Edge Cloud Computing for Geospatial Data Processing and Approximate Queries

Architecture for optimizing geospatial data processing pipelines in the cloud by making use of edge nodes deployed on containers in an urban moving taxi scenario (specifically Shenzhen, China). Edge nodes are using Geohash for efficient data preprocessing, including Geohash-based stratified sampling, and neighborhood location of incoming messages. Apache Kafka was then used to send data to a Spark cluster using a spatially-aware technique for data distribution. In particular, a Kafka topic for each neighborhood of the city considered was created, and each of these topics contained only messages originated in the same neighborhood.