The insecurity of home digital voice assistants -amzon alexa as a case study

通过亚马逊家庭数字声音设备的研究，发现其存在一定的不安全因子，并且给出了一定的解决方法。这些数字声控设备通过声音控制家里的各种设备。但是存在一个问题：就是如果有人窃取了账号，利用声音也能控制家里的设备，该怎么办呢？文中给出了一个解决方法：当设备接到声音的命令时，并不立即执行该命令，而是利用WiFi信号判断家里是否有人在，如果有人在，则才开始执行命令。

上述针对问题进行阐述了，这里主要学习一下如何侦测室内有人移动，如何识别室内室外人移动的差异性？

we believe that detecting human motions based on WiFi signals is a practical yet low-cost solution approach due to two reaons. First, home WiFi networks are commonly deployed, so no extra deployment cost is needed. Second, only a software upgrade is required for the alexa devices, since all of them have been equipped with WiFi.

从两个视角来分析人移动，在什么地方移动等问题。我感觉很有意思。

***multi-path effect for human motions detection

***multi-reflection effect for identifying where the motions are.

室内与室外衰减程度不一样，一般来说，室外信号经历两次穿墙反射，衰减程度更大。(实验结果展示室内变化比室外变化大，这个让我有点理解不了。按照常理分析，室外信号衰减大，应该信号变化大啊。经过多次理解发现，我们看的是最后结果，所以室外最后结果都是很小，即变化不大。室内信号衰减不大，但是变化相对来说比较大。)

Our results show that it makes outside motions to result in only a small variation of CSI values, compared with a significant variation caused by inside human motions. The variations degrees of CSI values can thus be leveraged to identify the human motions occuring inside and outside the wall.