在無線廣播環境中，頻寬與電源的供應常受到限制，因此如何提供一資料項的廣播排程，使客戶端能快速的存取資料就顯的非常重要。在無線廣播環境中，客戶端查詢的資料項常具有關聯性，因此將有關聯性的資料項加以集中廣播，可以降低客戶端的存取時間。其中，光譜法最常被用於關聯性資料項的聚集，但是在無線廣播環境下資料項會被伺服器週期性的播放，因此在不同廣播週期的資料項亦具有關聯性。本研究針對這關聯性提出分離查詢法，藉由分離查詢法聚集不同週期間具有關聯性的資料項，以進一步的降低客戶端的平均查詢時間。並且透過實驗結果顯示若考慮不同廣播週期資料項的關聯性將能降低平均存取時間。 Because bandwidth and power are limited in wireless environment, it is important to find a broadcast data placement that client can access data quickly. In wireless broadcast environment, data items which are required by users are usually related to each other and data items which are related can be broadcasted together in order to reduce the user's data access time. Spectral Method is an effective way to cluster related data item by using the relation among data item. Since data item are broadcasted periodically in wireless environment we can modify the Spectral method and then use a divide-query algorithm to cluster the related data item in different broadcast cycle to reduce average access time. Through experiments, our results show that the access time can be further reduced by using divide-query algorithm in Spectral method.
