Location–allocation strategies for traffic counters—a citywide deployment

Abstract

This study presents an innovative geostatistical approach, namely network regression kriging (NRK), for optimizing location–allocation of traffic counters within a road network. Long-term traffic observations for different road types are utilized by combining several semivariograms to better represent traffic behaviours. This location optimization problem minimizes the NRK estimation variance of traffic volumes via the simulated annealing algorithm. A real-world case study in Edmonton, Alberta, Canada, was used to illustrate the distinct features of our proposed approach. The results suggest that the generated traffic counters’ configurations can significantly improve the estimation accuracies. An effort was also made to reduce the number of traffic counters while maintaining the same estimation accuracies, and the result indicated potential equipment savings for up to 46.13%. The findings of this study imply that the proposed approach could be used as a decision-support and performance evaluation tool for transportation authorities to deploy traffic counters and assess alternative deployment strategies.