Monitoring best region in spatial data streams in road networks

Abstract

Given a set of spatial objects O, the search radius r and the submodular function f are specified by the user. The best region search (BRS) is to find an optimal area with fixed range size, in which the object set has the maximum submodular function value. The BRS problem has long been studied because of its wide application in spatial data mining, facility locating and so on. However, most of the existing work focus on either Euclidean space or motionless objects, which is not applicable in many real-life cases. In this paper, we propose the best region monitoring problem in the spatial data streams in road networks (MBRS). Many real life applications can obtain benefit from MBRS problem, such as monitoring traffic and tracking in ecology. We first propose an efficient algorithm to the static BRS problem in road networks, and extend the solution to a naive method to solve the MBRS problem. Then, we put forward effective pruning strategies and branch-and-bound algorithm GER on the basis of the preprocessing to monitor best region at different times. Finally, a large number of experiments verify the efficiency of the proposed method.