Filtering scheme for context-aware fog computing in cyber-physical systems

In the cloud-based Internet of Things, the amount of available network bandwidth can become a bottleneck, especially in real-time sensor network applications. This study presents an architecture and an algorithm for context-aware fog data filtering that can map data features and their appearance frequency. This reduces the amount of data sent using long-range communications. In this study, a Fast Fourier Transform (FFT)-based algorithm is presented, and a feature mapping technique is used. The filtering algorithm considers both historic data and adjacent sensor data to determine whether unexpected sensor outputs are caused by events in the system or faulty sensor readings. The type of data transmitted to supervisors is determined by the needs of the receivers. In this way, both events and raw data can be accessed from the proposed filters.