感知無線網路異質性頻譜之拍賣研究

Abstract

在本篇論文中，我們使用拍賣理論與賽局理論的角度來研究感知無線電網路 頻譜買賣。我們考慮一個多主要網路與次要網路所構成的感知無線電網路。在主 要網路的主要用戶擁有頻段的執照，為此拍賣的賣家，它們提供閒置的頻段來增 加獲利；在次要網路的次要用戶沒有頻段的執照，為此拍賣的買家，它們購買閒 置的頻段藉此來傳輸資料。為了處理此問題，我們提出在多個主要用戶與多個次 要用戶間的異質性頻譜拍賣來找出最佳的頻段分配藉此將利益最大化。在付款方 面，我們使用改良的Vickrey–Clarke–Groves 付款機制來決定當次要用戶分配 到主要用戶的頻段時該付多少錢。當主要用戶與次要用戶的數目增加時，找出最 佳的頻段分配將會變成一個NP 困難的問題，為我們提出了次最佳分配來解決此 問題，雖然效能可能無法與最佳份配一樣好，但可節省非常多時間且效能接近最 佳分配。

In this thesis, we study the problem of spectrum trading in cognitive radio (CR) networks using auction theory and game theory. Particularly, we consider a CR network consisting of a primary network and a secondary network. The primary users (PUs) who have licensed spectrum bands in a primary network work as the sellers to supply idle spectrum bands so as to increase revenue and the secondary users (SUs) who do not have licensed spectrum bands in a secondary network work as the buyers to buy idle spectrum bands so as to transmit data. To deal with this problem, we propose the spectrum trading auction between multiple PUs and multiple SUs with heterogeneous spectrum bands to find optimal allocation so as to maximize utility. For payment function, the modified Vickrey–Clarke–Groves payment to decide how much SU should pay to PU for the allocated spectrum band is used in this thesis. When the number of PUs and SUs increase, the problem of finding optimal allocation will become NP-hard. A sub-optimal allocation to tackle this problem is proposed. Although the performance of the sub-optimal allocation may not as well as optimal allocation, it can save much time and close to optimal allocation.