Understanding behavioral patterns in truck co-driving networks

Abstract

This paper examines the co-driving behavior of truck drivers using network analysis. From a unique spatiotemporal dataset encompassing more than 10 million measurements of trucks passing 17 different highway locations in the Netherlands, we extract a so-called co-driving network. In this network, nodes are truck drivers and edges represent pairs of trucks that are systematically driving together. The obtained co-driving network structure has various properties common to real-world networks, such as a dominant giant component and a power law degree distribution. Moreover, network distance metrics and community detection reveal that the network has a highly modular structure. We furthermore propose a method for understanding the network community structure through attribute assortativity. Results indicate that co-driving links are mostly established based on geographical aspects: truck drivers from the same country or the same region in the Netherlands are more inclined to drive together. The resulting improved understanding of co-driving behavior has important implications for society and the environment, as trucks coordinating their driving behavior together help reduce traffic congestion and optimize fuel usage.