Data schedule in broadcasting is playing a more importance role due to increasing demand for large client popularity and vast amount of information. For system with multipoint queries, data records which queried by same query are broadcasted contiguously to reduce the average access time. Several techniques have been used in clustering data by defining the affinity between them. The data affinity function defined was mainly aiming at minimizing the linear Query Distance. However, our work showing that in order to minimize the average access time, the objective function shall be in quadratic form. We propose a MinimumGap algorithm(MG) which merge relevant segments base on this new affinity function. Through extensive experiments, the results show not only the query’s access time can be reduced by using this new affinity function, by using a dummy segment to speed our algorithm, the scheme we proposed have significant saving on both time complexity and memory space complexity.
關聯:
Lecture Notes in Computer Science vol. 3391 pp.390-400
