| 摘要: | 近年來無線通訊系統的進步使得廣大的使用者可以透過無線網路擷取各式各樣的資訊服務。由於無線通訊系統本身具有通訊不對稱、通訊不預期中斷、頻寬有限、網路終端裝置具有移動性與電力容量有限等特性，這些因素嚴重影響到以主從式架構設計的傳統資訊傳播機制之效能。對此，先前的研究指出，資料廣播機制能夠提高系統效能的可層次性以及網路頻寬的使用效率。然而，我們注意到，過去在資料廣播領域的研究大多是基於傳統資料管理模式，在該模式中，資料基本是靜態、持續存放和各自獨立的。然而實際上，現今很多資訊傳播服務所傳遞的資訊內容卻是動態產生的，同時資料物件本身可具有組合性與相依性，對此，之前大多數為靜態資料所設計的廣播排序方法將無法有效的套用在動態資訊傳播上。因此，本計畫將研究在廣播頻道上具動態性與時間連續性物件之排序方法設計，在步驟與方法上，我們將會分析動態資料物件在廣播時間順序上的特性，並以代數方程式表示出動態物件排序最佳化之問題；利用廣播週期性的性質推演出資訊廣播服務平均擷取時間之理論最大值與最小值的數學表示式；根據理論與定量化分析，發展出一套連續性物件排序選擇策略與收益估計函式；最後期望經該策略所動態產生的連續性物件廣播排序能夠漸進地接近最佳化的廣播排序。 ; The advance of wireless networking technologies enables the ubiquitous information delivery. However, in wireless networks several critical conditions, limited bandwidth, communication asymmetry, mobility and frequent disconnection, and battery restriction, greatly impact the traditional data dissemination systems based on the client-server paradigm, hereby resulting in scalability problem and performance degradation. Many research works have demonstrated the use of data broadcasting is scalable and can achieve better bandwidth utilization. However, we notice that in the data broadcast area, substantial studies were based on either the broadcast disks or the transaction-based data broadcast model. A data object, specifically mapped to a pair of state and value in the database, is independent, persistent, and static in such settings. Many modern information broadcast applications, however, enable the generation and spread of dynamic data. Data objects in such scenarios are associated, dependent, or dynamically generated. Almost previous efforts on static optimization of data broadcasting cannot be put into effect the dynamic data and information dissemination in distributed, mobile network systems. Therefore, our research proposal aims to study the problem of scheduling dynamic time-sequential data objects in a generalized clients-providers-servers system. First, we will transform the task of the scheduling optimization as the procedure of finding a solution in solving the algebraic problem according to the principle of periodicity. Second, we will derive the upper and lower bounds of mean service access time. Third, in light of theoretical and quantitative finding, we will develop deterministic strategies associated with some gain measure functions to on-line approximate an optimal, periodic schedule with the specific gain measure function. Finally, experimental simulations will be conducted to show performance evaluation. ; 研究期間 9709 ~ 9807 |
