Efficiently monitoring nearest neighbors to a moving object

Abstract

Continuous monitoring k nearest neighbors in highly dynamic scenarios appears to be a hot topic in database research community. Most previous work focus on devising approaches with a goal to consume litter computation resource and memory resource. Only a few literatures aim at reducing communication overhead, however, still with an assumption that the query object is static. This paper constitutes an attempt on continuous monitoring k nearest neighbors to a dynamic query object with a goal to reduce communication overhead. In our RFA approach, a Range Filter is installed in each moving object to filter parts of data (e.g. location). Furthermore, RFA approach is capable of answering three kinds of queries, including precise kNN query, non-value-based approximate kNN query, and value-based approximate kNN query. Extensive experimental results show that our new approach achieves significant saving in communication overhead. © Springer-Verlag Berlin Heidelberg 2007.