Approximate search algorithm for aggregate k-nearest neighbour queries on remote spatial databases

Searching Aggregate k-Nearest Neighbour k-ANN queries on remote spatial databases suffers from a large amount of communication. In order to overcome the difficulty, RQP-M algorithm for efficiently searching k-ANN query results is proposed in this paper. It refines query results originally searched by RQP-S with subsequent k-NN queries, whose query points are chosen among vertices of a regular polygon inscribed in a circle searched previously. Experimental results show that precision of sum k-NN query results is over 0.95 and Number of Requests NOR is at most 4.0. On the other hand, precision of max k-NN query results is over 0.95 and NOR is at most 5.6. RQP-M brings 0.04-0.20 increase in PRECISION of sum k-NN query results and over 0.40 increase in that of max k-NN query results, respectively, in comparison with RQP-S.