A Linked Neighboring Leaves N-Tree to Support Distance Range Search

摘要

Mobile query processing is, actually, a very active research field. Range and nearest neighbor queries are common types of queries in spatial databases and location based services (LBS). In this paper, we focus on finding nearest neighbors of a query point within a certain distance range. An example of query, frequently met in LBS, is "Find all the nearest gas stations within 2 miles neighborhood of his/her current location ". We propose two approaches for answering such queries. Both are based on a recent indexing technique called N-tree. The first one is a branch and bound approach, whereas the second, called 'neighborhoods scanning', is based on a variant of N-tree, Leaves-Linked N-tree (LLN-tree). LLN-tree is an index tree structure that avoids visiting multiple paths during range search. Both techniques are presented, illustrated and evaluated. Experiments show that the latter approach outperforms the former in response time and disk access as well.