Parallelized Multilevel Arc Flags Improve Speedup in Shortest Path Queries

Abstract

Tremendous growth in online applications require shortest path queries to solve its problem where the data volume is very high. Among the various shortest path algorithms considered, Dijkstra's shortest path algorithm[1] proves to have better performance compared to other algorithms. The speedup techniques when applied to these shortest path algorithms will improve the performance of the system. The combined speedup techniques have higher impact over the shortest path queries. In this paper a new combined speedup technique based on arc flags and multilevel method(COAM) is implemented and tested on random graphs, planar graphs, and real world graphs. The pre-processing phase of arc flag and multilevel method is parallelised using OpenMP parallel programming constructs and effect of that produces an improved speedup in the individual basic speedup techniques and combined speedup technique(COAM).