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

Abstract

Landslides pose significant threats to infrastructure, ecosystems, and human lives, particularly in regions prone to intense rainfall. Traditional landslide prediction models often suffer from delayed data processing, limiting their real-time applicability. This study proposes a Mobile Edge Computing (MEC)-enabled landslide prediction framework that integrates remote sensing data, machine learning models, and real-time meteorological observations to enhance prediction accuracy and efficiency. By leveraging MEC, computational workloads are distributed to edge nodes near the data sources, reducing latency and enabling rapid decision-making. The proposed system processes rainfall-triggered landslide events using a hybrid deep learning model that combines Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks for spatial-temporal analysis. Synthetic Aperture Radar (SAR) and optical remote sensing imagery are fused with historical rainfall patterns and soil moisture data to improve predictive performance. A case study in landslide-prone regions demonstrates the effectiveness of the model, achieving a significant improvement in prediction accuracy compared to conventional centralized computing approaches. Performance evaluations reveal that the MEC-based framework reduces computational latency by 35% and increases prediction accuracy by 18%, ensuring timely alerts for disaster management authorities. The results suggest that integrating edge computing with AI-driven remote sensing analytics offers a scalable and real-time solution for landslide risk mitigation. This research contributes to disaster resilience strategies by enabling early warning systems that optimize resource allocation and minimize socio-economic disruptions caused by landslides.