Presentation 2018-02-26
Evaluation of Mobile Data Offloading Method Using Deep Reinforcement Learning
Daisuke Mochizuki, Yu Abiko, Takato Saito, Masaji Katagiri, Daizo Ikeda, Hiroshi Mineno,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In recent years, the demand for mobile data communication has been increasing as the technologies of IoT have been widely used, and it is known that the mobile data traffic has a locality in commuting time and urban areas. There is a problem that the cellular infrastructures are crowded and the mobile data communication demand makes the network accommodation efficiency lower. As a method for improving the cellular infrastructures bandwidth utilization efficiency, a transmission rate control method has been proposed such as MDOP which balances the load by focus on the delay tolerance of contents. However, mathematical models do not have perform well to maximize bandwidth utilization efficiency in various situations. In this paper, to improve bandwidth utilization efficiency, we propose mobile data offloading method using deep reinforcement learning considering the delay tolerance of contents. The proposal method can control transmission appropriately in situations where mathematical models do not work well. As a result, we confirmed that the proposal method reduces the excess data amount by 26% from the control target value compared to the time offloading of MDOP.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Offloading / Reinforcement Learning / Delay Tolerant / Mobile Data
Paper # MoNA2017-65
Date of Issue 2018-02-19 (MoNA)

Conference Information
Committee MoNA / ASN / IPSJ-MBL / IPSJ-UBI
Conference Date 2018/2/26(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Sophia University
Topics (in Japanese) (See Japanese page)
Topics (in English) Ubiquitous system, Ambient intelligence, Next generation wireless communicasion, Mobile network, etc.
Chair Ryoichi Shinkuma(Kyoto Univ.) / Hiraku Okada(Nagoya Univ.) / 河口 信夫(名大) / 寺田 努(神戸大)
Vice Chair Shigeaki Tagashira(Kansai Univ.) / Gen Kitagata(Tohoku Univ.) / Shigeki Shiokawa(KAIT) / Jin Nakazawa(Keio Univ.) / Satoru Yamano(NEC) / 深澤 佑介(ドコモ) / 久保 健(KDDI) / 北村 操代(三菱電気)
Secretary Shigeaki Tagashira(Kyushu Univ.) / Gen Kitagata(NTT) / Shigeki Shiokawa(NEC) / Jin Nakazawa(NICT) / Satoru Yamano(Sophia Univ.) / 深澤 佑介(愛知工大) / 久保 健(阪大) / 北村 操代(豊橋技科大) / (NTT)
Assistant Takayuki Nishio(Kyoto Univ.) / Takato Saito(NTT) / Hiroto Aida(Doshisha Univ.) / Tomoyuki Ota(Hiroshima City Univ.) / Tatsuya Kikuzuki(Fujitu Lab.) / Ryo Nakano(HITACHI) / Yoshifumi Hotta(Mitsubishi Electric)

Paper Information
Registration To Technical Committee on Mobile Network and Applications / Technical Committee on Ambient intelligence and Sensor Networks / Special Interest Group on Mobile Computing and Pervasive Systems / Special Interest Group on Ubiquitous Computing System
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Evaluation of Mobile Data Offloading Method Using Deep Reinforcement Learning
Sub Title (in English)
Keyword(1) Offloading
Keyword(2) Reinforcement Learning
Keyword(3) Delay Tolerant
Keyword(4) Mobile Data
1st Author's Name Daisuke Mochizuki
1st Author's Affiliation Shizuoka University(Shizuoka Univ.)
2nd Author's Name Yu Abiko
2nd Author's Affiliation Shizuoka University(Shizuoka Univ.)
3rd Author's Name Takato Saito
3rd Author's Affiliation NTT DOCOMO,Inc.(NTT DOCOMO)
4th Author's Name Masaji Katagiri
4th Author's Affiliation NTT DOCOMO,Inc.(NTT DOCOMO)
5th Author's Name Daizo Ikeda
5th Author's Affiliation NTT DOCOMO,Inc.(NTT DOCOMO)
6th Author's Name Hiroshi Mineno
6th Author's Affiliation Shizuoka University(Shizuoka Univ.)
Date 2018-02-26
Paper # MoNA2017-65
Volume (vol) vol.117
Number (no) MoNA-450
Page pp.pp.141-146(MoNA),
#Pages 6
Date of Issue 2018-02-19 (MoNA)