An efficient method for physical fields mapping through crowdsensing

被引：10

作者：

Dardari, Davide ^{[1
]}

Pasolini, Gianni ^{[2
]}

Zabini, Flavio ^{[2
]}

机构：

[1] Univ Bologna, DEI CNIT, Via Venezia 52, Bologna, Italy

[2] Univ Bologna, DEI CNIT, Viale Risorgimento 2, Bologna, Italy

来源：

PERVASIVE AND MOBILE COMPUTING | 2018年 / 48卷

关键词：

Crowdsensing; Gaussian processes; Spatial field estimation; Environmental mapping; WIRELESS SENSOR NETWORKS; GAUSSIAN-PROCESSES; ALGORITHMS; RECONSTRUCTION; PREDICTION; SIGNALS;

D O I：

10.1016/j.pmcj.2018.06.001

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Crowdsensing is an effective method to map physical spatial fields by exploiting sensors embedded in smartphones. Enclosing humans in the loop increases the amount of data available for the mapping process, with benefits in terms of accuracy and cost. On the other hand, the huge amount of data generated and the irregular spatial distribution of measurements are serious issues to be addressed. In this paper we propose a combined Gaussian process (GP)-State space method for crowd mapping whose complexity and memory requirements for field representation do not depend on the number of data measured. The method is validated through an experimental campaign and simulations. (C) 2018 Elsevier B.V. All rights reserved.

引用

页码：69 / 83

页数：15

共 46 条

[1]

Alzantot M., 2012, P 20 INT C ADV GEOGR, P99, DOI DOI 10.1145/2424321.2424335

[2]

[Anonymous], THESIS

[3]

Arias-de-Reyna E, 2017, INT CONF ACOUST SPEE, P2487, DOI 10.1109/ICASSP.2017.7952604

[4]

Bian J., CORR