AWS-Based Edge Computing Optimization for Industrial IoT Video Streams

Abstract

Industrial Internet of Things (IIoT) systems increasingly rely on video streams for real-time monitoring, quality control, and predictive maintenance. However, pro- cessing these video streams efficiently presents significant challenges, particularly regarding latency, bandwidth consumption, and computational costs. This thesis explores the trade-offs between edge and cloud computing for IIoT video stream processing, with a focus on optimizing performance while minimizing operational costs. The research is based on a series of controlled experiments that compare edge- only, cloud-only, and hybrid computing approaches. The experiments evaluate key performance metrics such as latency, processing speed, and bandwidth utilization. Results indicate that while cloud computing offers high processing power, it intro- duces significant network delays and higher data transmission costs. Conversely, edge computing reduces latency and decreases bandwidth usage but is limited by the processing constraints of local devices. A hybrid approach, where preprocess- ing occurs at the edge before sending refined data to the cloud for further analysis, proves to be the most efficient solution. This study demonstrates that distributing computational tasks strategically between edge and cloud environments can enhance IIoT video analytics. By preprocessing video streams at the edge, the amount of data transmitted to the cloud is sig- nificantly reduced, leading to improved efficiency and cost savings. The findings contribute to ongoing research on IIoT optimization, providing insights into how intelligent workload distribution can improve industrial video stream processing in smart manufacturing and other IIoT applications.