同儕網路(Peer-to-peer Networks)為近年來最熱門的研究議題之一,相關服務獲得廣泛地應用。同儕網路依架構特性的考量,一般採取用分散式雜湊表(Distributed Hash Table)的結構化方式建立其應用與服務。然而現存的分散式雜湊表大都未考量底層網路狀況,使得 Underlying Layer與Overlay Layer之間產生拓樸落差問題(Topology Mismatch Problem),使得實際在網路上的運作效能大打折扣。而在動態的網路環境下,拓樸落差問題的影響更趨嚴重。在此篇論文中以分散式雜湊常使用的Chord查詢架構作為探討的基本架構,以跨層(cross-layer)的角度來考量分散式雜湊表的建構與查詢方式,提出Hierarchical Swift RLM(HS-RLM)。HS-RLM利用Random LandMarking(RLM)所提出之動態地標技術建立相依的拓樸且結合MANET路由資訊來改善節點的查詢與資料的維護程序。因此,HS-RLM能夠有效降低拓樸落差問題的影響,提升分散式雜湊表查詢的效能。
In these days, most of the P2P systems use distributed hash table (DHT) on a structured architecture to provide services and applications. Most of the existing approaches using DHT did not consider the practical underlying routing overhead when executing the searching process on the logical overlay layer. The actual performance on latency and energy hence is degenerated. This is what we called the Topology Mismatch Problem. In this paper, we proposed a cross-layer design to minimize the effect of Topology Mismatching Problem in MANET. We integrated DHT with routing information and proposed a novel P2P system for MANET. We last validate our protocols through extensive experiments. In comparison with other algorithms, such as Chord, and RLM, our experimental results show that the proposed approaches outperform the others.