以賽局論為基礎之感知無線電動態頻譜接取

Abstract

動態頻譜接取伴隨著感知無線電的引進，被認為能有效解決現今固定頻譜分配政策下，所導致的頻譜使用率過低的情形。實際上實現動態頻譜接取機制必須是分散式的。感知無線電自我偵測未被使用的頻譜，並調整傳輸參數以在該頻帶上傳送。在動態頻譜接取中，最關鍵的問題之一是必須解決感知無線電之間對頻譜的競爭，同時必須保證有證照之使用者的服務品質。由於感知無線電分散式和自主的特性，賽局論提供在預測感知無線電的行為上，提供了一個適合的模型，使的我們可以藉由預測感知無線電的行為，設計出相應的系統參數，讓感知無線電能執行在我們所期望的工作區域。在這篇論文中，我們考慮動態頻譜接取中的兩個問題。在第一部分，我們處理與頻譜接取中的通道選擇問題。我們考慮了因為有限的頻譜偵測技術，造成感知無線電只能得到部份的頻譜資訊。在這樣的環境下，我們基於奈許平衡設計出的頻譜選擇策略，即使是在許多使用者的網路中，仍能維持好的表現。在第二部分，我們考慮如何解決在時槽式系統中，在主要系統和感知無線電不完美同步造成的時間誤差下，實現動態頻譜接取的方法。主要系統和感知無線電間的時間誤差，會造成感知無線電頻譜偵測時的錯誤。增加感知無線電偵測頻譜的時間以彌補時間誤差，是一個直覺上的做法。然而，從感知無線電自私的觀點而言，增加偵測頻譜的時間也會造成吞吐量的降低和易損期的增加。為了讓自私的感知無線電能自發地提高頻譜偵測時間，我們找出了相對應的系統最佳運作區域，藉由選擇適當的系統參數，主要系統的服務品質和頻譜使用率都能得到保證。即使在存在時間誤差的情形下，我們提出的方法仍可以保證系統的效能。

Dynamic spectrum access (DSA) with the introduction of cognitive radio (CR) is considered as a promising technology in resolving the underutilization of current fixed spectrum assignment policy. Realistic dynamic spectrum access mechanism for cognitive radio networks (CRN) shall be distributed. CRs sense for spectrum opportunity and adaptively change the transmitting parameters to utilize the spectrum. One of the most critical issue in DSA is to resolve the competition among CRs and guarantee the QoS of licensed users. Due to the distributed and autonomous property of CRs, game theory provides a well-suited model in predicting the behavior of CR users, making it possible for us to design the system parameters accordingly to enforce CRs act at the desired operating region. In this thesis, we consider two issues in DSA from game theoretical point of view. In the first part, we deal with the channel selection problem in DSA. The limitation of sensing capability which result in partial spectrum knowledge is considered. The channel selection strategy is proposed basd on Nash equilibrium solution, which performs well even in large networks. In the second part, we practically consider the DSA in slotted primary system in presence of timing misalignment among primary system users and CRs. Timing misalignment introduce spectrum sensing error for CRs to detect primary system users and hence result in degradation in system throughput. Increase the sensing time that compensate the timing misalignment may be a straightforward solution. However, for selfish CR users, increased sensing time means larger vulnerable period in spectrum competition and degradation in CR throughput. We identify the optimal operating region of asynchronous DSA in which selfish CRs self-enforcingly protract the sensing time such that the primary user is protected and spectrum utilization is maximized. We show that under the properly chosen system parameter, system performance is assured even in presence of timing error.