最近幾年，壓縮式取樣技術(compressive sampling theory)正被廣泛地使用在無線感測網路的資料收集應用上。藉由結合壓縮式取樣技術和路由路徑(routing path)，有些研究提供了集中式的演算法來最小化整體網路的資料傳輸。然而這些演算法通常需要完整的網路拓樸資訊和複雜的運算來取得最佳解。因此，當網路的拓樸變動時，這些集中式演算法往往需要耗費許多的傳輸來重建整個路由路徑。在這篇論文，我們提出了第一個分散式演算法來解決這個問題。我們首先介紹兩個分散式路由路徑建構演算法來算建立路由路徑。接著，我們發表一個最小化區域資料傳輸演算法來減少整體網路的資料傳輸量。模擬結果顯示出我們的演算法所耗費的建置成本遠低於集中式演算法。As compressive sampling theory has been extensively used for data aggregation in wireless sensor network, some researches provide a centralize protocol that can minimize the data traffic in the network through the combination of routing and compressive sampling. However, these protocols require the entire network topology information to compute the optimal solution. As a result, when the network environment is not stable, these protocols incur too much overhead. In this thesis, we investigate the decentralized scheme that can efficiently construct the routing path for compressive data aggregation. We first propose two distributed algorithms, namely MRT and MAT, to construct the routing path for compressive data aggregation. After works, an adjustment algorithm is proposed to locally redirect the data flow and further minimize the data traffic. The simulation results indicated that the construction overhead of our approaches is much lower than the centralize protocol.
