雙層棋盤架構Zigbee網路應用於任意室內面積定位法之研究

Study on An Efficient Error-Controlled Positioning Method by Dual-Grid Zigbee Network Applied to Free Size of Indoor Area

李紹瑋(Shao-Wei Li)

指導教授 ： 黃國鼎

國立虎尾科技大學/電機資訊學院/電機工程研究所/碩士(2015年)

摘要

近年來最常用ZigBee短距離無線通訊技術來做為室內定位的研究。在不同的室內環境中使用ZigBee系統進行定位時，由於無線電波在通道傳播常會受到各式障礙物的影響，因此針對不同室內環境將各別量測出不同距離實際相對應接收訊號強度來建立該環境模型曲線，做為估算待測點距離的參考基準。藉此可更契合各環境障礙物對接收信號強度的影響，估算出更為正確之待測點定位坐標。
在本文中所使用的定位方法為本實驗室過去所提出的自我調整誤差定位法，該方法須在一預設矩形面積下進行定位，其面積大小與採用Zigbee設備之接受信號強度能力有關。在本實驗室採用之設備規格下，選擇3公尺x 3公尺與3公尺x 5公尺的矩形面積下進行自我調整誤差法之定位實驗，從定位結果得知其誤差可達85公分至15公分的高精準度。但此定位方法其最大的缺點就是只能在預設的矩形面積下定位，若室內定位之面積大於該定位矩形面積時，將無法對該室內區域進行完整之定位。所以本實驗室過去提出擴展區域定位法以棋盤式細胞拓樸來擴展單一矩形面積(細胞)的定位範圍。棋盤細胞拓樸實際上是將單一矩形面積視為一棋盤方格向外擴展組合成一棋盤，其棋盤範圍可延伸至整個室內面積。在該棋盤拓樸下其定位方法分為兩階段: 第一階段先利用判定正確定位細胞法判斷待測節點位於整個棋盤(室內面積)中之正確棋盤方格(細胞)，得知待測點所在的單一矩形面積後，第二階段再利用原來提出之自我調整誤差法進行實際座標定位。在第一階段的判定正確定位細胞法實驗中，選擇4公尺x 4公尺的單一矩形定位面積做為一個細胞尺寸，待測節點判斷正確棋盤方格(細胞)統計成功率為82.35%；而4.5公尺x 4.5公尺做為一個細胞尺寸下，其待測節點判斷正確細胞的成功率下降至76.47%。從前面兩種不同細胞尺寸的實驗結果得知，雖然其判斷成功率高達76%以上，但其數字代表尚有20%以上的失敗率。
由於過去提出之判定正確定位細胞法無法達到100%成功率，因此當該方法無法成功判定正確細胞時，便無法進行第二階段之正確定位。因此本研究為將提出一個改善棋盤式細胞拓樸缺點之雙層棋盤Zigbee網路架構。雙層棋盤拓樸之做法為在整個室內面積建立兩個棋盤拓樸，兩個棋盤分別在水平與垂直方向以1/2棋盤方格距離交錯布置於整個室內面積上。本文將提出一新的判定正確定位細胞方法可同時判斷待測節點位於此雙層棋盤中之個別正確棋盤方格(細胞)，並可選擇其中任一細胞(矩形面積)執行自我調整誤差法進行定位。由於待測節點對應此二細胞(矩形面積)內之相對位置不同，因此利用自我調整誤差法進行實際座標定位時，其定位誤差精確度會有所不同。為了待測點在可二擇一細胞(矩形面積)中挑選出定位誤差較小的細胞進行定位，本文將更進一步提出動態三邊定位參考節點選擇策略找出優化的定位參考節點，從實驗結果驗證都能選擇正確的細胞(矩形面積)，而且可正確的選擇進行定位的三角形區域，執行自我調整誤差法定位。從實驗結果得知其優化定位參考節點後，選擇定位誤差較小的細胞進行定位之優化判斷率達71%。其優化判斷率未達100%之原因從研究中將更進一步發現待測節點位於其中一細胞(矩形面積)內之定位三角形中心區域，而位於另一細胞(矩形面積)內之定位三角形邊緣區域時其擇優效果較佳，反之擇優效果較差。
為更進一步分析擇優效果較佳三角形區域之範圍，將針對三角形定位區域明確劃分可擇優區域及較無擇優效果之模糊區域。最後，將進行一隨機擺放測試點之實驗，分別於兩個區域中隨機擺放12個測試點，結果顯示其優化定位判斷率高達 83%。從統計學的觀點，當不使用動態三邊定位參考節點選擇策略時，隨機擺放待測點進行定位之優化定位判斷率應該是50%。因此本文所提出動態三邊定位參考節點選擇策略確實具有明顯的優化定位判斷效果。而且從實驗結果得知在使用自我調整誤差定位法進行定位後，誤差值具有縮小的效果。所以我們提出的研究方法除了有效擴展室內定位面積也具有縮小定位誤差變動率的效果。

關鍵字

Zigbee室內定位 ； 自我調整誤差法 ； 擴展區域定位法

並列摘要

The ZigBee wireless Network is the considerable technique to use for indoor positioning applications in recent years. When the ZigBee systems are applied in the different indoor environments, the propagation of radio waves is often influenced by the various obstacles of the environments. It results in the variation of received signal strength (RSS) with the same distance but in the different environments. In our previous work, to estimate the positions more correct by ZigBee systems, the relation cure between distance and RSS, called channel model, is proposed to create separately for each indoor environment. Moreover, to promote the positioning accuracy, a self-adjustment error indoor positioning method is also proposed. The method must be applied to a bounded size of a rectangle area under the acceptable received power sensitivity of the ZigBee equipment. In the previous work, two rectangular areas of 3M x 3M and 3M x 5M were considered to experiment with self-adjustment error indoor positioning method. The experimental results show the high positioning accuracy with the estimated errors 15 cm and only up to 85 cm. However, the acceptable rectangular area is the main weakness of the proposed method where the measured positioning area is much larger than the bounded area. In order to upgrade the proposed method to be applied in any large indoor positioning area, we proposed an effective expanded coverage scheme for self-adjustment error indoor positioning method to extend the bounded rectangular area. In the scheme, the whole measured positioning area is divided into a grid topology where each square of grid is equal to a bound rectangular area (i.e. called a cell). Therefore, the procedure of the proposed scheme includes two phases. The first phase is to judge the correct rectangular area (cell) where the measured point is located actually. In the second phase, the self-adjustment error indoor positioning method is applied to measure the point using the cell judged in phase one. In the previous experimental results, the correctly judged rate is 82.35% with 4 M x 4 M cell in addition to the correct rate 76.47% with 4.5 M x 4.5 M cell after operating the phase one. It is obvious that over the 20% failed rate will degrade the positioning accuracy as applying the method in phase 2.
To overcome the drawback that cannot achieve 100% correct judged rate by the scheme, we will propose a dual-grid topology ZigBee network. Dual-grid topology is like the grid topology to divide the whole indoor positioning area into two independent grids where the two grids are separated in the horizontal and vertical directions to 1/2 square distance of grid respectively. In the dual-grid network, we will present a new expanded coverage scheme that can always correctly judge the measured point in each rectangular area (cell) of the two grids. Then, the point is measured by using either one of the two judged cells as in phase two. Since the point is located at the different related position within the two rectangular areas (cells), the positioning accuracy of the measured point would be different by using the two cells. In order to select the better one of the two positioning accuracy to promote the capacity of the new scheme, we will further propose a dynamic trilateral reference node selected method to find the premium reference nodes for positioning. In the experimental results, it will be verified that can always select a cell (rectangular areas) where the point is always located in, meanwhile, select the premium reference nodes as the apexes of the triangle area in the cell to positioning the measured point by self-adjustment error indoor positioning method. As shown in the experimental results, that the method can select premium reference nodes up to the ratio 71%. In order to find out the reason of the ratio under 100%, we will further consider the distribution of points actually located at the various positions within the two rectangular areas (cells) to check the area of points located with premium results and area without premium results. In addition, we will randomly scatter 12 points in the two areas with/without premium results to measure. The experiment results will show the ratio up to 83% with premium results. In comparison with use of dynamic trilateral reference nodes selected method, only the new expanded coverage scheme is used to positioning the points in the dual-grid topology ZigBee network that the ratio should be 50% with premium results. Obviously, the dynamic trilateral reference nodes selected method can effectually expanding the size of positioning areas. After using self-adjustment error indoor positioning method, the original estimated errors between 15 cm and 85 cm are reduced to between 20 cm and 44 cm. Consequently, we have proposed an efficient error-controlled positioning method by dual-grid ZigBee network to suitably apply to any size of indoor areas and also promote the positioning accuracy.

並列關鍵字

indoor positioning system ； self-adjustment error indoor positioning method ； expanded coverage scheme

參考文獻

[34] 黃俊元, “用ZigBee無線感測網路做為室內自我調整誤差定位方法之研究,”國立虎尾科技大學電機工程研究所在職專班碩士論文, 100年7月。

[35] 林佳瑩,“有效擴展「無線感測網路做為室內自我調整誤差定位法」定位區域之研究,”國立虎尾科技大學電機工程研究所碩士論文, 104年1月。

[2] E. Zhang, W. Jiang, Y. Kuang, M. A. Umer, “Active RFID Positioning of Vehicles in Road Traffic,” Communications and Information Technologies (ISCIT), pp. 222-227, 12-14 Oct. 2011.

[1] A. Krukowski, D. Arsenijevic, “RFID-based Positioning for Building Management Systems,” Circuits and Systems (ISCAS), pp. 3569-3572, May 30 2010-June 2 2010.

[3] W. M. Yeung, J. K. Ng, “An Enhanced Wireless LAN Positioning Algorithm based on the Fingerprint Approach,” TENCON 2006. 2006 IEEE Region 10 Conference, pp. 1-4, 14-17 Nov. 2006.