On the feasibility of estimating soluble sugar content using millimeter-wave
On the feasibility of estimating soluble sugar content using millimeter-wave
复制标题
DOI：
10.1145/3302505.3310065
复制
发表时间：
2019-04
期刊：
影响因子：
--
通讯作者：
Zhicheng Yang;Parth H. Pathak;M. Sha;Tingting Zhu;Junai Gan;Pengfei Hu;P. Mohapatra
中科院分区：
文献类型：
--
作者：
Zhicheng Yang;Parth H. Pathak;M. Sha;Tingting Zhu;Junai Gan;Pengfei Hu;P. Mohapatra
随着新型传感技术、连续监测、数据驱动推理、精确灌溉控制和智能物联网（IoT）系统的发展，农业部门正在经历一场革命。基于红外和激光的专业设备被开发出来，以帮助农民评估产品质量，特别是其糖含量。然而，这样的设备是昂贵的并且不容易为消费者所获得。在本文中，我们研究了使用60 GHz毫米波（mmWave）信号作为一种无处不在的和非侵入性的方式来估计水果中的可溶性糖含量（SSC）的可行性。随着毫米波技术的快速发展，60 GHz WiFi很可能在未来的移动的设备中变得普遍。我们的研究表明，当60 GHz的WiFi信号从水果反射时，反射可以用来推断水果的糖含量。我们确定的反射信号与不同的SSC的变化的根本原因，并研究反射的大小，形状和密度的水果的影响。然后，我们开发基于接收信号强度和幅度的统计特征，并使用它们来设计基于回归的估计模型。通过对300个水果样品的广泛评价，我们发现我们提出的技术可以估计三种不同类型水果的SSC，平均相关系数为85%。我们的预测误差在用户的味觉感知范围内。
With the development of novel sensing techniques, continuous monitoring, data-driven inferences, precision irrigation control, and intelligent Internet-of-Things (IoT) systems, agriculture sector is witnessing a revolution. Specialized devices based on infrared and laser are developed to assist farmers in assessing the produce quality, especially its sugar content. However, such devices are expensive and not readily available to consumers. In this paper, we investigate the feasibility of using 60 GHz millimeter-wave (mmWave) signal as a ubiquitous and non-invasive way to estimate the Soluble Sugar Content (SSC) in fruits. With the rapid development in the mmWave technology, 60 GHz WiFi is likely to become pervasive in future mobile devices. Our study shows that when 60 GHz WiFi signals reflect from a fruit, the reflection can be used to infer the fruit's sugar content. We identify the underlying reasons of variations in reflection signals with varying SSC and study the impact of size, shape and density of fruits on reflections. We then develop statistical features based on received signal strength and amplitude, and use them to design regression-based estimation models. With an extensive evaluation with 300 fruit samples, we find that our proposed technique can estimate SSC in three different type of fruits with an average correlation coefficient of 85%. Our prediction errors are within the range of user's taste perception.