Predictive Spatio-Temporal Query Processor on Resilient Distributed Datasets

Moving object prediction and indexing have been a well studied area of research and include applications in environment monitoring, traffic prediction, advertising, and efficient routing. Spark is a cluster computing framework, which utilizes Resilient Distributed Datasets (RDD) on a cluster of several commodity machines. Spark is popularly used for parallel processing of massive datasets. The modeling of cloud-based and distributed predictive spatio-temporal query processing framework for large-scale data is an interesting problem that has many practical applications. We propose a data-driven framework for moving region prediction using linear regression and distributed spatio-temporal query processing on RDDs. Our framework is designed to scale well to large-scale datasets, and process the predictive kNN and range queries with interactive query response times. Our experimental evaluation offer insight into properties of our framework and indicate that it fulfills its design goals, and enables scalable query processing for big spatiotemporal data.