以適應性水閘資料匯集系統用於長鏈狀無線感測網路之研究

Abstract

因為物理環境的限制，近幾年來無線感測網路中，長鏈狀網路晉升為許多實際部屬的拓樸結構。一般來說，長鏈狀拓樸是由多條長鏈狀分之所組成，每條分之由多個感測點相連而成，而每個感測點都只有一個父節點。在長鏈狀拓樸下，即使資料匯集可以降低無線感測網路中多餘的封包競爭，資料匯集還是會受到最大封包資料量的限制。這篇論文提出長鏈分之中應該放置都個資料匯集點，稱作水閘，負責匯集上游一般感測點的資料。這篇論文針對長鏈狀無線感測網路提出一個適應性水閘指定方法，該方法在無線感測網路回報效率以及資料收集壅塞程度中取得平衡，並且減輕無線感測網路中的漏斗效應。該方法更動態的調整水閘的位置及數量以適應網路中隨時間變化的流量。實作結果顯示出適應性水閘指定方法確實能改善長鏈狀網路的效能。

Constrained by the physical environments, the long-thin (LT) topology has recently been promoted for many practical deployments of wireless sensor networks (WSNs). In general, an LT topology is composed of long branches of sensor nodes, where each sensor node has only one potential parent node toward the sink. Although data aggregation may alleviate excessive packet contention, constraints imposed by the maximum payload size of a packet severely limit the amount of sensor readings that may be aggregated along a long branch of sensor nodes. This paper argues that multiple aggregation nodes, termed lock gates, need to be designated along a branch to aggregate sensor readings sent from their respective upstream sensor nodes (up to the other upstream lock gate(s) and/or the end of the branch). The paper further describes an adaptive lock gate designation scheme for LT WSNs, which balances the responsiveness and the congestion of data collection, and mitigates the funneling effect. The scheme also dynamically adapts the designation of lock gates to accommodate the time-varying sensor reading generation rates of different sensor nodes. Experimental results demonstrate the effectiveness of the scheme.