以分散式幾何演算法發展無線感測網路之室內定位技術

Abstract

無線通訊技術的發展讓環境監測與健康照護系統有更多的可能性，但是如果不知道感測器的位置，便降低了收集到的環境資訊與生理資料的參考價值。因此發展了許多室內定位方法以傳遞時間延遲或接受訊號強度指標進行定位。本研究在ZigBee無線感測網路中設計並實現以接受訊號強度指標為主的分散式定位系統，使用的幾何定位演算法稱為bounding boxes法，並以mass spring optimization改良。由於接受訊號強度指標容易受到干擾以及晶片本身的差異性，因此本文提出的定位系統採用卡爾曼濾波器和最大似然估計器改善訊號強度指標的穩定性並加入接收訊號強度指標校準機制。 本研究所提出的定位演算法及改良方法以透過模擬和實測驗證並評估效果。bounding boxes與加入mass spring optimization後在邊長為五公尺的正方型模擬空間中的平均誤差分別為1.1517公尺和0.67公尺。以線性迴歸法分析實際收集的訊號建立接收訊號強度指標與距離之間的數學模型。bounding boxes加入mass spring optimization並配合接收訊號強度指標校準和濾波處理後，在邊長為7公尺與5公尺的長方形測試空間中的是實測誤差為1.035公尺。 分散式幾何定位演算法實現於感測器上並提供了合理的準確性，降低了系統的封包傳輸量，增加感測器的使用時間。本文針對接受訊號強度指標的處理提升了定位的準確性，此外較少的運算量及封包傳輸量增加了與環境監測或健康照護系統的整合可能性。

The development of wireless communication technology brings more possibility in environment monitoring and health-care system. But the absence of sensor location information reduces the reliability of sensed environmental and biomedical data. Therefore, many localization algorithms have been proposed based on Time of Arrival (TOA) or Received Signal Strength Indicator (RSSI) in the indoor environment. This study designed and implemented a RSSI-based distributed localization algorithm in a ZigBee-based Wireless Sensor Network (WSN). A geometrical localization, Bounding boxes and Mass spring optimization for refinement is presented in this thesis. Because of the disturbance and chip-to-chip variation in RSSI measurement, Kalman filter and Maximum likelihood estimator and RSSI calibration have been applied in the system. The proposed localization algorithms and improving methods were verified and evaluated through simulations and experiments. The average error of bounding boxes and that with mass spring optimization in simulation are 1.1517m and 0.67m respectively in virtual square space of 5m edge. The collected RSSI data establish the model of RSSI to distance by linear regression. The average error of bounding boxes algorithm with RSSI filtering and calibration in the experiment are 1.035m in a rectangle space of 7m and 10m edges. The distributed geometrical localization algorithm implemented on the sensor device provides reasonable estimation accuracy, reduces the packet traffic load and extends the battery life. The processes for RSSI enhance the accuracy of localization. The lower computation and packet traffic load reveal the potential to merge with environment monitoring and heal-care systems.