Multi-core parallelism for plane sweep algorithms as a foundation for GIS operations

Abstract

The plane sweep algorithm is a foundational algorithm for many geometric and spatial computations; thus, improvements in the algorithm have far reaching effects in many applications. In this paper, we examine the performance of the serial plane sweep algorithm, and introduce a parallelization technique for the algorithm that is suitable to multi-core computers. The parallelization technique is described in detail and shown to be correct. Finally, experiments are performed using multiple data sets on computers with varying numbers of processing cores. We show that our algorithm achieves significant speedups over the serial plane sweep algorithm using a wide range of input parameters; thus, our algorithm achieves good performance without the need to tune the input parameters for specific input cases.