無線感測網路之事件周圍查詢處理機制

Abstract

在無線感測網路領域，事件的發生與偵測一直是很重要的ㄧ環。然而，事件有可能在任何時間點演化，並且往任意方向蔓延，及早偵測並且提供即時的資訊給使用者是非常重要的，使用者透過資訊的回報可以採取適當的動作，對事件發展的情況取得控制。因此，在本篇論文我們提出ㄧ種新型態的空間查詢，簡稱為『事件周圍查詢』，透過此查詢可以選出最靠近事件邊界的感測器節點來監控事件，而被選出來的節點就稱之周圍節點。另外，我們為此提出兩個演算法，一個是以RNN為基礎的演算法以及另一個Greedy演算法來處理事件周圍查詢。以RNN為基礎的演算法延伸現有的RNN查詢來選取周圍節點，然後找出它們之間不能被感測到的漏洞，再選取適當的感測器節點去彌補此漏洞。Greedy演算法每次都去選取最靠近事件的感測器節點作為周圍節點，然後藉由鄰居關係的檢驗避免選取多餘的感測器節點作為周圍節點。實驗結果顯示，Greedy演算法在選取周圍節點的數量跟所花的成本優於以RNN為基礎的演算法。

In many applications of wireless sensor networks, events evolve at anytime and in any directions. Early detection and providing timely information are especially important for users to take events under control. In this paper, we propose a novel type of spatial query, Event Surrounding query (abbreviated as ES query), in which sensor nodes that are near to the boundary of events are selected for monitoring. Those nodes selected by ES query are referred to as surrounding nodes. We propose two algorithms, RNN-based algorithm and Greedy algorithm for event surrounding queries. RNN-based algorithm extends RNN query to select surrounding nodes and then to identify the gaps. RNN-based algorithm will select proper sensor nodes to enclose the event. Greedy algorithm selects those are nearest neighbors to the event as surrounding nodes every time. Then it avoids redundant sensor nodes to join surrounding nodes via neighborhood relationship recognition. Experimental results show that Greedy algorithm outperforms RNN-based algorithm in CPU time and average number of surrounding nodes selected.