Efficient Group K Nearest-Neighbor Spatial Query Processing in Apache Spark

Aiming at the problem of spatial query processing in distributed computing systems, design and implementation new algorithms is a current challenge. Apache Spark memory-based framework suitable for real-time batch processing. Spark-based systems allow users to work on in-memory data, without worrying about data distribution mechanism fault-tolerance. Given two datasets points (called Query Training), group K nearest-neighbor (GKNN) retrieves (K) Training with smallest sum distances every point Query. This has been actively studied centralized environments several performance improving techniques pruning heuristics have also proposed, while, algorithm Hadoop was recently proposed by our team. Since, general, exhibits lower than Spark, this paper, we present first GKNN compare it against one Hadoop. incorporates programming features facilities that are specific Spark. Moreover, improve applicable incorporated. The results an extensive set experiments real-world presented, demonstrating solution, its improvements, efficient clear winner comparison