應用於多使用者、多輸入多輸出正交分頻多工存取感知無線電系統的自由度根基頻譜偵測器和基頻處理器

• Main Authors: Lin, Jung-Mao, 林榕茂
• Other Authors: Ma, Hsi-Pin
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/21127258976733714283

博士 === 國立清華大學 === 電機工程學系 === 99 === Ubiquitous wireless applications drive the growing demand for more spectrum access opportunity and higher data rate mobile communications. The scarcity of spectrum resources and the poor spectrum utilization urges the re-examination of old spectrum allocation policies. The concept of spectrum pooling enables public access to a common spectrum pool which is comprised of several licensed frequency bands hosted by licensed users. Rental users can access the unused frequency bands in the common pool with the permission of the licensed users. However, spectrum pooling requires the improvement of control and infrastructure framework for the licensed users, which makes the implementation inflexible. On the contrary, cognitive radio (CR), which allows unlicensed users (or usually called secondary users in CR networks) to access the available spectrum autonomously, provides a more feasible solution to improve spectrum efficiency. Moreover, a CR system incorporating orthogonal frequency division multiplexing (OFDM)/orthogonal frequency division multiple access (OFDMA) transmission can perform spectrum access in a highly flexible and adaptable manner to maximize spectrum efficiency; meanwhile, the multicarrier modulation technique can also achieve a reliable high data rate transmission. Therefore, an OFDM/OFDMA-based CR system has become a promising solution to solve the spectrum crowding and spectrum deficiency problems. In this dissertation, design challenges of an OFDM/OFDMA-based CR system are investigated. To overcome the interference problem caused by licensed users (or usually called primary users in CR networks) in spectrum sensing, a degree of freedom (DoF)-based spectral detector that is applicable to wideband strong interference environments is proposed. By exploiting the DoF obtained from spectral estimation, the proposed detector provides a higher DoF for the improvement of detection performance. Besides, the detector can perform two-phase sensing with adjustable spectral resolution. This makes the system robust to strong interference environments and compatible with wideband spectrum sensing applications. Moreover, to achieve a reliable and flexible OFDM transmission in a CR system, multiple access schemes and synchronization issues should be considered. In this dissertation, a multiuser OFDMA system, which can offer both uplink and downlink communications, is presented. For downlink communications, the proposed receiver consists of carrier frequency offset (CFO)/sampling frequency offset (SFO) estimation/compensation, channel estimation/equalization, and sampling add/drop function for symbol timing synchronization. As for uplink communications, a configurable (for single-input singleoutput (SISO)/multiple-input multiple-output (MIMO) and OFDM/OFDMA communications) and power efficient multiuser MIMO-OFDMA baseband processor will be considered. To solve the CFO problem in multiuser transmission, an inter-carrier interference (ICI)-based CFO estimator is implemented based on an iterative search criterion of maximum signal-to-interference-plus-noise ratio (SINR). Compared to state-of-theart methods, the proposed CFO estimator is more robust to transmission configurations (MIMO and multiuser) and CFO variations. Moreover, the authors propose an efficient architecture that reduces the hardware complexity by 78% compared to the direct implementation architecture by employing Taylor series expansion for ICI/multipleaccess interference (MAI) cancellation. Meanwhile, a 2D linear channel estimator is also proposed to assist the CFO estimator and track the time-variant multipath channel. Two kinds of MIMO detector, vertical Bell Laboratory layered space-time (V-BLAST) and V-BLAST with maximum likelihood (V-ML), are adopted to minimize output latency and achieve the best ML bit-error-rate (BER) performance. An application-specific integrated circuit (ASIC) fabricated by 0.13 μm 1P8M CMOS technology is measured with 2.31 Mbps/mW power efficiency and less than 1.5 dB implementation loss. In addition, the whole transceiver is integrated and verified by a system-on-chip (SoC) platform to demonstrate its efficacy.