This paper is concerned with the secrecy rate based optimization problems in a mixed radio frequency (RF) / free space optical (FSO) spectrum sharing network, wherein multiple single-antenna secondary users (SUs) are connected to a decode and forward (DF) single-antenna relay through RF links and the relay is connected with the secondary destination through an FSO link. The RF and FSO channels are assumed to follow Rayleigh and Gamma-Gamma distributions with the effect of pointing errors, respectively. The three schemes, i.e., direct transmission, single-user cooperative jamming and multi-user cooperative beamforming jamming, are proposed to improve the physical-layer information security in the presence of a single-antenna eavesdropper. The mutual interference between the primary and secondary networks is considered. The problems are to jointly optimize the SU transmit power and jamming power for the best secrecy performance under both the RF and FSO dominant cases. Such problems are nonconvex. The Dinkelbach approach is adopted to solve the optimization in the direct transmission and single-user cooperative jamming schemes and a two-level optimization approach with semi-definite relaxation (SDR) is applied to obtain the solution for the multi-user cooperative beamforming scheme. Moreover, the SDR optimality is proved by Karush-Kuhn-Tucker conditions analysis. Simulation results are provided, and the results show that the proposed multi-user cooperative beamforming jamming can attain the higher achievable secrecy rate than the zero-forcing beamforming scheme and other two suboptimal designs.

Wideband sensing-based cognitive radio with simultaneous wireless information and power transfer can be designed for efficient spectrum and energy usage. We investigate maxmin fairness energy-harvesting optimization problem in such a system by taking into account of the individual link fairness. In particular, we maximize the energy harvested by the worstcase link when the problem is subject to the rate requirements, transmit power constraint, interference power constraint and subchannels assignment constraint. Due to the nonconvexity of the formulated problem, we relax the integer variable and introduce an auxillery variable. The Lagrangian and subgradient methods are adopted to obtain a suboptimal solution. Simulation results are presented to verify the fairness performance of the proposed algorithm, and to reveal a new tradeoff between the network harvested energy and link fairness.