Research Project (Selected papers)
耐延遲網路環境下由於節點的高移動性及傳輸距離等限制,無法保證兩節點間持續存 在有一條固定的資料路由路徑,嚴重影響封包路由機制的運作,因此為了改善資料傳輸效能,傳統的耐延遲路由設計大多是利用增加封包複製數,或透過與相遇節點建立機會性連線時選擇有利的轉送節點等方式來增加封包抵達率或減低傳輸延遲。 然而,過度增加封包複 製數及選擇轉送節點所需額外記錄的節點資訊,將可能導致網路壅塞,並且延伸出路由協 定的低度層次性和高執行複雜度等問題。 本論文提出一套基於節點密度感知的路由機制,利用節點分佈具有疏密度的差異,將封包轉送至前往高密度或處於密度較高區域的節點,使之有更大的機會相遇更多節點,從而提高相遇目的地端的機率及縮短封包遞送延遲的時間。
Delay-tolerant networking technologies are characterized by high node mobility, uncertainty of node existence, and intermittent connectivity. Message routing in such networks thus resorts to new routing paradigms instead of conventional end-to-end routing paradigms in mobile ad hoc networks. When the use of erasure coding-based routing paradigm for delay-tolerant networks attracts lots of research interests, the study in this paper quantifies the cost measurement of erasure coding-based routing for both message unicasting and multicasting mechanisms in delay- tolerant networks. Upon delivery ratio and delay time, the examination of performance sensitivity results in many new inherent observations, properties and insights into the virtue of the erasure coding-based routing paradigm, providing significant information for the design of erasure coding-based routing mechanisms in support of multicasting in delay-tolerant network environments.
在耐延遲網路下由於節點密度稀疏及傳輸距離等限制,網路拓撲往往是破碎不連通的,大部份的時間中,節點間並不存在一條點到點的路徑,因此過去諸多使用在行動異質網路下的路由協定並不適合直接應用在耐延遲網路上。 本論文的研究首先分析耐延遲網路中資料傳遞延遲時間的變異性,提出可參考的定量分析結果,據以提出一套有效的時間戳記路由機制,此路由方法的設計藉由紀錄節點間廣播的時間戳記控制訊息,讓節點可估測出自身將訊息傳送到其它節點所需的訊息傳遞時間,並且精簡化的控制訊息表示格式能讓整體的控制訊息數量之複雜度抑制在 O(n),降低在網路資源缺乏時對訊息傳遞成功率的影響。 最後,本研究透過不同的節點移動模型來模擬驗證所提出的時間戳記路由演算機制在訊息到達成功率、訊息傳遞延遲時間與訊息複製數量等三個指標上的表現, 並與其它路由協定做比較,模擬結果顯示本論文所提出的路由方法有更佳的訊息到達率,亦能大幅降低訊息複製的數量及傳輸成本。
Message delivery in delay tolerant networks often resort to message replication to increase the successful delivery rate. Repeatedly replicating messages, however, consumes a vast amount of resources like local buffer space on nodes and limited bandwidth during inter-node communications. It is important to have efficient buffer management and scheduling policies in delay tolerant networks. Although previous studies proposed various buffer management and scheduling policies, their efforts mainly contributed to the simple scenario of message unicasting from a source to a singular destination. When message multicasting towards multiple destinations is considered, previous solutions perform inefficiently. Thus, this paper proposes a new buffer management and scheduling policy for message multicasting in delay tolerant networks. The proposed design elegantly extends an optimal knowledge-based scheduling and drop policy, and derives a new utility function to prioritize messages in buffer to maximize the successful delivery rate in a network. Simulation results show that the proposed scheme outperforms not only the original policy but also several buffer management policies under message multicasting in delay tolerant networks.
To achieve message delivery in delay- tolerant networks is challenging due to the intermittent connectivity between nodes. This paper proposes a novel message forwarding scheme with dynamic cluster awareness, abbreviated as MDCA, which exploits the aggregation phenomenon of nodes caused by implicit relationship among nodes in a network. Because the aggregation phenomenon of nodes will create some clusters in a network, the delivery probability can be improved if messages can be distributed to each cluster. Thus, the main idea of MDCA is as follows: (1) the node calculates the expected density of nodes; (2) the node determines whether they are in a cluster or not; (3) the node chooses the appropriate relay nodes to carry messages in a cluster; (4), accordingly, MDCA can control the quantity of message copies by referring to the message density. The simulation results show that MDCA results in comparable effects in terms of delivery probability when the mobility models are in accord with human behavior.
Students' Project (Under construction)
For paper review