Distributed Searching of Multi-dimensional Data: A Performance Evaluation Study

In this paper we present a data structure for searching in multi-dimensional point sets in distributed environments and discuss its experimental evaluation also through a comparison with previous proposals. The data structure is based on an extension of k-d trees. The technological reference context is a distributed environment where multicast (i.e., restricted broadcast) is allowed, but it is also shown how to avoid using it. The data structure supports exact, partial, and range search queries with a complexity that is optimal in a distributed sense. The set of multidimensional points is managed in a scalable way, i.e., it can be dynamically enlarged with insertion of new points. We also propose new performance measures for the comparative evaluation of the efficiency with which a data structure is distributed over a communication network. © 1998 Academic Press