Model-Based Operator Placement for Data Processing in IoT Environments

Abstract

The advances of the Internet of Things (IoT) lead to further challenges for data processing. Besides deriving meaningful information from a high amount of raw data, processing data in a timely manner is required as well, in order to enable the development of reactive IoT applications. Usually, the processing of IoT data is done in cloud-based infrastructures, which provide on-demand resources to process the data as needed. However, this affects timely processing, since sending data to off-premise cloud infrastructures increases latency and network traffic. In this paper, we propose a method to process data streams primarily on-premise in IoT environments, i.e., data is processed near to their data sources and the processing power already provided by IoT devices in the environment is explored.