Authors - Anika Yadav, Ananya V Holla, Animesh Giri Abstract - With the growing use of connected devices, geographic barriers diminish, enabling communication of critical data in near real-time. As the frequency of natural disasters increases, timely communication forms the backbone of efficient disaster management. With the number of recipients varying based on the population density, any disaster notification framework must be highly scalable. Big Data technologies like Apache Kafka, Apache Flink, RabbitMQ, Apache Spark, and Apache Hadoop can be employed to enhance scalability. This study proposes a hybrid framework that classifies disasters based on severity and then routes them through one of three pipelines. This approach ensures that the Quality of Service requirements for each disaster severity type are satisfied. The pipeline assigned to handle high-severity traffic demonstrates notification delivery with latency in the order of 0.88 seconds on average across the disaster types.
