臺灣大學: 電機工程學研究所陳光禎曾繁閔Tseng, Fan-MinFan-MinTseng2013-03-272018-07-062013-03-272018-07-062011http://ntur.lib.ntu.edu.tw//handle/246246/253945在無線網路，尤其是感知無線電網路(cognitive radio networks)及無線網狀網(wireless mesh networks)中因為傳輸比較不可靠使得網路整體效能降低。臨機路由(opportunistic routing)藉由動態調整傳輸路徑來解決因為固定傳輸路徑所造成的效能低落問題，在這樣的傳輸機制下，我們歸納其網路傳輸速率受限於傳給最後目標的封包是否對最後目標來說是新的。我們使用網絡編碼(network coding)的概念找出網路中的最小截流(min-cut)，以此來算網路的理論傳輸速率的界限。我們的分析結果顯示傳輸速率界線與網路中最多能夠同時傳輸的數量有關，並且傳輸的路徑數也會影響傳輸速率。 接著我們探討在多個具相關性的信源(correlated sources)時，路由機制該如何設計使得網路整體的傳輸量能夠最有效率。我們假設在無線網路中，信源能夠透過偷聽(overhearing)的方式做更有效率的傳輸，也就是說，信源能夠由偷聽到其他信源所傳送的訊息，判斷自身如何最有效率的傳輸。根據這個假設，我們最後想要透過路由的機制使得整體網路的傳輸量能最有效率。為了分布式的方式達到整體的最佳化，我們用吉布斯採樣提出一套路由機制。我們將我們所提出的演算法與貪心演算法做比較，模擬結果顯示我們的演算法比貪心演算法有著顯著的改善，改善的程度也與信源的相關程度有關。In wireless network, the broadcast feature makes the transmission link unreliable in certain network, like cognitive radio network or wireless mesh network. The opportunistic routing is therefore been proposed to solve the problem. In opportunistic routing, the end-to-end transmission path is not pre-determined. Every node would be potential forwarder. In this routing mechanism, we conclude that the capacity is bound in the situation that whether the forwarders to destination have received an un-transmitted packet so that it can transmit to destination. To solve this problem we use the concept of network coding that we find the min-cut of network from source to those forwarders. Our analytical result shows that the capacity is related to the total transmission links within T time slots, and the capacity will also increase with the number of transmission paths. The second part of our work does the multi-sources, single destination network while the sources are correlated, which is most likely the wireless sensor network. We use the overhearing information to improve traffic efficiency, i.e., every node can do data compression by the transmitting information message that it overhears from its neighboring nodes. We let each sensor choose its forwarder, and also do efficient transmission by the overhearing message of its possible forwarders. Our goal is to minimize the total transmitting information message from every source nodes and relay nodes. To obtain a distributed routing algorithm, we use the Gibbs sampler which theoretically can approach to global optimal by local information. We compare our algorithm with greedy approach, and the simulation result shows that our algorithm improves the total transmitting information message significantly. We also find that the improvement is related to how correlated of sources.1984145 bytesapplication/pdfen-US臨機路由網絡編碼相關信源吉布斯採樣opportunistic routingnetwork codingcorrelated sourceGibbs samplerthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/253945/1/ntu-100-R98942042-1.pdf