Coronavirus Contact Tracing: Evaluating The Potential Of Using Bluetooth Received Signal Strength For Proximity Detection
Coronavirus Contact Tracing: Evaluating The Potential Of Using Bluetooth Received Signal Strength For Proximity Detection
复制标题
DOI：
10.1145/3431832.3431840
复制
发表时间：
2020-10-01
影响因子：
2.8
通讯作者：
Farrell, Stephen
中科院分区：
文献类型：
--
作者：
Leith, Douglas J.;Farrell, Stephen
许多国家正在部署新冠病毒接触者追踪应用程序，这些应用程序使用低功耗蓝牙（LE）在15分钟内检测2米内的接近程度。然而，蓝牙LE是一项未经验证的技术，这引起了人们对这些应用程序有效性的担忧。事实上，测量表明，蓝牙LE接收到的信号强度会受到以下因素的强烈影响：(i)所使用手机的型号，（ii）手机的相对方向，（iii）人体、袋子等的吸收，以及（iv）墙壁、地板、家具的无线电波反射。对接收信号强度的影响与移动2m的影响相当，因此有可能严重影响近距离检测的可靠性。这些影响是由于无线电传播的物理特性造成的，这表明基于蓝牙LE接收信号强度开发精确的接近检测方法可能具有挑战性。我们呼吁在三个方面采取行动。首先，需要进行测量，以评估在整个接触者追踪系统中部署的应用程序的附加价值，例如，关于应用程序通知的人中有多少人不会通过手动接触者追踪被发现，以及应用程序通知的人中有多少人实际检测出Covid-19阳性的数据。其次，2米/15分钟的距离限制只是一个粗略的指导方针。真正的需求是使用手机传感器来评估感染风险，这需要一个活动来收集手机传感器数据和感染结果的测量。第三，需要齐心协力，在广泛的现实环境中收集可控蓝牙LE测量值，这里报告的数据只是朝着这个方向迈出的第一步。
Many countries are deploying Covid-19 contact tracing apps that use Bluetooth Low Energy (LE) to detect proximity within 2m for 15 minutes. However, Bluetooth LE is an unproven technology for this application, raising concerns about the efficacy of these apps. Indeed, measurements indicate that the Bluetooth LE received signal strength can be strongly affected by factors including (i) the model of handset used, (ii) the relative orientation of handsets, (iii) absorption by human bodies, bags etc. and (iv) radio wave reflection from walls, floors, furniture. The impact on received signal strength is comparable with that caused by moving 2m, and so has the potential to seriously affect the reliability of proximity detection. These effects are due the physics of radio propagation and suggest that the development of accurate methods for proximity detection based on Bluetooth LE received signal strength is likely to be challenging. We call for action in three areas. Firstly, measurements are needed that allow the added value of deployed apps within the overall contact tracing system to be evaluated, e.g. data on how many of the people notified by the app would not have been found by manual contact tracing and what fraction of people notified by an app actually test positive for Covid-19. Secondly, the 2m/15 minute proximity limit is only a rough guideline. The real requirement is to use handset sensing to evaluate infection risk and this requires a campaign to collect measurements of both handset sensor data and infection outcomes. Thirdly, a concerted effort is needed to collect controlled Bluetooth LE measurements in a wide range of real-world environments, the data reported here being only a first step in that direction.