本論文探討大量的相關研究，包含耐延遲網路路由與社群路由方法，回顧過去的文獻發現幾項人類的移動性質及社群互相接觸的特性。進而本論文以這些特性設計Leverage機制，在這個機制之中藉由節點與節點過往的接觸時間來判斷這個訊息是否要轉送。並以實驗佐證演算法的效能，使用訊息抵達率、訊息傳遞費用與成功傳遞比例進行量測。實驗結果顯示本論文提出的演算法不僅有較佳的訊息抵達率而且傳遞訊息的次數較少，能有效的降低傳輸費用。 In delay-tolerant networks, it is hardly possible to sustain any end-to-end data delivery paths between any two nodes because the networks suffer from various restrictions by non-uniform node distribution, high node mobility as well as limited transmission ranges. Nodes thus take a store-carry-and-forwarding method to send messages to destinations when they have any opportunistic contacts with other nodes in a network. Considering the recent advance of wireless and mobile networking systems, human beings possessing mobile devices are able to store data in such devices, carry the data along with them, and forward the data to encountered devices as encountering people during movement. As human beings appear like mobile nodes in a network context, the scenarios of human movement and contact may fall into the application domain of social-based delay-tolerant networks. Therefore, the research study of human mobility characteristics will contribute to the design of new social-based routing schemes in delay-tolerant network environments.
The study in this thesis investigates lots of related works, including not only delay-tolerant routing but also social-based routing methods for delay-tolerant networks. This literature review finds out several behavior characteristics about human mobility patterns and contacts by social communities. Accordingly, the study exploits these characteristics to design a leverage routing scheme. In this scheme, message forwarding decision is made by referring to the information of contacts between two nodes in the past. To examine the proposed scheme, simulations are conducted to the performance in terms of message delivery probability, overhead ratio, and successful relay ratio. Performance results indicate that the leverage routing scheme not only has better delivery probability but also results in lower amounts of message transmissions in the network.
