透過您的圖書館登入
IP:18.222.184.122
  • 期刊

應用機率假定密度濾波器與卡門濾波器進行目標追蹤之效能比較研究

A Comparison between the Target Tracking Performance of a PHD Filter and a Kalman Filter

摘要


本研究論文,係利用機率假定密度(probability hypothesis density, PHD)濾波器,卡門濾波器(Kalman filter)於無線感測網路(wireless sensor networks, WSNs)佈置之環境中,進行追蹤變動物件之效能的比較研究。雖然在本文中,真正的WSNs僅以模擬方式進行佈署,然而,就WSNs實務之建構環境仍然可行。之後,並以電腦模擬進行其兩演算追蹤目標物,所產生之均方根誤差(root mean square error, RMSE)的比較探討。最後,由模擬所得之數據分析,可以發現,雖然PHD之演算法可以產出較小誤差的追蹤效能,然而卡門濾波器之結果在硬體的實現上可以得到較優的益處。本研究結果,可以延伸作為WSNs環境中,追蹤變動性物件之設計時的參考。

並列摘要


This study adopts the probability hypothesis density (PHD) filter and the Kalman filter as the two algorithms for tracking the maneuvering objects deployed in wireless sensor network (WSN) environments. Their tracking performances with root mean square errors (RMSEs) are compared and simulated using computer programs. Superior performance can be obtained using the PHD filter algorithm; however, a single implementation of the Kalman filter outperforms the PHD filter. To improve the tracking of maneuvering objects, the results from this study can be used as a design reference for the deployment of mobile sensors within WSN environments.

延伸閱讀