Spectral/Power-Efficient Transceiver Design for Multi-cell MIMO Cognitive Radio Networks

Abstract

This thesis discusses how to design the spectral or power-efficient transceiver for multi-cell cognitive radio networks (CRNs) with multiple antenna techniques. The transmit and receive beamforming techniques in CRNs have been the subject of many recent studies to efficiently harmonize the primary system with the secondary systems. For spectrum-sharing-based CRNs, there are two fundamental criteria involved in optimizing the beamformer. One is the throughput maximization of the secondary system with the constraints of transmit power for secondary systems and interference power for the primary system. The other is the transmit power minimization of secondary systems under their minimum quality-of-service (QoS) constraint and the interference constraint for the primary system. The former approach provides the maximum achievable rate of the secondary system with a given transmit power from the information-theoretic perspective. In practice, however, each secondary user has a certain QoS requirement that depends on the amount of information provided by the secondary transmitter. When the QoS requirement is satisfied, the information for the secondary users can be successfully provided through wireless channels.In spectrum-sharing-based cognitive radio networks, multiple secondary systems can access a licensed spectrum to better utilize scarce radio resources. When the multiple secondary transmitters are co-located, the weighted sum-rate of the secondary users (SUs) is mainly limited by the inter-cell interference (ICI). With limited cooperation among co-located secondary transmitters, an algorithm for decentralized beamforming with power allocation via dual decomposition is proposed. To maximize the weighted sum-rate of the SUs, the proposed decentralized algorithm efficiently mitigates the ICI by the undesired leakage power limitation at each secondary transmitter. Because the channel information is not perfectly known at the transmitter in practical applications, I also develop a decentralized robust beamformer. To efficiently design the robust beamformer, a convex problem is formulated by semi-definite relaxation. Simulation results show that the proposed algorithm with perfect channel state information (P-CSI) efficiently maximizes the weighted sum-rate performance by the undesired leakage power limitation. For an imperfect CSI with a small error bound, the proposed robust beamformer approaches the performance of a P-CSI case, without causing harmful interference to the primary user (PU).For the transmit power minimization of secondary systems under their minimum QoS constraint, I propose both centralized and distributed beamforming algorithms. The proposed algorithms optimize the total transmit power of secondary systems, while maintaining the interference to PUs below a certain threshold and satisfying the QoS constraint for each secondary system. The centralized algorithm achieves the optimal transmit power by exploiting the virtual uplink-downlink duality using the knowledge of the channel state information for all the secondary links. However, the assumption of global channel knowledge at each secondary system may not be allowed in practical applications for multi-cell coordination. To address this problem, I design a distributed transceiver beamformer that satisfies the interference constraint to protect PUs. On the basis of this distributed beamformer, I also propose power allocation algorithms that guarantee the QoS for secondary systems. Distributed beamforming and power allocations operate iteratively to minimize the total transmit power. Simulation results show that the distributed algorithms achieve a near-optimal transmit power while satisfying both the QoS and interference constraints.To obtain the optimal transmit or receive beamforming vector, accurate CSI is required. For the performance improvement of linear minimum mean square error (MMSE) channel estimation, a power delay profile (PDP) estimation technique is investigated. For practical applications such as multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems, only the pilot symbols of all transmit antenna ports are used in estimating the PDP. The distortions caused by null subcarriers and an insufficient number of samples for PDP estimation are also considered. The proposed technique effectively reduces the distortions for accurate PDP estimation. Simulation results show that the performance of MMSE channel estimation using the proposed PDP estimate approaches that of Wiener filtering due to the mitigation of distortion effects.TV white space (TVWS) is perceived as the most suitable frequency bands for CR, although the technology is conceptually workable in any frequency band. Experimental and simulation results for the use of TV-band devices (TVBDs) in TVWS are presented, considering the presence of interference by incumbent services. Digital TV (DTV) services are major incumbent services currently operating in the TV bands. With DTV service, co-channel and adjacent channel deployment scenarios of TVBD networks are introduced. To safely protect the incumbent service, a minimum separation distance from the DTV protected contour, which is called the keep-out distance, is required. We estimate the keep-out distance for different ranges of TVBD transmit antenna height by using several propagation models and measurements of ultra-high-frequency signals in Korea. We also investigate the hidden node problem for the spectrum sensing operation mode of TVBDs. According to the results of these measurements, the hidden node margin should be at least 38 dB in order to protect DTV service. Finally, the service coverage reduction of TVBD networks caused by neighboring DTV service is discussed. It is shown that the service coverage of a wireless local area network system decreases about 50\% by co-channel interference from neighboring DTV service when the field strength of the DTV received signal is 41 dBu.