文章目录
title
3D City Map Reconstruction from UAV-based Radio Measurements
topic: UAV 3D 地图重建
本文笔记已停止更新
贡献
We propose a first attempt for reconstructing a whole 3D city map based on radio measurements made from a low altitude UAV with receive signal strength indicator (RSSI) sensing capability
概述
Traditionally, 3D city map reconstruction uses photogrammetry techniques which need high resolution stereo images and extensive processing capability . Over the past decade, a number of approaches have been developed which employ remote sensing techniques for obtaining geometry measurements of the environment using depth information
无人机的优势
The use of lightweight low-altitude UAV is very attractive as it provides a quickly deployable relay functionality in weakly covered areas, temporary hotspot areas, or as an alternative to expensive terrestrial ultra-dense networks.
Map Reconstruction 步骤:
RSSI如何获取
The RSSI levels are gathered at each selected UAV location from a large set of static outdoor ground users in a time-division multiplexing mode (TDM).
-
简:detect and classify line-of-sight (LoS) vs. non-line-of-sight (NLoS) channels towards ground users.
详:phase 1, machine learning tools are applied to learn the propagation parameters from the RSSI measurements from a target group of ground users. The learned parameters are in turn exploited to classify outdoor users in either LoS or NLoS.
从和ground user的交互中得到RSSI measurement,然后学习propagation parameter. paremeter是用来划分LoS和NLoS的依据。
-
简:the LoS vs. NLoS data is fed as input to a building position and height reconstruction algorithm.
详:phase 2, the LoS/NLoS classified data is exploited to yield a reconstructed city map via the low-complexity resolution of a large set of inequality equations.
根据LoS/NLoS数据生成city map.
原理——principle of reconstructing building shapes
The proposed principle of reconstructing building shapes and locations from radio measurements bears some intuitive analogy with the problem of reconstructing the shape of an object that is exposed under a series of light sources from the shadow images created from the object.
可以类比通过不同光源下的阴影图来重建一个object
In our context, the light/shade classification is replaced with a notion of LoS/NLoS classification for the radio channel.
只不过现在相当于LoS是被光照到,NLoS是背光
重建后的效果
建模
1.城市范围
2.楼的宽度
3.坐标和RSSI的表示
算法
通过SVM来学习channel model
Optimization of Target User Group
it is essential that maximum LoS/NLoS distinguishability is achieved on the target group. The following method is proposed for selecting the ring radius R.
合理地选择半径R使得 γ ‾ L o S \overline{\gamma}_{LoS} γLoS和 γ ‾ N L o S \overline{\gamma}_{NLoS} γNLoS相差尽可能大
大概意思是,R取的太大,因为远场区的LoS和NLoS损耗相差不大,所以会使得二者平均损耗减小,而取的太小视野又太小。
Radio Propagation Parameter Learning
求出 θ S \theta_{S} θS 后就可以得到 γ S l \gamma_{S}^{l} γSl的表达式。
user classification—正确划分LoS和NLoS用户
we proceed with classifying the remaining network users into LoS or NLoS categories. we use the SVM classifier for the sake of good performance.
训练出的这个SVM能根据 f i f^{i} fi预测 t i t^{i} ti
抽象成一个支持向量机的问题,通过训练得到 w T \mathbf{w^{T}} wT
Map Reconstruction
算法流程图总结如下