An Efficient Index Method for Moving Objects Databases in Fixed Networks

Abstract

Efficiency is an important subject in moving objects databases research area in recent decade. Particularly for fixed network, where many index structures have been proposed, such as the ones containing 2 D R-tree and 1D R-tree. For range query, these index structures are shown to be relatively efficient than that of 3D R-tree. But for the problem of searching the trajectory of individual moving object, it is not. In this paper, a simple but efficient index structure composed of 2D R-tree and lists of pointers is proposed, with set operations, intersection and union, on these pointers. This index structure is slightly less efficient for range query, but is significantly efficient for querying the trajectory of individual moving object. Besides, the insertion cost is kept in a small fixed range and is scalable with tree height. The problem of information overload caused by long time data accumulation can also be easily alleviated with this structure. As the empirical data shows, this scheme is one order of magnitude better than FNR-tree for trajectory query, but its insertion of each data is restricted to only 5 to 8 disk accesses.