无线室内定位大概分为两类：
1）基于指纹匹配的室内定位
2）基于模型的室内定位（These schemes calculate locations based on geometrical models rather than search for best-fit signatures from pre-labeled reference database. These approaches trade the measurement efforts at the cost of decreasing localization accuracy）
*log-distance path loss（power-distance mapping describe relationship between RSS values and RF propagation distances）(定位误差高于5米)
*ToA, TDoA, and AoA 来刻画发射端与接收端之间的几何关系。
在上述基础上，有些研究团队探索SLAM(simultaneous localization and mapping)同时定位和感知周围环境映射为地图
*****************************************************************************************************************************************************************************
清华大学刘云浩团队之成员杨峥：Locating fingerprint space: wireless indoor localization with little human intervention, mobicom'12
site survey exists several drawbacks:
1) intensive costs on manpower and time
2) limiting the applicable buildings of wireless localization worldwide.
3) inflexibility to environment dynamics
基于上述的缺陷，本文提出一个新方法即研究移动手机内置的传感器和利用用户移动来构建基于室内地图的无线指纹数据库。
该工作提出一个LiFS系统利用当前的WiFi设备和移动手机实现室内定位；指纹数据库的校准是众包技术和自动实现的。LiFS系统定位精度为5.88米， RADAR为3.42米，基于模型的定位精度为高于5米，EZ系统的高于7米。
**********************************************************************************************************************************************
北京大学张大庆团队，Dynamic-MUSIC: accurate device-free indoor localization, ubicomp'16（相关工作写的非常好，可以参考一下）
目前无源室内定位研究的现状：
1）劳动力密集的线下训练
2）特定设备（require dedicated special-purpose devices）
基于上述缺陷，本文提出了MaTrack， 利用dynamic0-MUSIC算法侦测人体引起的微弱信号变化，此外也能区分静态物体反射信号与人体的反射信号，进而识别人体目标的角度进行定位。亮点是无需进行训练，仅需要两个接收端，在人走动的情况下，均值定位精度在0.6米以下。该系统适合移动的对象，对于静态的对象则精度会大打折扣（后续工作想参考Non-invasive detection of moving adn stationary human with wifi来利用呼吸做静态目标对象的定位。With a much larger bandwidth in 802.11ac and more antennas attached to the commodity AP, the resolution of our system can be improved significantly. We leave this challenging problem as our future work.）。
[1]Nuzzer systems'30 (Nuzzer: a large-scale device-free passive localization system for wireless environment, TMC, 2013)using RSSI signature as a fingerprint
[2] Pilot'45(Pilot: passive device-free indoor localization using channel state information, ICDCS,2013) CSI information to build fingerprint map
[3] Mono-PHY'1(MonoPHY: mono-strea-based device-free WLAN localization via physical layer information, WCNC,2013) CSI information to build fingerprint map
[4] E-eyes'40 utilizes the amplitude pattern of CSI to build a fingerprint map to identify the target's moving trajectory and accordingly determine the destination room
[5] Ichnaea'28 (Ichnaea: a low-overhead robust wlan device-free passive localization system, 2014)中位值精度为2.5米 is a device-free tracking system based on RSSI with an offline background training phase.
[6] Ohara'22 (Transferring positioning model for device-free passive indoor localization, ubicomp , 2015)propose a fingerprint-based device-free method to locate the mobile target and apply model transformation scheme to reduce the offline training load.
多目标无源室内定位：
[1]Multi-entity devcie-free WLAN localization ,2012, GLOBECOM
[2]ACE:an accurate and efficient multi-entity device-free WLAN localization system, TMC, 2015
[3]SCPL：indoor device-free multi-subject counting and localization using radio signal strength, 2013, IPSN
[4]FitLoc: fine-grained and low-cost device-free localization for multiple targets over various area,
[5]MODLoc: localizing multiple objects in dynamic indoor environment, 2014, TPDS
[6]Fine-grained localization for multiple transceiver-free objects by using RF-based technologies,2014, TPDS
[7]Multi-person localization via RF body reflections, NSDI, 2015(利用雷达技术实现该功能)
以下不是多目标定位的工作
[8]Accurate and efficient indoor location by dynamic warping in sequence-type radio-map,ubicomp, 2018
[9] Device-free localization with multidimensional wireless link information, TVT, 2015
下面的无源室内定位或者跟踪需要部署多个无线设备或特殊设备：
[1]Wilson'42 部署多个无线设备构建 radio tomographic imaging(RTI ） method to achieve a high accuracy for moving target localization
[2]WiDeo'18
[3]WiSee'23
特殊应用的论文：
[1] iLocScan: harnessing multipath for simultaneous indoor source localization and space scanning,sensys'14
*************************************************************************************************************************************************
Pallas self-bootstrapping fine-grained passive indoor localization using wifi monitors, 2017,TMC
Pallas uses off-the-shell access point hardware to opportunistically capture WiFi packets to infer the location of smartphones in the indoor environment.
关键创新点（the key novelty ）：the passive fingerprint database for localization is automatically constructed and updated without any active participation of WiFi devices or manual calibration.
1) passive landmarks： WiFi RSS traces contain more than two landmarks
2) 根据室内地图和WiFi monitor的位置， Pallas 把RSS trace (RSS trend and RSS distribution 侦测到连续两个锚点)映射到特定的室内路线。
该工作把积极定位分为四类:
1) infrastructure based:3,15,16
2) wireless fingerprint based:1,2,17,18
3) RF propagation model based:1,19,20
[1] RADAR: an in-building RF-based user location and tracking system, 2000
[2] Indoor localization without the pain, 2010,
[3] Zero-configuration, robust indoor localization: Theory and experimental, 2006.
4)SLAM and crowdsourcing based:4,5,6,7,8,21,22,23
被动定位分为两类：
1）device-based passive localization
2) device-free passive localization
*****************************************************************************************************************************************************
（参考A
rrayTrack, 2013）
1) Known technologies: not accurate enough (WiFi)
2) require dedicated infrastructure (ultrasound)
3) require cameras and good light conditions (vision)
*******************************************************************************************************
人体动作识别相关工作的简介和总结
******************************************************************************************************
根据调研可以把当前人体动作识别应用粗略分为三类：
1）粗粒度的动作识别
(1) 粗粒度动作的种类
*特定的动作: 摔倒行为识别，吸烟行为的识别，睡觉行为的监控
*日常动作：在室内环境经常出现的动作，例如walking, sit-down, bend, squat
（2) 特定位置下的粗粒度动作识别
*预先选定一些动作和位置，在该位置下采集一些粗粒度动作，然后经过训练，接着使用测试集测试，最后得出分类精度（动作识别的固定设计思维）
CARM, WiFall等
（3）由动作类别推出所在的环境位置
*根据识别出的动作特点，推理出该动作经常发生的区域（E-eyes）
[1] E-eyes: device-free location-oriented activity identification using fine-grained wifi signatures, mobicom, 2014
[2] WiFi-Enabled Smart Human Dynamics Monitoring, sensys, 2017
(4) 移动状态下的粗粒度动作识别（跟踪问题）
*预先设定路线，每条路线上分别选取几个位置做特定动作，最后识别出动作
[1] 3D Tracking via Body Radio Reflections, NSDI, 2014
[2] Widar2.0: passive human tracking with a single wifi link, Mobisys, 2018
[3] Device-free localization for human activity monitoring, 2019
(5) 独立于固定位置的动作识别：选择几个位置，然后分别在不同位置下做一些预先设定的动作，然后分析和识别（这是独立位置的动作识别初期的研究方法，后来是利用深度学习、迁移学习方法来探索）。
[1]CSI-based device-free wireless localization and activity recognition using radio image features, TVT, 2017
[2]Device-free simultaneous wireless localization and activity recognition with Wavelet Feature, TVT, 2017
[3]Towards environment independent device-free human activity recognition, Mobicom, 2018
上述是目前研究人体动作识别的几种相关思路。最近一年，关于人体动作识别的论文，开始利用深度学习、迁移学习等方法来探索位置差异、个体差异、环境的影响，并尝试从数学模型来探索信号在遇到人体的具体变化（Towards a diffraction-based sensing approach on human activity recognition, 2019, ubicomp）。此外，基于声音的、可见光、毫米波、雷达信号等人体动作识别工作在当前*会议或期刊出现的比较多。
2）细粒度的手势识别
3）生理或生命特征（呼吸，唇语，心跳）识别