Presentation | 2019-06-19 Policy Gradient Reinforcement Learning for Reducing Transmission Delay in EDCA Masao Shinzaki, Yusuke Koda, Koji Yamamoto, Takayuki Nishio, Masahiro Morikura, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | This paper proposes a packet mapping algorithm among Access Categories (ACs) in Enhanced Distributed Channel Access (EDCA) scheme based on policy gradient Reinforcement Learning (RL).In EDCA scheme based on an autonomous distributed control, high priority packets obtain more transmission opportunity than low priority packets.The arrival rate of high priority packets can be higher than that of low priority packets.In such a situation, mapping high priority packets to the AC defined in EDCA scheme is not necessarily the best mapping algorithm to minimize the transmission delay of high priority packets. Therefore, we assume that EDCA scheme can map high priority packets to any AC.Although each AP sends high priority packets early, however, APs can not always send high priority packets early.This paper proposes policy gradient RL to empirically obtain optimal mapping algorithm.By using the mapping algorithm based on RL, simulation results reveal that the transmission delay can be reduced.The average transmission delay of the proposed mapping algorithm is 13.8% smaller than that of the conventional mapping algorithm.Moreover, the average transmission delay of the proposed mapping algorithm is 5.2% smaller than that of the heuristic mapping algorithm. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | IEEE 802.11e / EDCA / reinforcement learning / policy gradient |
Paper # | RCS2019-52 |
Date of Issue | 2019-06-12 (RCS) |
Conference Information | |
Committee | RCS |
---|---|
Conference Date | 2019/6/19(3days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Miyakojima Hirara Port Terminal Building |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | First Presentation in IEICE Technical Committee, Resource Control, Scheduling, Wireless Communications, etc. |
Chair | Tomoaki Otsuki(Keio Univ.) |
Vice Chair | Satoshi Suyama(NTT DoCoMo) / Fumiaki Maehara(Waseda Univ.) / Toshihiko Nishimura(Hokkaido Univ.) |
Secretary | Satoshi Suyama(NTT) / Fumiaki Maehara(Kyushu Univ.) / Toshihiko Nishimura |
Assistant | Kazushi Muraoka(NTT DOCOMO) / Shinsuke Ibi(Doshisha Univ.) / Koichi Adachi(Univ. of Electro-Comm.) / Osamu Nakamura(Sharp) / Shinya Kumagai(Fujitsu Labs.) |
Paper Information | |
Registration To | Technical Committee on Radio Communication Systems |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Policy Gradient Reinforcement Learning for Reducing Transmission Delay in EDCA |
Sub Title (in English) | |
Keyword(1) | IEEE 802.11e |
Keyword(2) | EDCA |
Keyword(3) | reinforcement learning |
Keyword(4) | policy gradient |
1st Author's Name | Masao Shinzaki |
1st Author's Affiliation | Kyoto University(Kyoto Univ.) |
2nd Author's Name | Yusuke Koda |
2nd Author's Affiliation | Kyoto University(Kyoto Univ.) |
3rd Author's Name | Koji Yamamoto |
3rd Author's Affiliation | Kyoto University(Kyoto Univ.) |
4th Author's Name | Takayuki Nishio |
4th Author's Affiliation | Kyoto University(Kyoto Univ.) |
5th Author's Name | Masahiro Morikura |
5th Author's Affiliation | Kyoto University(Kyoto Univ.) |
Date | 2019-06-19 |
Paper # | RCS2019-52 |
Volume (vol) | vol.119 |
Number (no) | RCS-90 |
Page | pp.pp.91-96(RCS), |
#Pages | 6 |
Date of Issue | 2019-06-12 (RCS) |