Mechanism for context-aware substitution of Smart-M3 agents based on dataflow network model

Abstract

Dataflow network is a concise well-understood and well-matching approach for building applications in Internet of Things (IoT) world, where a task of obtaining meaningful results in time strongly depends on efficiency of orchestration of dataflows. Processing units transform incoming signals into intermediate data suitable for further processing and analysis by using certain rules, which are further processed by the next levels of dataflow processing. As a result, the set of source signals get meaningful representation for the users or smart devices. This paper studies the reliability of dataflow networks in Smart-M3 platform - open-source platform for smart space application development. We developed a new method and implemented the corresponding mechanism for agent substitution, which allows smoothly transferring data processing from the compromised agent to a substitute. The transfer includes all details of the processing program and the computational context, so the user will not even notice the change. And when the original agent gets back to the normal working condition, the dataflow processing and new computational context is handed back to it. The developed mechanism allows to preserve the context in agent disconnection and physical device relocation situations, which are common for IoT environments.