感知無線電網路之信任模型

Abstract

具有偵測頻譜，進而更有效率使用頻譜的感知無線電網路，被視為未來無線通訊發展中的重要技術。當進一步將感知無線電系統及傳統的無線系統整合成感知無線電網路時，我們可對整體的網路使用效率最佳化。在感知無線電網路中，當我們從頻譜的偵測、取得、直到實際用來傳輸資料，我們考慮的問題已經從底層的連結建立提升到網路層的網路關係連結。在感知無線電網路的網路層設計中，有兩個重要的部份︰信任連結與路由路徑判斷。經由信任模型建立信任連結後，感知無線電可以取得傳統無線電或者其他感知無線電網路的網路參數，進而建立網路的合作關係。這樣的信任模型，讓感知無線電可以在網路上幫忙傳輸資料的同時，也辨認出潛在的風險，這是達成合作式傳輸路徑選擇的重要初始步驟。我們在本論文中提出的信任模型，可以讓網路的節點估計信任合作機率。其中，我們利用觀察網路上其他鄰近點的傳輸行為，進而從過去的紀錄去估計未來的行為模式，並且判斷是否要繼續維持相互的信任連結。 另一方面，感知無線電網路的信任系統必須要能夠即時的適應感知無線電網路的拓樸改變。在感知無線電網路的網路層設計中，最困難的部份在於，無法直接偵測到合作鄰近點的動態變化，為了提高網路的傳輸效率，信任模型必須能快速的對這些拓樸變化做出判斷並且回應。在這篇論文中，我們透過設計感知無線電網路的信任系統，改善網路的使用效率，進而期改善整體感知無線電網路的系統效能。

Devices with cognitive radio capabilities of spectrum sensing to fully utilize radio spectrum have been considered as a key technology in future wireless communications. Primary system and CRs, which can leverage and coexist with legacy systems, thus form cognitive radio networks (CRN) which originally design to improve the spectrum utilization. When we try to further consider network efficiency, the problem turns into optimization in a inter-network manner. In cognitive radio networking function, CRs should establish trust association with neighbors for construction of trusted route. This is an important initial step in network layer design of CRN since the cooperative routing transmission is allowable in the heterogeneous systems. In this thesis, we propose a general trust decision criterion for nodes as receiving association request from neighbors. We exploit unique ID to identify nodes and derive the decision criterion under system-defined constraint. It minimizes the risk of accepting probable malicious users. On the other hand, each node in CRN should observe, analyze, and learn the trust evidence such like packet loss rate, total time delay, and etc. The trust model in this thesis provides a methodology to measure the trust evidence and try to learn and response to the analyzed data meanwhile. It should be not only suitable for dynamic topology variation in CRN but also learn the behavior change in individual node. Then, we can build up the first step when we try to design the network architecture of CRN. When we learn the some special parameters from the trust model, we expect to move forward from trust association to trusted routing in CRN.