A PARALLELIZED DATA PROCESSING ALGORITHM FOR MAP MATCHING ON OPEN SOURCE ROUTING MACHINE (OSRM) SERVER

This paper discusses the multiprocessing algorithm, used for improving the performance of code that processes data on Open Source Routing Machine (OSRM) Server. OSRM is a routing engine that provides (shortest) routes between origins and destinations on Open Street Map (OSM) based road networks. The Nearest Service by OSRM may be used to map GPS data on OSM road networks. The latitudes and longitudes (coordinates) from Floating Car Data (FCD) are sent to the OSRM server using nearest API to find the pair of nodes for the specific coordinates. The performance of data preparation process depends on many factors. For the OSRM server available online, it takes tremendous amount of time therefore a local setup of server is a preferable choice. Even on a local server, it takes a considerable time to prepare large scale data. It is important to optimize the data fetching algorithm so that huge records can be prepared easily using the Open Source Routing Machine. A two-prong strategy has been discussed in this paper. Firstly, the performance can be improved by running processes in parallel and utilizing the maximum number of cores. The Second improvement is achieved by splitting files to avoid the memory swapping process. The parallelization process yields pronounced results which have been discussed in this paper.