Optimization approaches to mpi and area merging-based parallel buffer algorithm

Abstract

On buffer zone construction, the rasterization-based dilation method inevitably introduces errors, and the double-sided parallel line method involves a series of complex operations. In this paper, we proposed a parallel buffer algorithm based on area merging and MPI (Message Passing Interface) to improve the performances of buffer analyses on processing large datasets. Experimental results reveal that there are three major performance bottlenecks which significantly impact the serial and parallel buffer construction efficiencies, including the area merging strategy, the task load balance method and the MPI inter-process results merging strategy. Corresponding optimization approaches involving tree-like area merging strategy, the vertex number oriented parallel task partition method and the inter-process results merging strategy were suggested to overcome these bottlenecks. Experiments were carried out to examine the performance efficiency of the optimized parallel algorithm. The estimation results suggested that the optimization approaches could provide high performance and processing ability for buffer construction in a cluster parallel environment. Our method could provide insights into the parallelization of spatial analysis algorithm.

Approach to Accelerating Dissolved Vector Buffer Generation in Distributed In-Memory Cluster Architecture

“…The first strategy is based on the MPI (message passing interface) protocol, which supports the parallel execution of buffer analysis algorithms on HPC (high-performance computing) cluster computers [14][15][16][17] but suffers from weak extensibility when faced with the big spatial data scenarios. The second strategy comes from the distributed high-performance computing framework for big data, MapReduce and Spark, which have recently gained much attention and popularity in the field of geoanalysis.…”

“…The solution was applied to a two-node self-build cluster with a dataset containing 200,000 features. Fan [17] proposed a parallel buffer algorithm based on area merging and vertex partitioning to improve the performances of buffer analyses when processing large datasets. Fan implemented the algorithm in the MPI programming model, which is the mainstream architecture in high-performance computing.…”

“…We found that space-filling curves have good advantages in terms of the aggregation of neighboring spatial objects when compared with an ordinary grid [14]. Space-filling curves in parallel buffer analysis has not been addressed in recent research [16,17,32], though in his literature [17], Fan presented this topic as a future research direction. Therefore, we present a more efficient partitioning method based on space-filling curves, which essentially contains two steps.…”

“…This strategy is practically supported in two ways. Primarily, the splitting-and-conquer method is one of the ways to create a buffer zone [17]. Moreover, the topology relationship between split lines is insignificant for the dissolve-type buffer result.…”

“…Second, a single high-density partitioned series D may require a lengthy calculation time, thereby causing processing congestion. To solve this problem, an even decomposition strategy, which decomposes the data according to the number of vertices, is utilized [16,17,32]. Our goal is to minimize the chances of producing a single large data block as well as control the total number of data blocks.…”