Integration of Federated Learning and Edge-Cloud Platform for Precision Aquaculture
Wai Khuen Cheng, Jia Cheng Khor, Wei Zheng Liew, Khean Thye Bea, Yen‐Lin Chen
Abstract
This paper addresses the escalating demand for aquaculture products by proposing an edge-cloud solution for precision aquaculture. Despite technological advancements, challenges persist in the aquaculture industry, including labor shortages and inaccurate data analysis. The project focuses on water quality monitoring, employing IoT sensors in an edge computing environment to collect real-time parameters and using a camera module to analyze the growth of prawns. Local processing minimizes bandwidth requirements and latency, with data transmitted to the cloud for analysis and graphical visualization. In this paper, we present a unique edge-cloud architecture for precision aquaculture tailored to small-scale and resource-constrained farmers. The paper introduces a federated learning framework, utilizing a distributed computing architecture, where the edge environment acts as the client node, training a local machine learning model. The cloud server aggregates model weights from diverse clients, creating a robust global model without compromising individual data privacy. This comprehensive solution aims to improve aquaculture efficiency, address production challenges, reduce high consumption of network bandwidth, and ensure sustainable practices.