指向性感測網路上有效確保時間性覆蓋之感測器佈署以及其物體監控之應用

Abstract

於諸多無線感測器網路的應用中，由於應用需求或是設備限制，感測器之感測幅度為扇形範圍。此外，輔以器械的幫助，如步進馬達，這些感測器便可透過旋轉來覆蓋其周圍的目標物，這類型的感測器被稱為旋轉式指向性(R&D)感測器。在論文中，我們考慮R&D感測器的佈署問題，其決定如何放置最少數的R&D感測器，以符合覆蓋特定物體的條件，如此每個目標物皆可為δ-time covered，其中0 < δ ≦ 1。若在一段固定的周期時間T裡，目標物可被一個感測器覆蓋至少達到δT時間，則可稱此目標物為δ-time covered。R&D感測器的佈署問題是NP-hard，因此我們提出了兩個有效的探索法，第一個方法為最大覆蓋佈署(MCD)演算法，其核心在於佈署感測器以覆蓋最多的目標物；第二個方法為圓盤重疊覆蓋佈署(DOD)演算法，利用處理感測器覆蓋範圍間的重疊問題，以降低感測器的數目。模擬實驗中，針對不同目標物分布情形可顯示出各方法的有效性。此外，為了實證本篇論文所提出的時間性覆蓋模型之特性，我們利用R&D感測器開發一個事件導向視覺監控系統，於此系統中，R&D感測器利用紅外線偵測與攝影機來定期地監測目標物，當目標物失竊時，偵測到的感測器會回報含有快照細節的警告訊息給使用者。

In many wireless sensor applications, sensors may possess sector-like sensing coverage due to application requirements or equipment constraints. In addition, with the help of machinery such as stepper motors, these sensors can rotate to cover the objects around them. This type of sensor is called a rotatable, directional (R&D) sensor. In the paper, we consider the R&D sensor deployment problem, which determines how to place the minimum number of R&D sensors to cover a given set of objects, such that each object can beδ-time covered, where 0 < δ ≦ 1. In particular, an object is said to beδ-time covered if during a fixed period T, the object can be covered by one sensor for at leastδT time. The R&D sensor deployment problem is NP-hard and we thus propose two efficient heuristics. The first heuristic, called the maximum covering deployment (MCD) algorithm, always places sensors such that the maximum number of objects can be covered. On the other hand, the second heuristic, called the disk-overlapping deployment (DOD) algorithm, exploits the overlap between sensors’ coverage to save sensors. Simulation results show the effectiveness of the proposed heuristics under different distributions of objects. Moreover, to demonstrate the feasibility of our temporal coverage model, we develop an event-based visual surveillance system by R&D sensors. In this system, objects are periodically monitored by R&D sensors equipped with infrared detectors and cameras. When one object is taken away, the monitoring sensor will report a warning message along with the detailed snapshots to the user.