無線網路下以細胞格為主的定位方法

Abstract

在無線網路下決定節點的座標位置是一個很有挑戰性的問題。這個問題十分的重要，尤其是對於體積很小的節點，如感測節點。因此，有很多的研究都在探討無線網路下的定位問題。節點定位的問題大致可分成兩大類：集中式及分散式定位方法。集中式定位方法需要有中央控管的伺服器，其功能是負責接收節點所偵測資料及其定位的要求，並根據此資訊執行定位演算法以決定節點位置資訊並回傳給發出詢問的節點。此方法的優點是系統架構簡單且節點負擔較輕。但缺點為伺服器與節點間的通訊成本大，且伺服器的運算負載重。另一方面分散式的定位方法，讓節點可以自行完成定位而不需依靠伺服器的協助。本篇論文針對規則及不規則的網路架構下，提出一個利用訊號涵蓋範圍區域重疊的特性來讓節點自行估計目前的位置。節點只需接收來自參考節點的訊框資訊，並根據收集的資訊來執行簡單的運算即可獲得位置資訊。我們討論所提方法的定位能力及對於不同環境下，評估系統的效能。另外，為了進一步提升定位系統的準確度，我們將具單一訊號強度的參考節點改採用具有多重訊號強度的參考節點。將此改進方法應用於各種嚴苛的狀況，如參考節點損壞、訊號不穩定、訊框封包遺失等，所提出的方法均能維持良好的運作。對於這樣的結果，我們相信將有助於提供一個簡單、準確且低成本的定位系統。

One challenging issue in wireless networks is to determine where a given node is physical located. This problem is especially crucial for very small nodes such as sensor nodes. Therefore, many researchers have given much attention to positioning problem of wireless networks. This problem can be solved by centralized or distributed positioning methods. The centralized positioning methods need a central server to receive the sensed data, accept positioning queries, perform positioning algorithm and reply the coordinate back to the querying node. The advantages of centralized methods are that their system architecture is simple and the computing load of querying node is light. However, both of the communication overhead between central server and querying node and computational load of central server are very heavy. Therefore, the distributed positioning methods are presented to provide a self-positioning mechanism to reduce these defects. This dissertation develops a distributed positioning method using transmission signal overlapping region that called as cell-based positioning method in regular and irregular network structures. The node receives beacon frames from reference node(s) and performs simple operation to estimate its position. We discuss the characteristics of the proposed method and evaluate the performance in different environment. To improve the positioning accuracy, the radio model with multiple power-levels is applied in the proposed positioning method. In critical situations such as reference node failure, unstable radio transmission range and beacon collision, the proposed method still performs well. We believe that the results are useful for designing a simple, accurate, and low cost positioning system for wireless networks.