利用訊號強度為基礎之位置追蹤及感測器佈署問題

Abstract

中文摘要 在無線隨意感測器網路，利用各種測量距離的技術，例如，TDOA (Time Difference of Arrival)和RSSI (Received Signal Strength Indicator)，一群感測器可以透過合作的機制，來估算物體的所在位置。在此論文中，我們將利用RSSI (Received Signal Strength Indicator)的path loss model來追蹤環境中物體的所在位置。固定一對傳送端和接收端的距離，訊號的強度會隨著距離衰減，而且易受環境因素影響，例如環境中的障礙物，或是訊號傳播固有的multipath和fading的特性。此外，在訊號空間的觀點來看，固定一對傳送端和接收端的距離，各種訊號強度出現的機率的是呈現Gaussian Normal Distribution，因此利用RSSI (Received Signal Strength Indicator)的測量距離的技術是不準確的。為了解決利用RSSI估算距離時受環境因素影響造成的不準確，在此我們建立一個機率的位置估算模型來找出環境中的物體位置。更進一步的，利用這機率的位置估算模型來建立Error model，評估各種感應器佈署方式，在對環境中物體作定位時的誤差程度，並評估定位系統的效能。另外，應用我們的模型來改善定位系統的效能，在某些特定的位置增加新的感測器來幫助追蹤與對環境中的物體作定位，再者，結合此model和awake protocol和sleep protocol可以喚起某些特定位置的感測器改善效能，和關掉環境中多餘的感測器，以節省cost。

關鍵字: 隨意網路，環境監督，網路部署，感測器網路

ABSTRACT In wireless ad hoc sensor networks, according to distance measure technologies such as TDOA (Time Difference of Arrival) and RSSI (Received Signal Strength Indicator), a number of sensor nodes could cooperate to estimate the location of the object. In this paper, we approach location tracking based on the RSSI (Received Signal Strength Indicator) path loss model. Although the signal strength is decreasing with the distance between transmitter and receiver, the signal propagation is various depended on the environment condition such as obstacles, multipath fading. In addition to the signal strength is probabilistic vibration, and it follows Gaussian normal distribution in signal space. Therefore using RSSI distance estimate technology is inaccurate. According to the view, we derive a log-normal probability function of signal strength, and take it to translate the probability of signal strength from signal space into distance space for location tracking problem. Furthermore we apply our location estimation process to sensor deployment, taking our algorithm to evaluate the error of whole environment, and add some sensor nodes on some locations to reduce the error degree of the location estimation system. Finally, we also demonstrate the experiment and application of our error estimate process for sensor deployment.

Keywords: Ad hoc network, environment monitoring, pervasive computing, network deployment, sensor network.