感知無線電網路下考慮傳輸干擾之頻譜偵測與共享

Abstract

感知無線電已被提出來作為提升頻譜使用效率的技術之一，其中頻譜偵測與頻譜共享為其核心技術。次要用戶在使用主要用戶的頻譜之前，必需先進行頻譜偵測以確認頻譜的使用狀態。在先前的研究中著重於最小化偵測漏檢和假警報的機率，然而頻譜共享造成對主要用戶的影響卻未善加考慮，導致傳統的頻譜偵測在次要用戶對主要用戶干擾機率小時無法提升頻譜的使用效率。因此我們提出一個考慮傳輸干擾機率的頻譜偵測模型，當次要用戶傳輸而干擾主要用戶的機率小時，該次要用戶可容許較高的漏檢機率，而採取較大偵測閾值以降低假警報、提高頻譜使用率。除了單節點的頻譜偵測之外，我們也將此感測模型推廣到合作式偵測，並提出一個有效率的演算法來找尋各個合作成員的最適偵測閾值。合作式偵測透過融合數個次要用戶的偵測結果，做出一個統一的決策以降低偵測錯誤的機率，然而當不同的次要用戶對主要用戶有不同的干擾機率時，不同次要用戶對漏檢與假警報的機率將產生不同的取捨與要求。合作式偵測雖然可以降低偵測錯誤，但同時也失去對各個次要用戶的差異做出最佳化，因此我們進一步探討次要用戶之間如何產生適當的分群來進行合作式偵測以達最佳利益。利用群組的獨立性、次序性及連貫性，我們設計出一個更有效率的演算法來找出最佳分群方式，同時我們也以賽局的角度切入此問題，探討次要用戶間有意願進行合作偵測的條件，達到簡化判斷合作意願所需的次數。此外，我們也透過賽局設計以促進群組內部成員的合作性，以避免利己主義的成員和無效率的納許平衡。 針對頻譜共享技術，我們探討次要用戶如何主動地透過人數控管及傳送功率控制，使得即使主要用戶仍在頻譜上使用，次要用戶也能與之共存。透過連結最佳化理論與非合作賽局分析，我們在各個次要用戶的利益函數中設計出一個最優的補償代價，該設計同時考慮次要用戶對主要用戶的干擾量，也考量次要用戶之間對彼此的干擾，使得納許平衡點可以接近整體網路利益最優的情況。此外，我們也重新設計次要用戶在每回合賽局中的決策函數，將對其他用戶的反應預測納入考量，並避免自身決策的劇烈改變，使得設計出的分散式干擾控制可以在更少的回合數內收斂到更高的頻譜使用效率。為了因應頻譜環境的快速變化，我們還發展出一套快速修正的演算法，使得次要用戶可以更快速地適應頻譜環境，避免經常性地執行冗長而精細的干擾控制。綜合言之，本論文將次要用戶傳輸對主要用戶的干擾納入頻譜感測的設計，透過鄰近主要用戶的次要用戶量測頻譜與回報，部分次要用戶可藉由更精細的干擾控制與主要用戶同時共享頻譜。 透過各種模擬實驗驗證，我們所提出的方案的確能有效提升1∼2倍的頻譜使用效率。

Spectrum sensing is the first important step to discover available spectrum holes left by primary users. Spectrum sharing is the second step to enable dynamic spectrum access of secondary users. In this dissertation, we design spectrum sensing and sharing with the consideration of each other in order to improve the performance of both important steps. In spectrum sensing, most conventional approaches consider the metrics of probabilities of missed detection and false alarm to ensure protection of primary users and maximum utility of secondary users. We find that spectrum sensing based entirely on the two metrics is unable to maximize spectrum utilization for dynamic spectrum access. We show that, to meet the requirement of the probability of missed detection, conventional approaches can unnecessarily increase the probability of false alarm in scenarios with good opportunity of spectrum reuse, thus lowering the ability to leverage spectrum holes. To address this problem, we propose an interference-aware metric for spectrum sensing that considers both probabilities of interference and missed detection. By using the tool of monotonic programming, we develop an efficient algorithm that can solve the optimal detection threshold either for a single detector or for cooperative sensing of multiple detectors. Although cooperative sensing can make one accurate decision, secondary users with different interference probabilities desire different trade-offs of probabilities of missed detection and false alarm. We then consider that secondary users can form different coalitions for cooperative sensing and adjust the corresponding detection threshold. An efficient algorithm based on combinatorial optimization is proposed to solve the optimal coalition structure that maximizes the total utility of secondary users. The tool of coalitional games is applied to analyze the cooperation between secondary users, derive the rules for cooperation and provide a distributed algorithm of self-organized coalition formation. Missed detection itself is not harmful to primary users, but the interference caused by spectrum sharing of secondary users does. In spectrum sharing, we study the problem of secondary users who can actively control the interference to primary users. A hybrid cooperative scheme for overlay and underlay secondary users is proposed. Overlay users who have detected primary users can provide the measured interference temperature to underlay users for better interference control. The interference control is performed through selecting a subset of users for spectrum sharing and controlling the transmission power of the admitted users. An optimization of joint admission and power control is formulated and as the base for designing the optimal price in the price-based game theory. Distributed decision-making of admission and power control is designed by jointly considering the bast response of the game, intended movement and smooth reaction such that a near-optimal equilibrium converges within a small number of iterations. To deal with highly dynamic spectrum environments, a quick algorithm of interference-margin adaption is proposed to re-optimize any unfeasible or suboptimal solutions. For protection of undetectable primary receivers, a lower bound of the sentinel density is derived given a tolerable error of interference temperature. Compared against conventional techniques, interference-aware spectrum sensing considers the cost of transmission and can potentially discover more spectrum holes. Spectrum sharing with cooperation of overlay and underlay users can achieve more accurate interference control and result in better spectrum utilization. We show through evaluation results that the proposed techniques can achieve a significant performance gain compared to conventional approaches that neglect the effect from the other step of spectrum sharing or sensing.