發展無線感測網路子集廣播式之時間同步機制應用於聲源定位系統

Abstract

隨著無線科技的發展，感測器的微小化，無線感測網路的應用範圍越來越廣，其中，可以應用在軍事、生活上的聲源定位引起研究團隊高度興趣。使用無線感測網路進行聲源定位使麥克風位置擺放的可調整性提高並可擴展其應用範圍。本研究使用無線感測網路結合麥克風感測器，基於每個麥克風之聲音到達時間差資訊來實現聲源定位。為了計算到達時間差，感測節點之間必須完成時間同步，因此，本研究提出網路子集廣播式之時間同步機制應用聲源定位系統。本研究使用牛頓法來計算雙曲線交點，即為聲源所在位置，並計算雙曲線之漸近線交點取得最佳的初始位置，以計算出正確的聲源位置。本研究並加入二維卡爾曼濾波器用於即時聲源位置追蹤。 本研究所提出的網路子集廣播式之時間同步機制已透過模擬和實測驗證並評估效果。在重新同步週期維持在二十分鐘以內時，平均誤差維持在1.3毫秒以內。結合牛頓法之定位演算法經實測驗證，在25平方公尺的空間中，平均誤差約為1.2公尺。本研究分析不同參數對無線網路時間同步的影響，依不同的誤差容忍度，由實驗結果提供使用者最適合的參數設定，並由實驗結果得知，對網路子集廣播式之時間同步機制影響最大的為重新時間同步的週期。 網路子集廣播式之時間同步機制將所有網路節點同步至網路協調者並提供合理的時間估測準確性，以供聲源定位演算法使用。未來除了聲源定位外，也可以將此同步機制應用在其它需要精確時間同步的無線網路系統。

Recently, miniature low-power devices which are capable of micro-sensing, actuation with onboard processing, and wireless communication have attracted considerable attentions in developing the wireless sensor network (WSN) and found numerous applications. Among of the applications with WSN, acoustic source localization is the popular research. The microphone sensors do not form a structured formation but often are deployed randomly when locating the acoustic source in WSN. This offers more flexibility in setup and expands the application range. This study performs acoustic source localization based on time difference of arrival (TDOA) in WSN. To utilize the localization algorithm based on TDOA, all the nodes in the network should be synchronized. Hence, this study proposed subnetwork broadcast time synchronization (SBTS) for acoustic source localization system. The acoustic source location is the intersection of hyperbolas calculated by Newton’s method, and the intersection of hyperbola asymptotes is the proper initial position for Newton’s method. This study adopts the two dimensional Kalman filter for location tracking of acoustic source. The proposed time synchronization protocol was verified and evaluated through simulations and experiments. The average of estimated time offset errors is less than 1.3 milliseconds when the re-synchronization period is less than twenty minutes. In localization experiments, the average estimated location errors of acoustic source calculate by Newton’s method is about 1.2 meters in a 25-square-meter virtual space. The effect of different indices on SBTS is analyzed in this study, and the adequate parameters are recommended form the experimental results. Moreover, the re-synchronization period affects SBTS significantly from the experimental results. SBTS synchronizes all the nodes in the network to PAN Coordinator automatically. Moreover, SBTS provides reasonable estimation accuracy for acoustic source localization system. Besides the acoustic source localization system, SBTS could be applied in other systems which need accurate time synchronization in the wireless network.