4.5 Article

OXI: An online tool for visualization and annotation of satellite time series data

期刊

SOFTWAREX
卷 23, 期 -, 页码 -

出版社

ELSEVIER
DOI: 10.1016/j.softx.2023.101476

关键词

Annotation tool; Annotations; Time series; Ground truth; Satellite telemetry

向作者/读者索取更多资源

Satellite telemetry data presents challenges in terms of volume, sampling rates, redundancy, and interconnections. Existing tools do not efficiently handle satellite telemetry and lack a user-friendly interface. To address this gap, we propose the web-based OXI tool, optimized for handling real satellite telemetry data, with improved data loading, visualization, and annotation capabilities. The tool was designed and evaluated in collaboration with domain experts from Airbus and ESA.
Satellite telemetry data is a special case of multivariate time series characterized by large volumes (in terms of both the number of series and samples), varying sampling rates (including time gaps), redundant sensors, and many interconnections between the series. Special tools are needed to handle visualization, analysis, and, most importantly, annotation of such data. Manually prepared annotations are crucial when designing and evaluating algorithms for satellite telemetry analysis, including anomaly detection in telemetry data. Although there are many applications for time series analysis, there are no tools that would smoothly handle typical satellite telemetry and at the same time provide a free user-friendly interface accessible from any computer without the need for installation, registration, data sharing, or special training. As a solution to this technology gap, we propose the web-based OXI tool written in JavaScript and publicly available at https://oxi.kplabs.pl/. Based on the TRAINSET application (https://trainset.geocene.com/), we optimized each part of the interface to handle real satellite telemetry data from European Space Agency (ESA) missions. The most important improvements include the redesigned data loading and visualization to effectively handle large datasets (up to 2,000,000 samples), different annotation modes (1D/2D and single/multi-series), autosaving annotation status in the browser storage, and dozens of user experience enhancements (autoscaling of axes, color-coded selectors, additional hotkeys, and many others). The tool was designed and evaluated in cooperation with domain experts from Airbus and ESA. Finally, we provide an extensive description of the software, design choices, and their impact on creating and evaluating machine learning models for anomaly detection in satellite telemetry.& COPY; 2023 Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.5
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据