[Back]

H. Truong, M. Halper:
"Characterizing Incidents in Cloud-Based IoT Data Analytics";
Talk: 42nd IEEE International Conference on Computers, Software & Applications (COMPSAC 2018), Tokyo, Japan; 2018-07-23 - 2018-07-27; in: "Proceedings of the IEEE 42nd Annual Computer Software and Applications Conference (COMPSAC 2018)", S. Reisman, S. Ahamed, C. Demartini, T. Conte, L. Liu, W. Claycomb, M. Nakamura, E. Tovar, S. Cimato, C. Lung, H. Takakura, J. Yang, T. Akiyama, Z. Zhang, K. Hasan (ed.); IEEE, Volume 1 (2018), ISBN: 978-1-5386-2667-2; 442 - 447.

Systems for big Internet of Things (IoT) data analytics are extremely complex. Different software components at different software stacks from different infrastructures and providers are involved in handling different types of data. Various types of incidents may occur during execution of such systems due to problems in software stacks, the data itself, and processing algorithms. Here incidents reflect unexpected context-specific situations that might happen within data themselves, machine learning algorithms, data analytics pipelines, and underlying big data services and computing platforms. It is important to address any incident that prevents the pipeline running correctly or producing the expected quality of analytics. In this paper, we show the need to characterize incidents for IoT data analytics in the cloud with real-world examples. We characterize incidents based on various aspects in IoT data analytics, including analytics phases, status of data, software services, and stakeholders. We introduce a meta-model for capturing knowledge about incidents.

http://dx.doi.org/10.1109/COMPSAC.2018.00068