無線感測網路於各種領域的應用主題已廣泛地被研究與討論，近年隨著行動感應器的應用概念與技術加入，行動感測網路的相關議題也陸續被提出。其中行動感應器可移動的特性，使得無線感測網路的偵測範圍可以動態調整，讓應用上更具彈性，並可進一步使用於各種異常環境區域的偵測，如火場、海洋油污區域偵測等。傳統無線感測網路是以大量隨機散佈的設置方式進行環境偵測，然而由於行動感應器成本較高，行動感測網路並不適合採用這種應用模式。 本研究的目的，在於嘗試設計一套能以少數的行動感應器，進行環境中未知形狀與位置之目標區域偵測的應用模式。為了能有效地進行偵測，各個行動感應器首先應組織成網路，以集團的方式進行活動，隨後再以此感應器網路為基礎，搭配有系統的偵測移動機制進行環境的偵測。在此研究目的與操作概念下，本研究提出小型行動感測網路架構，以及以此架構為基礎所設計的偵測移動機制。在行動感測網路架構主題方面，本研究首先提出基礎之架構 － SensorGroup，隨後再透過融合無線定位機制，提出進階的Centralized SensorGroup架構，以降低建構SensorGroup時對行動感應器定位能力需求之門檻，並進而降低網路之建構成本。在偵測移動機制主題方面，本研究首先提出基礎的S形偵測移動機制，透過讓SensorGroup在偵測環境中進行規律的S形移動，以達成偵測目標區域之功能。為進一步提升偵測之效率，本研究另提出一套利用漸進式分群演算法之偵測移動機制，藉由偵測紀錄的群集分析，讓SensorGroup可以依據目標區域分佈狀態不同，動態改變其偵測移動方式，以期能以較高之效率收集目標區域邊緣之資訊。 The application issue of wireless sensor network in various areas had been widely discussed. With the development of concepts and technologies of mobile sensor, the issues of mobile sensor network are also arising. The mobility feature of mobile sensors makes the sensing area of wireless sensing network can be adjusted dynamically. Hence, mobile sensor network can be used in detection operations of abnormal environments such as conflagration or oil spill areas. Traditional wireless sensor networks detect environment by the way of random deployment. Due to the higher cost of mobile sensors, it is inappropriate to use mobile sensors by the same way. The aim of this study is to design an application model for small mobile sensor network to discover unknown target areas in the environment. To detect the targets efficiently, the mobile sensors should be formatted to a network and operate as a group based on a systematic sensing mechanism. With this concept, this study is composed with two parts － mobile sensor network structure and sensing mechanism. In the part of mobile sensor network structure, we firstly propose a basic network model called SensorGroup. To reduce the requirement for the localization ability of mobile sensor and also the construction cost of network, an advanced structure called Centrlized SensorGroup is proposed by combining a wireless localization mechanism with SensorGroup. In the part of sensing mechanism, we firstly propose S-shaped sensing mechanism. In this basic mechanism SensorGroup senses the target area with a regular S-shaped movement. To improve the efficiency of sensing operation, we then propose an incremental clustering aided sensing mechanism. In this advanced mechanism, SensorGroup can adjust its movement dynamically based on the cluster analysis of sensing record of target area to achieve higher sensing efficiency.
