| 講演抄録/キーワード |
| 講演名 |
2025-11-20 14:55
[依頼講演]ローカル5Gにおける深層学習を用いた干渉予測と複数L5Gネットワーク間サービスエリア構築手法 ○表 昌佑・森山雅文・村上 誉・沢田浩和・松村 武(NICT) SRW2025-35 |
| 抄録 |
(和) |
This paper proposes a deep learning-based wireless prediction framework to enable interference-aware service area expansion in Local 5G (L5G) networks. At the core of the framework is RadioResUNet, a neural model that learns indoor building layouts and antenna placement information to accurately predict radio signal strength (RSS) and co-channel interference. This capability facilitates proactive and efficient service area design, even in complex or dynamically changing environments, thereby reducing deployment costs and enhancing network reliability. To guide the optimization of service area boundaries, the concept of service gain (G) is introduced as a quantitative metric for evaluating the trade-off between the performance improvement of the interfering network and the performance degradation of the interfered network. Simulation results demonstrate the effects of transmission power, user density, and traffic load on coverage, transmission success rate, and service gain, confirming that the proposed approach enables optimal service area configuration without inducing inter-network interference, while achieving higher overall system throughput compared to the conventional L5G service area design. |
| (英) |
This paper proposes a deep learning-based wireless prediction framework to enable interference-aware service area expansion in Local 5G (L5G) networks. At the core of the framework is RadioResUNet, a neural model that learns indoor building layouts and antenna placement information to accurately predict radio signal strength (RSS) and co-channel interference. This capability facilitates proactive and efficient service area design, even in complex or dynamically changing environments, thereby reducing deployment costs and enhancing network reliability. To guide the optimization of service area boundaries, the concept of service gain (G) is introduced as a quantitative metric for evaluating the trade-off between the performance improvement of the interfering network and the performance degradation of the interfered network. Simulation results demonstrate the effects of transmission power, user density, and traffic load on coverage, transmission success rate, and service gain, confirming that the proposed approach enables optimal service area configuration without inducing inter-network interference, while achieving higher overall system throughput compared to the conventional L5G service area design. |
| キーワード |
(和) |
/ / / / / / / |
| (英) |
Co-channel Interference / Deep learning / Local 5G (L5G) / RadioResUNet / Service Area Configuration / Service Gain / / |
| 文献情報 |
信学技報, vol. 125, no. 256, SRW2025-35, pp. 35-40, 2025年11月. |
| 資料番号 |
SRW2025-35 |
| 発行日 |
2025-11-13 (SRW) |
| ISSN |
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 |
SRW2025-35 |
|