Hierarchical algorithm for KNN query in spatial network

This paper proposes a new algorithm for finding K-nearest neighbors in large spatial networks. It employs a strategy of processing a network hierarchically. The new strategy is based on the CCAM algorithm for spatial network representation, the hierarchical approach for shortest path search and the INE algorithm for KNN search. It overcomes the problems of existing approaches: either the storage cost or the computational cost is too expensive. Our algorithm shows its potential application in large spatial networks with the requirement of real time response. A real spatial network data set, CA, will be used to compare the performance of the proposed algorithm with a Dijkstra-based approach and a pre-compute-based approach, measuring of storage cost and computational cost.