An intelligent edge computing-based scalable architecture for large-scale smart farm system

Abstract

The recent smart farm system, which can be controlled remotely and automatically, is demanding for an intelligent strategy due to its limitations in data processing and storage capacity. Therefore, cloud server computing approach is generally used for such functions with complex algorithms, but the explosive increase of data traffic causes network congestion and overhead in central cloud. To overcome the limitations of the cloud-centric systems, the concept of edge computing has become important. In this paper, we propose an intelligent edge computing-based scalable architecture for smart farm, which is based on two-stage control system for intelligent management using conditional and correlation analysis. Since the cloud server collects preprocessed data from correlation stage, it processes smaller capacity data than the original of the edge computer, thereby reducing the congestion of the entire network and the processing burden on the server. It is possible to determine the optimal next state of farm environment according to the growth of the crop from the previous state in consideration of the correlation of input parameters using deep learning model. Cooperation or standalone management of large-scale farmland can be possible using network interface with lower conditional control stage and intelligent filtered data from upper cloud server. Our proposed edge computing-based architecture can be referenced as an architecture framework for intelligent smart farm standardization in the future, with scalable structure that can be connected and extended to other correlation control stages of edge computers.