無線感測網路中利用空間關聯性之節能資料收集方法

Abstract

隨著科技的發展與網路的普及，無線感測網路日益展露其重要性與實用性。因此，本論文在無線感測網路中，提出一利用空間關聯性之節能資料收集方法。第一，我們利用空間關聯性將一大群感測節點分類為一個個叢集，每個叢集中的節點感測相似的環境讀數並僅挑選出一個代表點回報該叢集的讀數。第二，為了進一步節省能源的消耗，各叢集回報少量的資訊給伺服器，由伺服器重新選擇可合併的叢集，並把選擇組合叢集的方式模組化為infer-graph set問題，提出一貪婪式演算法去解決。第三，我們提供每個感測節點一個safe region，避免經常性的溫度改變導致叢集的重建以增加叢集存在的持久性。實驗顯示本論文提出之方法明顯降低了資料傳輸量以及叢集的數目，因此增加了無線感測節點的生命週期，實驗結果並証明了提出之方法較適用於動態的感測環境。

In this thesis, we propose an energy-efficient data gathering scheme over a highly spatial-correlated region in wireless sensor networks. First, we present a mechanism to group sensor nodes into a set of clusters, which have monitored similar phenomena by exploiting spatial correlation. The member readings in a cluster are bounded within an user-tolerable threshold and each cluster is represented by one clusterhead to answer queries. Second, in order to further reduce energy consumption, we merge the clusters which got resembled readings. The problem to merge clusters is modeled as an infer-graph set problem. We devise a greedy-based heuristic algorithm to acquire the near-optimal solution of choosing the new representative clusters. Third, we design a safe region for each sensor (even clusterhead or member sensor) not to casually leave the cluster and strengthen the persistence of clusters. It avoids the overheads of frequently rebuilding clusters. Experimental results show that our work is superior to previous techniques by saving 11 percent of bytes count and decreasing 15 percent of clusterhead count. Furthermore, our scheme is more suitable for dynamic sensing environment than competing techniques.