在無線隨意網路中以分區為基礎之連線穩定模型與適應性繞徑策略

Abstract

在無線行動隨意網路中，由於網路連線與拓樸狀態具有隨時都在改變的特性，而這樣的情況將使得網路效能明顯地下降。在這篇論文裡，我們建立了連線生命週期的模型並藉由分區的觀念區分為穩定區域與不穩定區域的網路連線生命週期，同時我們提供一個理論分析的數學型態表示式去分析連線的生命週期。而我們也藉由NS2的模網路模擬程式去驗證我們的連線生命週期模型與模擬的結果相當一致。 目前已存在的行動無線隨意網路的繞徑演算法，只能在連線中斷以後才能夠察覺網路中的某一段連線已經中斷並重新初始另一個繞徑要求的訊息。然而一旦發生這樣的狀況，傳輸的封包必須馬上重新建立並往另一條繞送路徑傳送，而這樣因連線中斷而導致須重新繞送，將會導致一個潛在的且嚴重的封包重新傳輸成本與網路負擔。基於這樣的現象，我們善用以分區為基礎的連線穩定模型的公式與理論分析，發展出一套可以預先察覺網路連線即將中斷的狀態的適應性繞徑策略 ，並於網路連線真正中斷前，提出一個繞徑更新的訊息，進而通知來源端將封包的繞送路徑更新，藉此避免重新繞徑尋找與維護的成本，同時亦能提供一個更穩定的傳輸連線。

Since the characteristic of link dynamics and network topology changing frequently in wireless Mobile Ad Hoc Networks (MANETs) and such a situation will cause the network performance decrease significantly. In this thesis, we establish an analytical link lifetime model considering link state in different zones (strong zone and weak zone) and derive formal expression for an expected link lifetime in strong and weak zones, respectively. In the meantime, we simulate this formulation model by using ns2 network simulator [19] to verify that the simulation results which are highly agreement with our model. The existing routing algorithms in ad hoc networks initiate a route discovery only after one or more links of the routing path breaks. In case of this happens, data packets must be re-routed quickly, which potentially involves a serious overhead for all ad hoc networks and lead a high retransmission cost for all data packets. In view of this phenomenon, we make use of the formulation of the zone-based link stability model to develop an adaptive routing strategy, which could initiate a routing update message in advance and before link breaks, to warn the source to find another stable path which can avoid the overhead on both routing discovery and routing maintenance while providing more reliable transmission connections.