個人通訊網路之智慧型位置追蹤策略

Abstract

傳統的位置管理策略是指用在GSM上的位置管理方法，而非傳統的位置管 理策略則是被提出來改進傳統的位置管理策略。非傳統的位置管理策略讓 有線網路端增加一些判斷任務，以大量降低位置更新所需要的無線訊號頻 寬。當手機的移動模式有很高的可預測性，非傳統策略會記錄手機的相關 資訊來做預測，以節省大量的系統資源。然而，非傳統策略只適用於長期 且固定的移動路徑。若手機改變其移動習性或遇到一些短期的突發事件， 非傳統策略將要花費較多的位置追蹤成本。在本篇論文中，我們提出一種 降低位置追蹤成本的新方法，以適用於這些情況，其主要的想法是考慮手 機近期的移動資訊（例如上一個所在位置區域和進入此區域的起始時間） ，即所謂的呼叫資訊記錄，來決定那一個位置區域要先被呼叫。我們用一 個能反應實際城市環境的城市區域模式來模擬並作效能分析。實驗數據顯 示我們的方法與傳統和非傳統策略相比較，可以分別降低14%-55% 和6%-39% 的位置管理訊號成本。除了維持與非傳統策略一樣低的位置更 新成本外，我們的智慧型位置追蹤策略比非傳統策略有較低的位置追蹤成 本。我們的方法只需增加額外的儲存空間和計算時間來分別儲存和處理手 機的資料及呼叫資訊記錄。然而，若與降低無線訊號流量以得到較多的使 用頻寬來比較，額外的儲存空間及計算時間需求對現今記憶體的大容量與 處理機的快速度來說是較不重要的。 An alternative strategy (AS) for location management has been proposed for improvingthe classical strategy (CS) for location management used in GSM. It significantlyreduces the location update signaling traffic on the radio link by increasing theresponsibility of the fixed network. The AS stores the mobile related informationand brings great savings in system resources when a mobile host (MH) has a highpredictable mobility pattern. However, the AS is only suitable for long term eventsand fixed movement tracks. It's location tracking will cost more when the MH changesits movement habit or encounters some short term events. In this paper, we proposea novel method which aims at reduction of signaling overhead resulting from locationtracking in the above situations. The key idea is taking the recent user movementinformation (i.e., last connection's location area and start time of the lastconnection), that is called paging information record, into account to determinewhich location area to be paged first. Performance analysis has been conducted usinga city area zone model which can model a realistic city area environment. Experimentalresults show that our method can reduce 14\% - 55\% and 6\% - 39\% location management's signaling cost compared with the CS and AS, respectively. Besidesmaintaining low location update cost as that of AS, our intelligent location trackingstrategy achieves a significant reduction in location tracking cost than the AS. Theoverhead of our approach is the additional storage space required (for storing MHprofiles and paging information records) in network databases and the additionalprocessing time of profiles and paging information records. However, neither the storage nor the processing time requirements can be considered significant in termsof today's memory capacity and processor speed as far as the reduction of signalingtraffic implying more bandwidth available.