在無測感測網路中，定位問題是一個基本且重要的研究議題。許多定位方法假設少部分感測器(錨節點,anchor node)擁有位置座標資訊，其他感測器(一般節點, normal node)利用收集周圍錨節點位置資訊，進而計算出自己所在位置。大多數的方法是針對靜態式感測網路所設計，有些方法也會利用移動的錨節點廣播訊息來增加定位精準度，但是一般節點仍然是靜止不動的。然而，在許多應用當中利用移動的節點增加感測與偵測的區域，進而節省所需的感測器成本。因此，針對移動式感測網路所設計的定位方法是必須的。這篇論文，我們提出一個新的定位方法改善先前所提出的蒙地卡羅定位法(Monte Carlo Lolicazation scheme, MCL)。在我們提出的方法中，一般感測器藉由收集錨節點的實際座標和鄰近的座標所預估的位置，進而推算出自己的座標位置。除此之外，我們提出一個預測節點的移動方向的方法，並藉此降低定位誤差。模擬結果顯示，不論在不同的節點移動模式或不同的移動速度之下，我們所提出的方法都比MCL擁有較準確的估計位置。 Localization is a fundamental and important research in wireless sensor networks. Most localization schemes focus on static sensor networks. Some of them increase the accuracy by using mobile anchor nodes, but normal nodes are still static. However, mobile sensors are required in some applications such that the sensed or monitored area can be enlarged. Thus, a localization scheme designed for mobile sensor networks is necessary. In this thesis, we propose a localization scheme to improve the Monte Carlo Localization (MCL) scheme [1]. In our scheme, the normal nodes can estimate their own locations by gathering the positions of neighboring anchors and the estimative locations of neighboring normal nodes. In addition, we propose a scheme to predict the moving direction of sensor nodes to increase the localization accuracy. Simulation results show that the localization error in our proposed scheme is less than the MCL scheme in various mobility models and moving speeds.
