Large-Scale Water Quality Prediction Using Federated Sensing and Learning: A Case Study with Real-World Sensing...

Area of Science:

• Environmental Science
• Data Science
• Machine Learning

Background:

• Green tides represent a significant water pollution challenge.
• Complex interactions between flow rate, water quality, and weather drive green tide formation.
• Current prediction methods are inadequate for accurate forecasting and damage mitigation.

Purpose of the Study:

• To develop an advanced system for predicting green tide occurrence.
• To minimize potential damage by enabling early detection of green tides.
• To address limitations in existing green tide prediction methodologies.

Main Methods:

• Implementation of a novel network model utilizing smart sensor-based federated learning.
• Leveraging distributed observation data from geographically dispersed local models.
• Design of an optimal scheduler to efficiently process real-time big data arrivals.

Main Results:

• The proposed scheduling algorithm enhances data usage efficiency.
• The system demonstrates improved performance in green tide occurrence prediction models.
• Experimental validation using real-world water quality big data confirms the algorithm's effectiveness.

Conclusions:

• The developed federated learning system offers a robust solution for green tide prediction.
• The optimal scheduling algorithm significantly boosts network system efficiency.
• This approach provides a valuable tool for proactive management of water pollution events.