在無線感測網路中，許多應用皆需要用到位置資訊以完成任務。近幾年已有許多針對移動式感測網路的定位演算法被提出，其中比較常見的定位方法為根據蒙地卡羅演算法而設計。這些基於蒙地卡羅方法的演算法需要產生樣本點以表示節點的位置，然而，產生並且維持這些樣本點的成本相當高。這篇論文中，我們提出一個非基於蒙地卡羅的定位方法，因此我們可以避免產生樣本點的花費。在我們提出的方法中，當一般節點(沒有位置資訊的節點)位於錨節點(已知位置資訊的節點)的通訊範圍內時，我們分析錨節點與一般節點的相對移動方向，並且利用這個資訊幫助一般節點定位。此外，我們也分析當一般節點鄰居只有單一錨節點的情形。模擬結果顯示我們的計算複雜度低於蒙地卡羅演算法，同時定位誤差也維持在可接受範圍。In wireless sensor network (WSN), location information is required to achieve the goal in many applications. Many localization algorithms are proposed to deal with the mobile sensor network in the past year. It is common that existing localization algorithms are based on the Sequential Monte Carlo (SMC) method. The SMC methods need to generate the samples to estimate the location of the sensor node. However, generating and filtering the samples is cost for each sensor node. In this paper, we propose a non-SMC distributed localization method. Thus, we can avoid spending computation cost on generating samples. In our proposed scheme, we analyze the relation between the moving direction of anchor node and normal node when the normal node (with unknown location) encounters the anchor node (with unknown location), and use the information to calculate the position of normal node. Besides, we also discuss the situation that normal node only has single neighboring anchor node in order to enhance the localization accuracy. The simulation shows that the computational complexity in our proposed method is lower than Monte Carlo Localization algorithm, and we also have acceptable localization error.
