無線感測網路下以差分進化演算法提升區域定位準確性之研究

本論文中使用人工智慧(Artificial Intelligent, AI)室內定位(Indoor Positioning)演算法，提升無線感測網路之室內區域定位準確度。本文採用樣本比對法(Pattern matching)來做室內定位，因為樣本比對法無需量測待定位目標點與定位參考點之間的距離，而是比對待定位目標點的訊號強度與環境樣本中訓練點的訊號強度並估測目標點位置，但在室內定位中，訊號傳遞容易受到環境的影響造成訊號不穩定而影響定位的準確度。本論文使用訊號強度指標(Received Signal Strength Indicator, RSSI)建立訊號傳播通道模型與環境通道模型，考慮到訊號在實際室內環境會受到遮蔽效應與距離影響，我們在通道模型中加入遮蔽效應與路徑損耗效應，並藉由調整高斯隨機變數標準差模擬簡單環境與複雜環境。本文首先使用K-最近鄰居(K-Nearest Neighbor Average, KNN)演算法，因為RSSI的浮動以及鄰近區域RSSI相似，會造成KNN區域定位的效能不佳，所以我們提出KNN結合差分進化演算法(Differential Evolution, DE)DE-KNN進行區域定位，而DE是建構於將個體之間的差異加到另一個個體中，因此RSSI的浮動較小，能有效提升區域定位的效能，本論文將探討在6m×6和3m×3m的室內環境中定位參考點數量以及K-NN的K值對於區域定位的準確性，並與我們提出KNN結合DE區域定位方法，對兩種方法進行效能分析。另外本論文使用Matlab GUI建立人機介面，可以在此人機介面中，調整不同參數建立不同通道模型與樣本資料庫，也可以畫出通道模型波形圖，並於定位模擬人機介面中可以清楚看見整個模擬定位過程，不論是待測定位目標點實際位置以及定位演算法所估測目標點的位置都能顯示於人機介面中。

關鍵字

無線感測網路；區域定位；最近鄰居演算法；差分進化演算法；接收訊號強度

並列摘要

This thesis aims to utilize the artificial intelligence algorithm to enhance indoor area positioning accuracy for Wireless Sensor Networks (WSNs). In indoor positioning, signals would be affected by environment during signal transmission. This thesis uses the pattern matching to do indoor area positioning. Pattern matching does not need to measure the distance between target node and positioning reference nodes but compare target RSSI and training points RSSI in environment model to estimate the target location. In thesis, we utilize received single strength indicator (RSSI) to establish the single transmission channel model and simple database. Considering that the signal is affected by the shadowing effect and distance in the reality indoor environment, we added the effect of shadowing and path loss in our channel model and adjusted the Gaussian random variable standard deviation to simulate a simple environment and complex environment. In this thesis, we compare the K Nearest Neighbor Average (KNN-AVG) algorithm with the KNN combine Differential Evolution KNN (DE-KNN) algorithm on the indoor area positioning accuracy. We discussed how many positioning reference nodes and K values to improve the positioning accuracy in 6 by 6 and 3 by 3 meter indoor environment. However, if the area RSSI is similar to the neighbor area RSSI, KNN won’t estimate the incorrect area. Therefore, we propose the KNN combined with the DE method to improve KNN positioning accuracy. DE is constructed by difference between the individual signals and the other individual signals. To see if it can improve the accuracy of area positioning. When using Matlab GUI to establish the man-machine interface, we can adjust the parameters to establish the different model of the channel and the sample database. Through the interface the positioning simulation, we can obviously understand how the interface process, both the reality target location and the algorithm which estimates the target location can be display on the interface.

並列關鍵字

WSNs ； Area localization ； KNN ； DE ； RSSI

參考文獻

[1] Élodie Morin, Mickael Maman, Roberto Guizzetti, and Andrzej Duda, “Comparison of the Device Lifetime in Wireless Networks for the Internet of Things,” IEEE Access, Vol. 5, pp. 7097-7114, April 2017.

[7] Yehua Wei, Tun Chen and Wenjia Li, “An Indoor Localization Algorithm Based on Dynamic Measurement Compressive Sensing for Wireless Sensor Networks,” 2015 International Conference on Identification, Information, and Knowledge in the Internet of Things (IIKI), 22-23 Oct. 2015

[9] Pratap Kumar Sahu, Eric Hsiao-Kuang Wu, and Jagruti Sahoo, “DuRT: Dual RSSI Trend Based Localization for Wireless Sensor Networks,” IEEE Sensor Journal, vol. 13, pp. 3115-3223, Aug 2013.

[10] Suining He and S.-H. Gary Chan, “INTRI: Contour-Based Trilateration for Indoor Fingerprint-Based Localization,” IEEE Transactions on Mobile Computing, Vol. 16, pp. 1676-1690, 1 June 2017

[12] Yaqin Xie, Yan Wang, Arumugam Nallanathan, and Lina Wang, “An Improved K-Nearest-Neighbor Indoor Localization Method Based on Spearman Distance,” IEEE Signal Processing Letters, Vol. 23, no. 3, pp. 351-355, March 2016.

國際替代計量

無線感測網路下以差分進化演算法提升區域定位準確性之研究

未授權

主題瀏覽