Enhancing Landslide Prediction: Mobile Edge Computing in Rainfall-Triggered Remote Sensing

Abstract

Landslides pose significant risks to infrastructure, human lives, and ecosystems, particularly in regions prone to heavy rainfall. Traditional prediction models rely on centralized cloud-based processing, which often suffers from latency and inefficiencies in real-time hazard assessment. This study explores the integration of Mobile Edge Computing (MEC) with remote sensing technologies to enhance the accuracy and timeliness of rainfall-triggered landslide prediction. By leveraging MEC’s localized processing capabilities, real-time sensor data—such as satellite imagery, ground-based precipitation measurements, and soil moisture indices—can be analyzed closer to the source, reducing transmission delays and improving predictive performance. We propose an adaptive framework that utilizes machine learning algorithms at the edge to assess landslide susceptibility dynamically. Comparative analysis with conventional cloud-based models demonstrates improved response times and predictive accuracy. The findings highlight MEC's potential in transforming landslide early warning systems, offering a scalable and efficient solution for disaster risk mitigation in vulnerable regions.