適用於農業之無線行動感測網路技術之研究---子計畫一：無線感測網路之佈建及行動感測車派遣問題

Abstract

本子計畫主要的目的是在室內農場以及戶外田野佈建感測器網路，使得感測器能 有效涵蓋整個農場/田野，以便協助其他子計畫完整收集環境資訊，以便將來可以協助農 業的生產。為了配合總計畫「適用於農業之無線行動感測網路技術之研究」以及協調其他 的子計畫，子計畫一制定了兩大工作重點：(1)無線感測網路的佈建，以及(2)行動感測 車的派遣，在這兩大主軸下，我們制定了三大研究方向。 第一個研究方向是針對固定式感測器的佈建研究，目標是要能達到所佈建的 網路可以涵蓋整個農場/田野，此外這個網路是連通的；針對這個目標，我們將考慮兩種 策略，第一種是採用密集但任意的灑置方式佈建網路，然後再提出一套有效率的演算法 可以決定只由部份感測器醒來執行感測與通訊的工作，如此一來，網路的生命週期將可 被延長；另外一種策略則是提出一套決定擺放感測器位置的方法，使得所使用的感測器 數量能夠最少，這種策略不但可以降低網路佈建成本，由於知道感測器的位置，將來在 替換電池上也較為方便。 第二個研究方向則是開發移動式的感測器，至於如何付予感測器移動的能力是一 大挑戰，除了重新設計感測器的硬體外，另一種可行且經濟的方案就是選用具有移動能 力的載具，將固定式的感測器放置在上頭，並且透過小型的嵌入式電腦來控制整個系統 的運作；我們預計選用Lego Mindstorm 作為感測器的移動載具，並且利用Crossbow 與 Intel 所開發的Stargate 作為系統的運算處理中心；藉由此移動載具的觀念，我們不但能 免去重新設計硬體的問題，同時也能減少硬體開發的成本。 至於第三個研究方向則是針對移動式感測器的派遣問題，為了有效管理電源以延 長系統的使用時間，因此我們不能任意地派遣感測器去指定地點，相反地，我們應該根 據指定位置(或事發地點)以及感測器的位置來判斷由誰前去執行任務，因此這就衍生出 感測器的派遣問題；根據感測器的能力與性質，我們會將這個問題再分成同質性感測器 的派遣問題以及異質性感測器的派遣問題，我們將會針對這兩個問題分別提出有效率的 電量管理之派遣演算法。

The objective of this sub-project is to deploy sensors in a greenhouse or farm, so that the greenhouse or farm can be fully covered by the sensor network. This can help other sub-projects effectively gather the environmental information, which can be fed back to the farmers so that they can increase their production in the future. In order to follow the ultimate goal of the main project, 「Wireless Mobile Sensor Network Technologies for Agricultural Application」, and cooperate with other sub-projects, we classify the related issues as two classes, (1) deployment of a wireless sensor network and (2) dispatch of mobile sensor car. For these two issues, we derive three research topics. The first topic is to deploy static sensors in a greenhouse or farm. The goal is to fully cover the greenhouse or farm and the network is guaranteed to be connected. We consider two possible strategies. The first strategy is to deploy sensors densely and randomly. Then we will propose an efficient algorithm to make only a subset of sensors to wake up for detection and communication jobs, so that the overall network』s lifetime can be prolonged. Another strategy is to design an efficient sensor placement scheme to decide the locations to place sensors on. This scheme will use fewer sensors to deploy the network so that the cost can be reduced. Besides, since we can obtain the locations of sensors, we can easily replace their battery when they run out of energy. The second topic is to develop the mobile sensor cars. How to design a mobile sensor car is a big challenge. One possible solution is to redesign the hardware, but the cost is too high. Another possible and cheap solution is to integrate a mobilizer, a static sensor, and an embedded computing device into a mobile sensor car. We will select the Lego Mindstorm as the mobilizer and Crossbow/Intel Stargate as the embedded computing device. By adopt this solution, we can reduce the cost to develop a mobile sensor car. As for the last topic, we will focus on the sensor dispatch problem. In order to efficiently manage the energy, we cannot arbitrarily dispatch mobile sensor cars to the designate locations. In the contrary, we have to dispatch mobile sensor cars according to their present locations and the designate locations. And this will derive the sensor dispatch problem. According to the sensors』 capabilities and properties, we can classify the problem into homogenous sensor dispatch and multi-functional sensor dispatch problems. We will propose two efficient dispatch algorithms to solve these two problems.