An Architectural Approach to Data Stream Processing Based on the Current State of Data

Abstracts. In peripheral systems, the characteristics of a sensor stream and available resources may fluctuate during processing. Under these conditions, it is advisable to make decisions regarding the processing method based on the current state of the data. This work aims to develop an architectural approach for structuring such decision-making. An approach is proposed wherein the assessed state of a stream segment determines processing requirements, while resource availability informs the selection of the processing configuration. The functions of components – specifically stream state assessment, requirement formulation, resource monitoring, configuration selection, and execution – are defined, along with the information flows connecting them. Using the transition from a stable stream state to a series of anomalies as an example, the study illustrates the decision-making workflow across these components and the feedback of execution metrics into the processing loop. The proposed approach establishes an architectural foundation for implementing sensor stream processing that accounts for data states and execution conditions. Introduction. The main goal of the research is to develop an architectural approach to sensor stream processing that takes the current data state into account when determining processing requirements and selecting the execution method within an Edge system. Materials and Methods. Conceptual modeling was used to define the components of adaptive sensor stream processing, their functions, and the information flows between them. The approach links data state assessment, processing requirement formulation, resource monitoring, configuration selection, and execution with feedback. Component interactions were illustrated through a transition from a stable stream state to recurring deviations, showing how changes in data state prompt a reassessment of processing requirements and how available resources inform configuration selection. Results. We propose an architecture that integrates stream state assessment, processing-requirement formulation, resource monitoring, configuration selection, and feedback-driven execution. The functions of the components, the information links between them, and the decision-making workflow following a change in data state are defined. Conclusions. The proposed approach shows how the assessed state of the sensor stream determines processing needs and how resource information informs the selection of implementation methods. This architecture serves as a foundation for the future development of specific methods for state assessment, requirement formulation, and configuration selection.