在隨建即連行動網路中利用階級隨機網路編碼設計之群播協定

Abstract

現今，在手持裝置普及的環境中，進行多方通訊的需求與日俱增，由於行動節點的移動造成頻寬及連線的不穩定，網路拓樸更動較為頻繁，導致在使用群體協定(multicast protocol)傳輸資料時，資料到達各個接收端的正確率容易隨著節點的移動速度而起伏。在此環境中，該如何設計一具有低延遲與高正確率的多方通訊機制，便是十分值得研究的方向。 本論文提出一套植基於隨機網路編碼(random network coding)的分散式多方通訊機制，各節點可藉由在封包傳送時夾帶其統計的網路狀況及封包收送情形，動態地決定封包的發送時機。相較於固定時間發送封包的隨機網路編碼(timer-based random network coding) – Codecast技術，本論文提出之方法可更加有效地利用有限的網路頻寬。為驗證本方法的效能，我們以接收端接收到的資料正確率、點與點的傳輸延遲(end-to-end delay)與封包負擔(packet overhead)三項做為評比多方通訊服務品質的指標，模擬實驗結果顯示，採用本方法後，接收節點所接收到的平均資料正確率與Codecast相近，點與點的傳輸延遲比Codecast降低了10%，封包負擔更改善了100%，顯示本論文所提出之方法能更有效的利用行動隨意網路的有限資源，滿足在行動隨意網路中進行多方通訊的應用需求。

Nowadays, the growing popularity of mobile device causes the stronger demand of multiparty communication services than before. Due to the node mobility, the bandwidth and the link quality are unstable and the network topology changes frequently. Moreover, unexpected packet loss may result from fading, environment interference, or packet collision. These characteristics attract many research interests of designing a multiparty communication protocol with low end-to-end delay and high data delivery ratio. In this paper, we propose a distributed multiparty communication protocol based on rank-based random network coding. Mobile nodes collect nearby network information and transmit encoded packets with the collected statistical data. The node receiving the encoded packets can adaptively determine when to forward the encoded packet according to the rank of current generation and collected information from neighbors. Comparing with timer-based random network coding – Codecast, our approach could utilize network bandwidth more efficient. The simulation results showed that our proposed method improves both the network packet overhead and the end-to-end delay while maintaining similar data delivery ratio. Comparing with Codecast, our proposed method can improve the packet overhead by up to 100% and reduce the end-to-end delay by 10%. The experimental results demonstrated that our approach accommodate dynamics of mobile ad hoc network and utilize limited resources efficiently.