在蜂巢式網路結合道路拓樸的架構下提供強健性的允入控制機制

Abstract

在無線通訊中，由於每個細胞基地台所配置的通道數目有限，因此如何使通道達到最有效的利用成為一個重要的問題。在此篇論文中，我們研究在蜂巢式網路結合道路拓樸的架構下，將環繞於基礎細胞(base cell)周圍鄰接細胞(adjacent cell)的通話資訊與通話位置列入考慮，利用這些資訊完成允入控制機制。對於每個細胞，我們使用二維狀態的馬可夫鏈(Markov chain)來模擬此系統，第一維代表基礎細胞使用通道的狀態，第二維是代表鄰接細胞使用通道的狀態。對於以最小化新連線的阻斷率(new call blocking probability)和連線交遞的失敗率(handoff dropping probability)為目標函數的問題，可以使用馬可夫決策過程(Markov Decision Process)來描述。在模型中，我們利用了一些變數來描述整個系統，這些變數是非線性、難以預測的，然而模型所能展現效能的優劣與這些變數估測的精準度息息相關。為了避免所取得的模型與真實系統相差甚遠，變數的估測是我們成為我們關注的主題。我們提出利用Cost Match Update(CMU)來調整參數，使參數估測能較接近實際狀態。經由實驗結果，我們證實了使用CMU確實可以使模型的效能更加提升。

In wireless communication, the number of channels allocate to each cell is limited, so how to use channel efficiently is an important topic. In our thesis, we study the call admission control in road topology based cellular networks, and we take the mobile communication information and position in adjacent cells into consideration. For each cell, we formulate the system by Markov Chain with two-dimensional states where the first dimension represents the base cell’s state and the second dimension stands for the adjacent cells’ state. The problem of minimizing a linear objective function of new call blocking rate and handoff call dropping rate can be formulated as a Markov Decision Process. In our model, we use several parameters to model our system, and these parameters are non-linear and difficult to measure and predict. How the model performs depending on whether these parameters are estimated precisely. In order to avoid our model very different from the reality, we use Cost Match Update (CMU) rule to help us to estimate parameter. In simulation result, it shows our proposed scheme has lower average cost than others.