講演抄録/キーワード |
講演名 |
2018-03-02 11:15
確率的ニューラルネットワークを用いた無線LANチャネルのアイドル状態予測とそのFPGA実装 ○ウェバー ジュリアン・メーボドニヤ アボルファズル・侯 亜飛・矢野一人・臼井 誠・熊谷智明(ATR) SR2017-134 |
抄録 |
(和) |
(まだ登録されていません) |
(英) |
WLAN traffic is becoming increasingly congested particularly in the 2.4,GHz band.
Simultaneous utilization of multiple-bands has been proposed as one solution to this problem. To take advantage of the scattered resources across channels an agile transceiver is required that operates on multiple channels simultaneously.
To realize this concept a busy/idle (B/I) predictor is required in the sensing unit of the multi-band WLAN. In this paper we report on the prediction of upcoming WLAN B/I status using the probabilistic neural network (PNN) algorithm based on pattern matching technique. Using actual captured IEEE~802.11 wireless data frames, the prediction performance is compared for different PNN configurations as well as interval of retraining and results show that the PNN is a suitable prediction algorithm. The PNN firmware development including functional design and resource requirements are described. |
キーワード |
(和) |
無線WLAN / 確率的ニューラルネットワーク / 機械学習 / アイドル予測 / FPGA / 実装 / / |
(英) |
WLAN / probabilistic neural network / machine-learning / idle prediction / FPGA / implementation / / |
文献情報 |
信学技報, vol. 117, no. 457, SR2017-134, pp. 117-124, 2018年2月. |
資料番号 |
SR2017-134 |
発行日 |
2018-02-21 (SR) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
SR2017-134 |