K-Closest Pairs Queries in Road Networks

Given two sets of nodes P and Q on a road network, a k-Closest Pairs Query (k-CPQ) finds the pairs from P × Q which have the k smallest network distances. Although this problem has been well studied in the Euclidean and metric spaces, this is the first time it is being investigated in the more realistic case of road networks. As our first contribution, we present a new indexing structure, named G -tree, which is designed to support our proposed algorithms. Then, we propose, as our main contribution, two different approaches for processing k-CPQs. While the first approach applies a top-down traversal paradigm by applying a best-first search strategy, the second approach looks for the k-closest pairs by traversing the G*-tree in a bottom-up manner. Both of the these approaches employ an effective pruning strategy for shrinking the search space based on the minimum network distance between sub-graphs, which is main driver for the G*-tree's construction. Finally, we investigate the efficiency of the proposed approaches under a number of different parameters using real road networks.