neural Networks (BRNN) in correcting the feeding data. We evaluate our model based on the Abilene Dataset which contains the real traffic matrices. The experiment results show that the proposed approach can achieve significantly better prediction accuracy in term of several metrics such as error ratio, mean absolute error and root mean square error, even when only 30% of the traffic flows in the network are measured. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | traffic prediction / recurrent neural network / long short-term memory |
Paper # | CQ2018-40 |
Date of Issue | 2018-07-12 (CQ) |
Conference Information | |
Committee | CQ |
---|---|
Conference Date | 2018/7/19(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Tohoku Univ. |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Assessment, Measurement and Control of QoE and QoS, etc. |
Chair | Takanori Hayashi(Hiroshima Inst. of Tech.) |
Vice Chair | Hideyuki Shimonishi(NEC) / Jun Okamoto(NTT) |
Secretary | Hideyuki Shimonishi(NTT) / Jun Okamoto(Nippon Inst. of Tech.) |
Assistant | Chikara Sasaki(KDDI Research) / Yoshiaki Nishikawa(NEC) / Ryo Yamamoto(UEC) |
Paper Information | |
Registration To | Technical Committee on Communication Quality |
---|---|
Language | ENG |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Traffic Matrix Prediction based on Bidirectional Recurrent Neural Network and Long Short-Term Memory |
Sub Title (in English) | |
Keyword(1) | traffic prediction |
Keyword(2) | recurrent neural network |
Keyword(3) | long short-term memory |
1st Author's Name | Van An Le |
1st Author's Affiliation | The Graduate University for Advanced Studies(Sokendai(The Graduate University for Advanced Studies)) |
2nd Author's Name | Phi Le Nguyen |
2nd Author's Affiliation | The Graduate University for Advanced Studies(Sokendai(The Graduate University for Advanced Studies)) |
3rd Author's Name | Yusheng Ji |
3rd Author's Affiliation | National Institute of Informatics(NII) |
Date | 2018-07-20 |
Paper # | CQ2018-40 |
Volume (vol) | vol.118 |
Number (no) | CQ-140 |
Page | pp.pp.51-56(CQ), |
#Pages | 6 |
Date of Issue | 2018-07-12 (CQ) |