| Abstract: | 無線網狀網路 (WMNs) 由於它的self-organized, self-configure和較低的建置成本等優點吸引許多學術和企業投入研究和探討。 IEEE 802.11 TGs積極的制定無線網狀網路的標準也使得WMNs研究更受到重視，並預定在2008年7月完成802.11s的標準。無線網狀網路的主要特性在於使用multi-radio和 multi-channel 並且應用於multi-hop的無線網路環境下。在過去的Ad-hoc Network 針對提升無線網路的傳輸效能做了許多研究，然而Ad-hoc 下所做的研究並不完全適合於WMNs，因為在ad-hoc下設計的routing protocol並沒有把WMNs的特性考慮進去(multi-radio,multi-channel etc.)。另外，目前針對WMNs的研究大部份主要針對不同的網路協定層獨立研究，以提高網路效能，即使IEEE 802.11 TGs 雖然提出了一個預設的繞徑方法 HWMP和一個可選擇的繞徑方法 RA-OLSR，但並未充分利用各個網路層協定的資訊來選擇最佳的繞送路徑。但是封包在網路中繞送時，封包的傳輸是由各網路協定層交互影響，對單一網路協定層做最佳化無法完整的考量整體網路的變動對封包的影響，尤其video streaming對packet loss和delay的高敏感。因此透過Cross Layer design方法來設計routing protocol，利用各網路協定層的資訊，例如搭配動態調整physical layer參數或application layer的編碼技術等使影音封包等在網路中傳送時能有最好的效能，並提升整體網狀網路的效能。我們第一年的研究目的，將針對physical 和 network layer 設計一套cross layer 的routing protocol，每個flow將利用不同網路協定層的資訊決定我們的繞送路徑。我們將分別先以inter/intra flow interference考量繞送路徑，再以physical層使用multi-power為考量，研究intra-flow的每個hop該如何選擇power，以達到最佳效能。最後，再把inter/intra flow interference與不同power的影響，在作繞路演算時同時考量，以達到最佳網路效能。第二年的研究目的，將針對支援P2P streaming的應用服務為考量，設計一個與P2P Application跨層routing protocol。為達到更好的streaming QoS，我們建議video的編碼技術將原本的MDC修改成bias MDC，以便我們在計算P2P的繞路時，可以試著找出兩個(或以上)disjoint trees (Steiner tree)，根據disjoint trees上的path來個別傳送不同MDC video streaming資料。在建構此trees時，我們將整合第一年的研究結果，設計更完善的支援P2P streaming資料傳送的跨層網路路徑演算法。透過第一年和第二年的研究結果結合，我們希望設計出一套cross layer routing protocol 使得點對點video streaming的應用，在multi-power, multi-radio wireless mesh network 下傳輸時，能夠有好的服務品質保證。
Wireless Mesh Networks (WMNs) have the properties of self-organization, self-configuration, and low buildup cost. These advantages attract many academics and industry for research and discussions. Thus, the IEEE 802.11 TGs develops the standard for WMNs and will be finalized in July, 2008. WMNs can be operated in multi-radio, multi-channel, multi-hop environments. For traditional ad hoc networks, there are many routing protocols in the literatures that can improve the throughput only focused on single network protocol layer independently. Though the IEEE 802.11 TGs proposed a default routing method, called HWMP, and an optional method, called RA-OLSR, they wouldn’t utilize information from various network protocol layers to find the routes. Here, we try to propose a cross-layer routing protocol design that takes into account physical and application layers to enhance the overall network performance more efficiently. For the fist year, we will design a cross-layer (physical & network) routing protocol on a per-flow basis. We will first base on intra/inter flow interference to compute the path. Then, we further adjust different power levels to reduce the interference and enhance the overall throughput while computing the routing path simultaneously. For the 2nd year, we will focus on supporting P2P streaming applications. We will design a cross-layer (network & application) routing protocol on a per-application basis. To achieve better QoS, we suggest the MDC (multiple Description Coding) for video to be modified as bias MDC. In this way, we can compute two (or more) disjoint trees (Steiner tree), and then transport the 2 bias MDC streams into these two trees based on their quality or resource availability to support P2P streaming with QoS. The computation of trees will be further integrated with the results from the first year. Through the integration of results from the first and second years, we will develop a cross layer routing protocol for P2P streaming applications over multi-power, multi-radio wireless mesh networks. |
