Temporal Range Exploration of Large Scale Multidimensional Time Series Data

• Authors: Joseph JaJa;Jusub Kim;Qin Wang
• Venue: SSDBM '04 Proceedings of the 16th International Conference on Scientific and Statistical Database Management
• Year: 2004

We consider the problem of querying large scale multidimensionaltime series data to discover events of interest,test and validate hypotheses, or to associate temporalpatterns with specific events.Large amounts of multidimensionaltime series data are currently available, andthis type of data is growing at a fast rate due to the currenttrends in collecting time series of business, scientific, demographic,and simulation data.The ability to exploresuch collections interactively, even at a coarse level, willbe critical in discovering the information and knowledgeembedded in such collections.We develop indexing techniquesand search algorithms to efficiently handle temporalrange value querying of multidimensional time seriesdata.Our indexing uses linear space data structuresthat enable the handling of queries very efficiently, invokingin the worst case a logarithmic number of queriesto single time slices.We also show that our algorithmis ideally suited for parallel implementation on clustersof processors achieving a linear speedup in the number ofavailable processors.A particularly simple data structurewith provably good bounds is also presented for the casewhen the number of multidimensional objects is relativelysmall.These techniques improve significantly over previoustechniques for either the serial or the parallel case,and are evaluated by extensive experimental results thatconfirm their superior performance.