臺灣大學: 電信工程學研究所陳光禎李厚勳Lee, Hou-HsunHou-HsunLee2013-03-272018-07-052013-03-272018-07-052011http://ntur.lib.ntu.edu.tw//handle/246246/252773除了統計物理中的理論研究外，由於缺乏共同的設施及介面，大型異質性網路中的同步問題一直是一個有待克服的挑戰。然而，時間的誤差卻會顯著的降低系統效能。有鑒於現存的演算法無法有效的處理具有錯誤時間的結點，於是我們提出一個改進的時間同步演算法，適用於異質性無線網路，並達到可信賴的同步效果。我們討論了這個演算法在不同網路拓樸以及網路大小條件下的效能，並進一步提出了一個可用於特定網路下的加強型演算法。我們也把我們的演算法和另一種二元狀態共識演算法做比較，結果證明我們的演算法在許多方面有其優點。最後，我們提出了兩個可用此演算法解決的實際網路問題。我們將此演算法應用於進階長程演進系統(LTE-A)中微型基地台(femtocell)的時間同步，證明可有效限制時間誤差影響整個網路。另外在社群網路中，此演算法也提供了一種方法以謠言散播者對群眾的影響。In addition to theoretical study in statistical physics, synchronization in large heterogeneous networks has always been a technology challenge, due to lacking of common infrastructure and interface. However, mis-alignment of timing can significantly degrade the performance. In light of existing algorithms failing to effectively respond to node failure in timing, we develop an improved time synchronization algorithm enabling reliable synchronization in heterogeneous wireless networks. We discuss the performance of our algorithm on different network topologies and network size in order to understand the behavior of our algorithm, and further propose an enhanced algorithm that can be used on specific network topology. We also compare our algorithm with another binary-state consensus algorithm, the results turn out that our algorithm has many advantages. At last, we give two realistic network applications that can be well handled by our algorithm. We apply this improved algorithm to femtocells in LTE-A systems to successfully confine the timing errors in a more effective way, and this algorithm also decreases the influence of rumor spreader in social networks.7954438 bytesapplication/pdfen-US時間同步投票者模型合併型隨機漫步複雜網路進階長程演進系統共識問題time synchronizationvoter modelcoalescing random walkscomplex networksLTE-Advanced systemconsensus problemthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/252773/1/ntu-100-R98942034-1.pdf