Abstract: Histogram is one of effective methods for construction of synopsis data structures on landmark windows over data streams. This paper presented a new framework for incremental maintenance of approximate equal-depth histograms by merging and splitting the buckets,and compared three different merge & split strategics. The experimental resups show that the algorithms are effective and efficient for continuous streaming data processing over landmark window model.

Key words: Data stream, Landmark window model, Synopsis data structure, Histogram