Due to the ever-growing powers in sensing, computing, communicating and storing, mobile devices (e.g., smartphone, smartwatch, smart glasses) become ubiquitous and an indispensable part of people’s daily life. Until now, mobile devices have been adopted in many applications, e.g., exercise assessment, daily life monitoring, human-computer interactions, user authentication, etc. Among the various applications, Human Activity Recognition (HAR) is the core technology behind them. Specifically, HAR gets the sensor data corresponding to human activities based on the built-in sensors of mobile devices, and then adopts suitable recognition approaches to infer the type of activity based on sensor data. The last two decades have witnessed the ever-increasing research in HAR. However, new challenges and opportunities are emerging, especially for HAR based on mobile devices. Therefore, in this paper, we review the research of HAR based on mobile devices, aiming to advance the following research in this area. Firstly, we give an overview of HAR based on mobile devices, including the general rationales, main components and challenges. Secondly, we review and analyze the research progress of HAR based on mobile devices from each main aspect, including human activities, sensor data, data preprocessing, recognition approaches, evaluation standards and application cases. Finally, we present some promising trends in HAR based on mobile devices for future research.

中文翻译：

---

基于移动设备的人类活动识别的系统综述：概述、进展和趋势


由于传感、计算、通信和存储能力的不断增强，移动设备（例如智能手机、智能手表、智能眼镜）变得无处不在，成为人们日常生活中不可或缺的一部分。目前，移动设备已被应用于许多应用中，例如运动评估、日常生活监测、人机交互、用户身份验证等。在这些应用中，人体活动识别（HAR）是其背后的核心技术。具体来说，HAR根据移动设备内置的传感器获取与人类活动相对应的传感器数据，然后采用合适的识别方法根据传感器数据推断活动类型。过去二十年见证了 HAR 的研究不断增长。然而，新的挑战和机遇正在出现，特别是对于基于移动设备的HAR。因此，本文对基于移动设备的HAR研究进行综述，旨在推进该领域的后续研究。首先，我们概述了基于移动设备的 HAR，包括一般原理、主要组成部分和挑战。其次，从人类活动、传感器数据、数据预处理、识别方法、评估标准和应用案例等主要方面回顾和分析了基于移动设备的HAR研究进展。最后，我们提出了基于移动设备的 HAR 的一些有前景的趋势，以供未来研究。