基於修改線性規劃法及偏差修正法之無線感測網路定位

Abstract

在無線感測網路發展快速的現今，其應用的範圍十分廣泛，其中又以定位為一個熱門的議題。定位可以應用在室內以及戶外區域，藉由基地台接收手機或是待測物的訊號去推測出之間的距離，來求得最終待測物的位置。而室內或是大都會定位系統中，最容易造成較大定位誤差的干擾為NLOS(非直視性)情形，由於手機待測物及基地台之間有障礙物，因此使得量測的距離有過長的現象。NLOS效應之減緩可以分成兩種方向：知道統計特性及不知道統計特性。不知道NLOS統計特性情況下則是藉由非線性最小方差將NLOS偏差當作未知數估計，或是藉由一組粗估位置尋求殘餘量中帶有NLOS 偏差的訊息，並且加以估計NLOS偏差量來修正開始粗估的位置，過去也有提出利用線性規劃法利用LOS量測建立目標方程式，NLOS量測建立可行域來輔助定位。本論文將著重於不知道NLOS統計特性的情形，修改過去所提出的基於泰勒線性化並且帶限制式的NLOS偏差修正方法，藉由不考慮NLOS限制式，能有較短的求解時間以換取定位準確度，以及用泰勒線性以及新增變數法來修改線性規劃法中基於相減法的目標方程式，並且進一步提出基於泰勒線性化法之可行域之建立，改善定位準確度，最後則是將此可行域之建立，套用在現有線性規畫法之延伸，最後將藉由電腦模擬驗證我們所提出之方法。

Localization in wireless sensor network in a Non-Line-of-Sight environment, , has become an important issue recent years. NLOS (Non0Line-of-Sight) is one of the most crucial biases when localizing the mobile at indoor or urban area. It can cause large errors in our solution of localization because the measurements are much longer than the actual distance. There are two common ways to mitigate the NLOS situation: known and unknown statistics. In the past, there are some localization algorithms with unknown statistics. Use nonlinear weighted LS treating NLOS biases into unknowns with constraints. Or roughly localize the mobile pretending all the BS are LOS, and use residual concept to estimate the NLOS bias. Then correct the “rough” position. Linear programming use LOS measurements constructing cost function, and use NLOS measurements build feasible region to help positioning. This thesis will focus on the unknown statistic properties. By using different linearization methods and without constraints to modify the bias correction method. Shortening the computation time in exchange for positioning accuracy. And a significant improvement can be achieved by modifying the cost function of the linear programming method and the establishment of the feasible region. We will verify our proposed method by computer simulation.