An Adaptive Shared Filter Ordering Algorithm for Data Stream Systems

In multimedia stream filtering scenario, there usually exist many filtering queries that specify the filtering objectives and many filters that estimate the filtering queries. A filtering query may connect to several different filters and a filter may connect to several different filtering queries. An open problem in such a filtering scenario is how to order the filters in an optimal sequence so as to decrease the filtering cost. Existing methods are based on a greedy strategy which orders the filters according to three factors of the filters, i.e., the selectivity, popularity, and cost. Although all these methods reported good results, there are still some problems that havent been addressed yet. Firstly, the selectivity factor is set empirically, which can not adaptively adjust with stream passing by. Secondly, the relationships among the three factors are not considerably explored. Under these observations, in this paper, we propose an adaptive hierarchal ordering (AHO) algorithm which executes a two-stage ordering strategy. In the first ordering stage, AHO clusters all the filters according to their popularities and costs factors. In the second stage, for each cluster all the filters are then ordered according to their selectivity. Experiments on both synthetic and real life multimedia streams demonstrate that our AHO method outperforms other simple filtering methods.