计算机科学
副载波
聚类分析
目标检测
无线传感器网络
方案(数学)
实时计算
信道状态信息
无线
天线(收音机)
频道(广播)
人工智能
正交频分复用
计算机网络
模式识别(心理学)
电信
数学分析
数学
作者
Kazuki Noguchi,Osamu Muta,Tomoki Murakami,Shoichiro Otsuki
标识
DOI:10.1109/vtc2021-fall52928.2021.9625223
摘要
Machine learning based object detection that utilizes channel state information (CSI) in wireless local area network (WLAN) systems is an effective approach for indoor positioning. In this paper, we propose a real-time CSI-based object detection scheme using interleaved subcarrier selection techniques for WLAN systems with distributed antennas, where CSI frames are collected and used as data-set for machine learning and object detection. To improve real-time detection performance, we investigate two approaches; interleaved sampling (IS), and interleaved sampling and clustering (ISC). In the IS scheme, a part of subcarriers are selected among all subcarriers in an interleaved manner to reduce data-set size while maintaining the object detection accuracy. In the ISC scheme, all subcarriers (their CSI) are grouped into several clusters in an interleaved manner and detect a target by integrating cluster-by-cluster machine-learning results. Furthermore, we demonstrate the effectiveness of the proposed approach through real-time experimental evaluations in an indoor environment scenario. Experimental results show that the ISC scheme improves object detection probability than the case without clustering, while the IS scheme is effective in reducing data-set size for obtaining almost the same performance. The results also indicate that the improved detection performance is obtained by using the proposed scheme with a distributed antenna array.
科研通智能强力驱动
Strongly Powered by AbleSci AI