Compressed Sensing Construction of Spectrum Map in Cognitive Radio Networks and Its Applications

感知無線電網路(cognitive radio networks)由主要網路(primary network)與次要網路(secondary network)所組成，藉由次要網路執行頻譜偵測(spectrum sensing)來充分使用被主要網路浪費的無線電資源(radio resource)，解決現今越來越多無線裝置需要使用無線電資源、然而現有頻譜分配政策卻導致頻譜使用效率低落的棘手問題。另一方面，現存合作式中繼技術(cooperative relay)使感知無線電網路能夠進行多點傳輸(multi-hop transmission)，然而為達到成功的網路功能，則必須仰賴與地理位置相關的無線電資源使用狀況，因此如何獲取此資訊則為傳統頻譜偵測的一大挑戰。 本論文提出具有頻譜與地理位置資訊的頻譜地圖。有別於常見感知無線電網 路受限於臨機鏈結而只能討論統計特性，頻譜地圖成功幫助在高度動態環境下的 多點傳輸路由機制。由於頻譜偵測只能提供感知無線電使用者局部區域的資訊， 使得頻譜地圖的建立面臨困難；我們採用前瞻性的壓縮取樣技術，藉由極少數的 局部偵測結果即可完成頻譜地圖的建立。除此之外，我們使用消息理論分析建立 頻譜地圖的通訊開銷(communication overhead)理論值，並使用擴展圖 (expander graph)將頻譜地圖的概念具體化。最後我們運用頻譜地圖進行可靠的 感知無線電端對端(end to end)封包傳輸，同時保證主要使用者的通訊品質並最 大化次要網路的網路吞吐量。Cooperative relay enables general multi-hop cognitive radio networks (CRN) over cognitive radios (CR) and nodes of primary system (PS). However, successful CR networking such as routing relies on the knowledge of radio resource availability associated with location, which is a challenge of traditional spectrum sensing. We introduce the concept of spectrum map encompassing spectrum and location information to deterministically assist multi-hop routing in highly dynamic environment, while common routing of CRN paying attention to statistical nature of opportunistic links. Since it is not feasible to construct the entire spectrum map as traditional spectrum sensing only knows local information of CR transmitter, we adopt the novel compressed sensing technique to establish spectrum map based on a small number of available local sensing results. Besides directly apply Compressed Sensing theory, we theoretically analyze the communication overhead and practically construct the spectrum map via expander graph. Finally we use spectrum map to reliably route packets of CRs in an end-to-end way, under the guaranteed outage for nodes in PS and maximizing the throughput among cooperative CRs