在感知小細胞網路中不同系統拓樸下具有能源效益的頻譜選擇及資源分配

Abstract

小細胞網路(small cell networks)是一個有很前景的方法，用以解決使用者位於覆蓋空洞或室內環境中接收訊號強度太弱的問題。小細胞網路是由中央的小細胞閘道(small cell gateway(SC-GW))所控制的，它採用了感知無線電的技術，希望能夠藉由存取多個已授權頻譜來增強系統效能。在本篇論文中，我們引進了四種系統拓樸，然後分別去探討它們的頻譜選擇及資源分配的決策。考慮到了小細胞基地台(small cell access point(SAP))的硬體限制，一個完善的頻譜選擇機制是不可或缺的。因此，為了求出目標頻譜，在有限的天線數量的頻譜選擇(spectrum selection with limited number of antennas (SSLNA))方法被提出，並且被應用在大部分的系統拓樸中。根據已經決定好的頻譜選擇策略，並且為了改善系統的能源效益，在完美的資訊回饋下具有能源效益的最佳資源塊及能量分配(energy-efficient joint resource block (RB) and power allocation scheme (EJRP) with perfect information feedback (EJRP-P))和在不完美的資訊回饋下具有能源效益的最佳資源塊及能量分配(EJRP with imperfect information feedback (EJRP-I))以及具有能源效益的聯合資源塊及能量分配(EJRP with no information feedback (EJRP-N))被提出來。當EJRP-P或EJRP-I被採納時，為了避免SC-GW和SAP大量的溝通負載及計算複雜度，具能源效益的最佳化問題被轉變成雙層的問題，分別由SC-GW和SAP來解。除此之外，具有能源效益的聯合頻譜選擇及資源分配(energy-efficient joint spectrum selection and resource allocation (EJSR))被應用在中央控制的系統，它表現的比其他系統拓樸還好，因為SC-GW收集所有的資訊，然後將頻譜選擇及資源分配的決策一起解出來。

The adoption of small cell networks is considered as a promising solution to resolve the poor received signal strength problem experienced by the user equipment (UE) located in the coverage hole or in indoor environment. The architecture of small cell networks, which adopt cognitive radio (CR), centralized controlled by small cell gateway (SC-GW) is further proposed to enhance the system performance by accessing more licensed spectrums. In this thesis, we introduce four types of system topology, and then probe into spectrum selection and resource allocation policies based on each of them. With consideration of hardware limitation of both small cell access points (SAPs) and UEs, a sound spectrum selection mechanism is indispensable. Therefore, the spectrum selection with limited number of antennas (SSLNA), which is applied to most system topologies, is proposed to derive the target spectrum. According to spectrum selection policies, energy-efficient joint resource block (RB) and power allocation scheme (EJRP) with perfect information feedback (EJRP-P), EJRP with imperfect information feedback (EJRP-I), and EJRP with no information feedback (EJRP-N) are proposed to improve the system energy efficiency. To avoid large amount of communicational load and complex computation for both SC-GW and SAPs, the energy efficient optimization problem is transformed into a two-layer problem solved respectively by SC-GW and SAPs when EJRP-P or EJRP-I is adopted. Furthermore, energy-efficient joint spectrum selection and resource allocation (EJSR), which is applied to centralized system, perform better than others because SC-GW collects all information and jointly solves spectrum selection and resource allocation policies.