Suppr超能文献

亚分米无调制可见定位光信号机会的综合研究

A Comprehensive Study on Light Signals of Opportunity for Subdecimetre Unmodulated Visible Light Positioning.

机构信息

WAVES, Department of Information Technology, Ghent University/imec, Technologiepark-Zwijnaarde 126, B-9052 Ghent, Belgium.

出版信息

Sensors (Basel). 2020 Sep 29;20(19):5596. doi: 10.3390/s20195596.

Abstract

Currently, visible light positioning (VLP) enabling an illumination infrastructure requires a costly retrofit. Intensity modulation systems not only necessitate changes to the internal LED driving module, but decrease the LEDs' radiant flux as well. This hinders the infrastructure's ability to meet the maintained illuminance standards. Ideally, the LEDs could be left unmodulated, i.e., unmodulated VLP (uVLP). uVLP systems, inherently low-cost, exploit the characteristics of the light signals of opportunity (LSOOP) to infer a position. In this paper, it is shown that proper signal processing allows using the LED's characteristic frequency (CF) as a discriminative feature in photodiode (PD)-based received signal strength (RSS) uVLP. This manuscript investigates and compares the aptitude of (future) RSS-based uVLP and VLP systems in terms of their feasibility, cost and accuracy. It demonstrates that CF-based uVLP exhibits an acceptable loss of accuracy compared to (regular) VLP. For point source-like LEDs, uVLP only worsens the trilateration-based median p50 and 90th percentile root-mean-square error p90 from 5.3cm to 7.9cm (+50%) and from 9.6cm to 15.6cm (+62%), in the 4m × 4m room under consideration. A large experimental validation shows that employing a robust model-based fingerprinting localisation procedure, instead of trilateration, further boosts uVLP's p50 and p90 accuracy to 5.0cm and 10.6cm. When collating with VLP's p50=3.5cm and p90=6.8cm, uVLP exhibits a comparable positioning performance at a significantly lower cost and at a higher maintained illuminance, all of which underline uVLP's high adoption potential. With this work, a significant step is taken towards the development of an accurate and low-cost tracking system.

摘要

目前,可见光定位(VLP)需要昂贵的基础设施改造才能实现照明。强度调制系统不仅需要更改内部 LED 驱动模块,还会降低 LED 的辐射通量。这会降低基础设施满足维持照度标准的能力。理想情况下,LED 可以保持未调制状态,即未调制可见光定位(uVLP)。uVLP 系统本质上成本较低,利用机会光信号(LSOOP)的特性来推断位置。本文表明,适当的信号处理可以将 LED 的特征频率(CF)用作基于光电二极管(PD)的接收信号强度(RSS)uVLP 的判别特征。本手稿研究并比较了基于 RSS 的 uVLP 和 VLP 系统在可行性、成本和准确性方面的能力。结果表明,与(常规)VLP 相比,基于 CF 的 uVLP 具有可接受的精度损失。对于类似点源的 LED,在考虑的 4m×4m 房间中,uVLP 仅将基于三角测量的中位数 p50 和第 90 百分位数均方根误差 p90 从 5.3cm 恶化到 7.9cm(增加 50%)和从 9.6cm 恶化到 15.6cm(增加 62%)。大量实验验证表明,采用稳健的基于模型的指纹定位过程,而不是三角测量,可进一步将 uVLP 的 p50 和 p90 精度提高到 5.0cm 和 10.6cm。当与 VLP 的 p50=3.5cm 和 p90=6.8cm 进行比较时,uVLP 的定位性能相当,但成本明显更低,维持的照度更高,所有这些都突出了 uVLP 的高采用潜力。通过这项工作,朝着开发准确且低成本的跟踪系统迈出了重要一步。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/898d/7583036/142551f4691a/sensors-20-05596-g001.jpg

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验