Cognitive beamforming for secondary spectrum access

• Main Author: Cai, Shengming
• Other Authors: Gong Yi
• Format: Theses and Dissertations
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Wireless communication systems
• Online Access: https://hdl.handle.net/10356/62088
• Institution: Nanyang Technological University

It has been widely accepted that cognitive radio can substantially improve the spectrum utilization compared with traditional spectrum allocation strategies. Cognitive radio technology enables secondary users to coexist with primary users opportunistically or concurrently. In the latter scenario, the primary user allows the secondary user to access its licensed spectrum, as long as the quality of service (QoS) of the primary transmission is guaranteed. Beamforming is a practical way to achieve this goal by utilizing multiple antennas to limit the interference at the primary receiver (PR) to a tolerable level. With transmit beamforming and receive beamforming, a multi-antenna secondary transmitter (ST) with the knowledge of cross channel state information (CSI) can project signals to appropriate directions such that no harmful interference is received at PR, while the secondary receiver (SR) can separate and recover the intended signals reliably. Perfect or imperfect knowledge of cross CSI at ST results in different interference conditions as well as different transmit strategies. The interference received at PR in the perfect cross CSI case is controllable, and hence ST can precisely determine the transmit beamforming directions to gain the maximum benefit under the primary interference constraint. However, such a strategy is not always suitable in the imperfect cross CSI case, due to the uncertainty of CSI error directions and amplitudes. In this case, ST needs to ensure that the interference constraint would not be violated even with the worst CSI error. In this thesis, we first study cognitive beamforming subject to the secondary power constraint and the primary interference power constraint. With bounded error model for the cross CSI, we rewrite the error boundary expressions as a set of linear matrix equalities, and then convert the whole problem into a convex optimization problem to calculate the optimal solution. When no cross CSI is available at ST, we aim at maximizing the secondary sum throughput subject to the interference expectation constraint and interference outage probability constraint at PR, respectively. We prove that under the former constraint, the beamforming problem can be formulated as a conventional MIMO transmission problem, which can be easily solved via the water-filling method. Under the latter constraint, we find that the interference power is related to the eigenvalues of the covariance matrix of the secondary signals, and thus it can be expressed as a linear combination of multiple chi-square distributed variables. By exploiting the impact of the eigenvalues on the interference outage probability and establishing the relationship between the vibrations of secondary covariance matrix and its eigenvalues, we design an algorithm for this beamforming problem. In the second part of this thesis, we focus on beamforming in a multi-antenna cognitive radio system with transmission of multiple secondary data streams subject to the individual signal-to-noise ratio constraint per secondary data stream and the primary interference power constraint. The objective is to minimize the secondary transmit power consumption. Both perfect and imperfect cross CSI cases are considered and in both cases, the beamforming feasibility is tested. By exploiting the individual SNR constraints, we formulate the cognitive beamforming problem as an optimization problem on the Stiefel manifold. For zero forcing beamforming, we derive a closed form beamforming solution. For nonzero forcing beamforming, we prove that the strong duality holds for the nonconvex primal problem and thus the optimal solution can be easily obtained by solving the dual problem. In the case of imperfect cross CSI, we apply the S-procedure method to reformulate the problem and provide an algorithm to obtain the beamforming solution. Finally, we raise some interesting open problems for the future work.