Aggregate Nearest Neighbors Queries with Service Time Constraints in Time-Dependent Road Networks

In this paper, we consider the problem of processing aggregate nearest neighbor (ANN) queries in time-dependent road networks (TDANN) taking the service time constraints of points of interest (POIs) into account. Given a time-dependent road network where travel time along each edge is a function of the departure time, TDANN query is to find the nearest POI which minimizes the time-dependent aggregate travel time from each query point to the POI and the waiting time to be served. The existed Time-Dependent Aggregate Nearest Neighbor Hub Label( TD-ANNHL) algorithm for answering TDANN query assumes that each POI is served for 24 hours and have no waiting time, and the algorithm is inefficient when POIs have service time constraints due to the lack of consideration of the waiting time. To solve the TDANN query under service time constraints, an improved TD-ANNHL algorithm based on service time heuristic values (TD-ANN-STC) was proposed. In the process of generating the candidate set phase, TD-ANN-STC will search the nearest POI in the lower bound time-dependent road network, and those POIs who cannot provide service in time based on the arriving time will be filtered out which could reduce the online expansion area. In the verification phase, A* algorithm is utilized to compute the estimation of time-dependent travel time from each query point to the candidate POIs. To find the best POI, both the estimation of travel time and waiting time are considered by the heuristic function to avoid expanding the network to POI that can be reached quickly but cannot be served immediately. The experimental results show that the number of expanded nodes of TD-ANN-STC is 54.52% lower than the TD-ANNHL algorithm, and the processing time of TD-ANN-STC is 73.61% lower than TD-ANNHL.