現今發展健全的網際網路架構，以及成熟的通訊協定，在某些特殊的環境下，也許並不能有好的發揮，延遲容忍網路則是一個例子。延遲容忍網路是一種網路的概念型態，其擁有高延遲以及高度不可預期性。在延遲容忍網路中，已優先假設可能會發生較長的等待、傳輸不穩定、不會有發送端至目的端完整的網路傳輸路徑。 許多偏遠的村落並沒有網際網路，然而他們可能有學生需要吸收新知，也可能需要收發電子郵件。建立以公車為基底之延遲容忍網路，是一個可行的辦法。本論文的目標為增進公車為基底之延遲容忍網路的效能，幫助偏遠村落或者是開發中地區需要幫助者，使之也能生活在資訊社會。 本論文先介紹讀者關於延遲容忍網路，以及其延遲容忍網路路由演算法。同時，也介紹其他類似的網路，以期能夠讓讀者較容易瞭解延遲容忍網路之功能。本論文提出以公車為基底之延遲容忍網路調適性交通路由演算法，來預測哪些時間可能會是公車之間能夠有傳輸機會。有了預測資訊，整體網路效能將會提升，且訊息傳送的成功率會增加。本論文提出之路由演算法，不但有較高的傳輸送達機率，同時也能夠適用於其他以公車為基底之延遲容忍網路。在本論文後段，有一系列的實驗，證明本論文提出之演算法，能夠改進以公車為基底之延遲容忍網路傳輸效率。與其他參考文獻比較，最少可提升 50% 最多可提升 326%。 Under some special circumstances, today’s fully developed Internet structure may have less opportunity to operatic as it normally do. Taking Delay Tolerant Networks (DTN) for example, it is a concept of networks, recognized for its unpredictability and high delay quality. It is assumed that DTN must experience frequent, long-duration partitioning and an end-to-end contemporaneous path may never occur. Many remote villages could not access to Internet. However, there may be students who need to read news or send e-mails. One of the solutions is to establish a Bus-Based DTN. The goal of this paper is to enhance the performance of Bus-Based DTN to help remote villages or developing areas of the world to become a part of the information society. This paper gives a survey of DTN first. We not only introduce DTN but also the DTN routing algorithm. We also present other networks that are familiar to DTN in order to make readers understand DTN more easily. This paper proposes “Adaptive Traffic Routing Algorithm” to predict possible timing for buses to communicate with each other. If chances are provided for message to transfer among the buses, its function as DTN may have a higher efficacy. Not only does our algorithm have a high deliver rate but it is also suitable for other bus-based DTN. At the end of this paper is a series of simulations to prove that this paper’s algorithm can better the performance of Bus-Based DTN. Compare to other DTN algorithm, this paper’s algorithm can raise the efficiency from at least 50 percent up to 324 percent.
