Abstract
Tsunami early warning systems are critical for minimizing the impact of seismic events on coastal communities. However, existing systems face challenges in terms of latency, scalability, and real-time data processing. This research introduces RT-EDAP (Real-Time Event-Driven Data Aggregation and Processing), a novel framework designed to address these issues by enhancing the efficiency of tsunami prediction and alert dissemination. The primary objective of this study is to develop a robust, low-latency, and scalable TEWS that can effectively handle multi-source data streams and deliver timely tsunami alerts. RT-EDAP utilizes Edge Computing and Stream Processing Frameworks, such as Apache Kafka and Apache Flink, to process data locally at edge nodes, reducing latency and optimizing network bandwidth usage. The event-driven architecture prioritizes computational resources for critical seismic anomalies, ensuring fast and accurate detection of tsunami events. The framework integrates real-time data from seismic sensors, tide gauges, and GPS, and employs lightweight edge models combined with centralized machine learning techniques, such as Temporal Convolutional Networks (TCNs), to improve event classification accuracy. The system's performance is evaluated using key metrics, including latency, throughput, scalability, and prediction accuracy. The results demonstrate that RT-EDAP achieves high accuracy (95%) and low processing latency (50-60ms), outperforms traditional methods, and can scale to handle high event rates. In conclusion, RT-EDAP is a scalable, fault-tolerant solution that enhances tsunami early warning systems by improving real-time data processing and response times, offering a significant advancement in disaster management capabilities.