實測式智慧型手機資料建模以及LTE DRX省電機制之效能分析

Abstract

在今日的通訊網路中，可攜式裝置，如智慧型手機、平板電腦大量成長並且佔據了網路世界的重要地位。 這些裝置藉由應用程式產生與傳統的通話、網頁瀏覽不同的封包資料， 因此能提供使用者更好的使用體驗。然而，這些有別於傳統資料的應用程式封包使得裝置的電池快速耗盡，有時候甚至是在使用者並未操作該裝置的情形下耗盡。 為解決類似的電量問題，多數通訊系統都有相應的省電機制，但這些機制對於手機的應用程式所產生的資料來說可能不是最佳的。在這篇論文中我們研究了智慧型手機應用程式的資料特性，特別是週期性 更新的應用程式。我們捕捉從應用程式發出或是接收的封包，觀察其行為，將之分類，並且提出了週期性更新的數學模型。利用這個模型，我們進一步分析LTE系統中的省電機制─不連續接收 機制(DRX)─的效能，然後針對封包延遲、休眠比例等等效能評測指標的最佳化問題，提出最佳的DRX機制參數解。

Nowadays portable devices such as smartphones, tablets have lead to explosive growth in traffic over communication networks. These devices generate traffic from applications, which behave differently from traditional web browsing and voice call, and thus allow the device users to have better experience. However, the non-traditional application traffic drains the device battery quickly, even when devices are without user intervention. To resolve such energy problem, most communication systems have power-saving mechanisms, but these mechanisms may not be optimal for traffic generated by applications run on smartphones and tablets. In this paper, we study the traffic properties of smartphone applications, especially the periodically-update applications. We capture packets sent and received by applications running on smartphones, observe their behaviors, classify applications, and then provide mathematical model for periodically-update applications. With this model, we analyze the performance of Discontinuous Reception (DRX) mechanism in 4G LTE system, and then provide solutions to two DRX parameter-optimization problems constrained on delay and device sleep ratio, respectively.