在電池式無線網狀網路上做最大化總收益並以基因演化為基礎之群播繞徑研究

隨著近年來無線網狀網路（WMNs，Wireless Mesh Network）的蓬勃發展，相關的研究及應用也逐年的在增加，由於群播（Multicast）傳輸的方式在WMNs架構中可以同時提供影音串流、即時多媒體等服務給大量特定的終端使用者，因此在WMNs中無干擾的群播路由應用儼然成為電池式多頻道多網卡WMNs最重要的問題之一，由於節點的電量會隨時間消耗而中斷傳輸服務，傳統的研究在探討電池式無線網路時主要著墨在如何拉長整體的網路維持時間，本文探討如何將服務用戶數N及整體維持時間T做最大化，假設在商業行為中每秒服務一位客戶可收取一單位利潤，可知若T*N越大，則所獲得的收益就會越多。因此本文提出以基因演化為基礎之演算法，並自定義函數Total Profit用來檢驗演算法在建立群播樹時其總收益的高低，在各節點的電量會隨時間消耗及無線網路頻道數量有限的情況下，本文所提之演算法可建立無干擾的群播路由，同時在維持整體網路時間及最大化服務用戶數兩方之間取得平衡，進而得到最大化總收益之效，最後並藉由實驗結果證實以本文所提出之演算法建立群播路由其效能優於其他現有的方法。

In this paper, we consider the following scenario: an interference-free multicast tree is required to be deployed on a battery-powered multi-channel multi-radio wireless mesh networks, so that a video stream can be delivered to several destinations where a number of clients may subscribe the stream simultaneously. For various multicast routing problems studied on the battery-based ad hoc wireless networks, the traditional design aim is to maximize the life-time of the network. In our problem, however, we are required to maximize both the life-time T and the number serviced clients N at the same time. Given one unit of profit per second per client, it would be more reasonable to maximize T∗N instead of T for the owners of networks to have maximum profit. Hence, to support this business model, a term called profit which is defined to be T∗N is used as the optimization goal for constructing the multicast tree. A genetic-based algorithm is devised in this study to solve this problem. According to our experimental results, the GA-based approach can significantly outperform the other previously proposed methods.