Map-Reduce for Processing GPS Data from Public Transport in Montevideo, Uruguay

Abstract

This article addresses the problem of processing large volumes of historical GPS data from buses to compute quality-of-service metrics for urban transportation systems. We designed and implemented a solution to distribute the data processing on multiple processing units in a distributed computing infrastructure. For the experimental analysis we used historical data from Montevideo, Uruguay. The proposed solution scales properly when processing large volumes of input data, achieving a speedup of up to 22× when using 24 computing resources. As case studies, we used the historical data to compute the average speed of bus lines in Montevideo and identify troublesome locations, according to the delay and deviation of the times to reach each bus stop. Similar studies can be used by control authorities and policy makers to get an insight of the transportation system and improve the quality of service.