第二優先的使用者們要有效的分享頻道在感知型網路是一個很大的挑戰. 為了提升效能, 過去在實體層與媒體層已有許多不同的頻道選擇技術的研究被提出. 在這些技術中, 第二優先的使用者觀察頻譜的空洞, 並且根據頻道的狀態改變他的傳輸頻道. 因此, 第二優先的使用者隨時可以得到更大的頻寬. 雖然使用媒體層來選擇頻寬可以讓第二優先的使用者迅速的得到頻譜的狀態, 但受到傳輸層的流量控制機制所導致的狀態較慢回應, 可能會讓用媒體層來選擇頻寬的方式產生一個較差的效能. 這篇論文中, 使用傳輸層感知頻道選擇技術(TACS)以最佳化傳輸速度. TACS採用跨層式設計, 亦即實體層做頻譜感測, 傳輸層根據其狀態做出頻道選擇的決定, 如傳輸控制協定(TCP)的來回時間, 或是視窗大小, 及流量控制機制. 我們用公式表示TACS的問題為兩個不同的遊戲理論: 非合作及合作. 並且呈現新穎的演算法來提升TCP的傳輸速度. TACS允許沒有執照的使用者可以分散式的選擇頻道. 由模擬結果有看出TACS的方式比目前用MAC的方式解決頻道選擇問題的好.
Effectively sharing channels among secondary users is one of the greatest challenges in cognitive radio network (CRN). In the past, many studies have proposed various channel selection schemes at the physical layer or MAC layer to optimize performance of CRN. In these schemes, secondary users monitor spectrum hole or white space [1] and change its transmitted channel according to states of channels so that they can obtain large channel bandwidth throughout the time. Although selecting channels at the MAC layer can allow secondary users swiftly respond to spectrum states, it may not result in good performance perceived at the transport layer due to slow response of the flow control mechanism employed at the transport layer. This thesis presents a Transport Aware Channel Selection (TACS) scheme to optimize transport throughput of CRN. TACS adopts a cross-layer design framework, in which the Physical layer performs spectrum sensing while the transport layer makes channel decision based on states, such as RTT, congestion window size, and state of flow control mechanism of TCP connections. We formulate the TACS problem as two different games: non-cooperative and cooperative, and present novel heuristic algorithms to optimize TCP throughput. TACS allows secondary users select their channels distributedly. Computer simulation shows that TACS performs much better than current MAC layer based channel selection schemes in TCP throughput.