A k-nearest Neighbour Query Processing Strategy Using the mqr-tree

This paper presents a k-nearest neighbour query processing strategy that utilizes a recently proposed spatial access method, the mqr-tree, to narrow down the search for candidate nearest neighbours. It requires the traversal of only one path of the tree, and in the majority of cases, does not require backtracking all the way to the root. The mqr-tree-based k-nearest neighbour strategy is evaluated both individually and compared against the brute-force method for running time. It is shown that the proposed approach performs better in many cases. In addition, the running time is not affected by the number of points being indexed or the number of nearest neighbour being sought.