Analysis of ndimensional quadtrees using the Hausdorff fractal dimension.pdfVIP

Analysis of ndimensional Quadtrees Using the Hausdor Fractal Dimension  Christos Faloutsos Volker Gaede Department of Computer Science Institut fur Wirtschaftsinformatik University of Maryland Humb oldtUniversitat zu Berlin College Park MD Spandauer Str christoscs umdedu Berlin Germany gaedewiwi hub erlinde February Abstract There is mounting evidence Man Sch that real datasets are statistically selfsimilar and thus fractal This is an imp ortant insight since it p ermits a compact statistical description of spatial datasets subsequently as we show it also forms the basis for the theoretical analysis of spatial access metho ds without using the typical but unrealistic uniformity assumption In this pap er we fo cus on the estimation of the numb er of quadtree blo cks that a real spatial dataset will require Using the the wellknown Hausdor fractal dimension we derive some closed formulas which allow us to predict the numb er of quadtree blo cks given some few parameters Using our formulas it is p ossible to predict the space overhead and the resp onse time of linear quadtreeszordering OM which are widely used in practice In order to verify our analytical mo del we p erformed an extensive exp erimental investigation using several real datasets coming from dierent domains In these exp eriments we found that our analytical mo del agrees well with our exp eriments as well as with older empirical observations on d Gaeb and d