Human activity monitoring is an exciting research area to assist independent living among disabled and elderly population. Various techniques have been proposed to recognise human activities, such as exploiting sensors, cameras, wearables, and contactless microwave sensing. Among these, the microwave sensing has recently gained significant attention due to its merit to solve the privacy concerns of cameras and discomfort caused by wearables. However, the existing microwave sensing techniques have a basic disadvantage of requiring controlled and ideal settings for high-accuracy activity detections, which restricts its wide adoptions in non-line-of-sight (Non-LOS) environments. Here, we propose a concept of intelligent wireless walls (IWW) to ensure high-precision activity monitoring in complex environments wherein the conventional microwave sensing is invalid. The IWW is composed of a reconfigurable intelligent surface (RIS) that can perform beam steering and beamforming, and machine learning algorithms that can automatically detect the human activities with high accuracy. Two complex environments are considered: one is a corridor junction scenario with transmitter and receiver in separate corridor sections and the other is a multi-floor scenario wherein the transmitter and receiver are placed on two different floors of a building. In each of the aforementioned environments, three distinct body movements are considered namely, sitting, standing, and walking. Two subjects, one male and one female perform these activities in both environments. It is demonstrated that IWW provide a maximum detection gain of 28% in multi-floor scenario and 25% in corridor junction scenario as compared to traditional microwave sensing without RIS.

人类活动监测是一个令人兴奋的研究领域，可以帮助残疾人和老年人独立生活。已经提出了各种技术来识别人类活动，例如利用传感器、相机、可穿戴设备和非接触式微波传感。其中，微波传感最近因其解决相机的隐私问题和可穿戴设备引起的不适的优点而受到广泛关注。然而，现有的微波传感技术具有一个基本缺点，即需要用于高精度活动检测的受控和理想设置，这限制了其在非视距 (Non-LOS) 环境中的广泛采用。这里，我们提出了智能无线墙（IWW）的概念，以确保在传统微波传感无效的复杂环境中进行高精度活动监测。IWW 由可以执行波束控制和波束成形的可重构智能表面 (RIS) 以及可以高精度自动检测人类活动的机器学习算法组成。考虑了两种复杂的环境：一种是走廊交叉点场景，发射器和接收器位于不同的走廊部分，另一种是多层场景，其中发射器和接收器放置在建筑物的两个不同楼层。在上述每个环境中，都考虑了三种不同的身体运动，即坐、站和走。两个科目，一名男性和一名女性在两种环境中执行这些活动。结果表明，与没有 RIS 的传统微波传感相比，IWW 在多层场景中提供 28% 的最大检测增益，在走廊交叉口场景中提供 25% 的最大检测增益。