NNCluster: An Efficient Clustering Algorithm for Road Network Trajectories

With the advent of ubiquitous computing, we can easily acquire the locations of moving objects. This paper studies clustering problems for trajectory data that is constrained by the road network. While many trajectory clustering algorithms have been proposed, they do not consider the spatial proximity of objects across the road network. For this kind of data, we propose a new distance measure that reflects the spatial proximity of vehicle trajectories on the road network, and an efficient clustering method that reduces the number of distance computations during the clustering process. Experimental results demonstrate that our proposed method correctly identifies clusters using real-life trajectory data yet reduces the distance computations by up to 80% against the baseline algorithm.