[en] In this paper, an RF-powered cognitive radio network is considered, in which the secondary users are powered by an RF energy harvester (Rectenna). Unlike most existing works, we consider a realistic Rectenna characteristic function, and derive the actual amount of harvested energy and thus, the resulting actual energy level of the secondary users. We consider a system architecture at which simultaneous energy harvesting and data transmission for each secondary user is possible. We introduce a strategy to manage the challenge of network throughput decreasing due to lack of the secondary users’ energy, via selecting the best possible channels for energy harvesting and simultaneously by allocating the best channels for data transmission. Therefore, we implement cognition in spectrum utilization and in energy harvesting. We show that the amount of harvested energy affects the available energy of the secondary user and consequently the throughput, therefore, the channels selection to maximize energy harvesting affects the network throughput. To maximize the network throughput, the Hungarian algorithm is employed, and then, an algorithm with lower complexity based on the matching theory is proposed. Finally, we compare our proposed approach with some existing benchmarks and show its high performance in energy harvesting and system throughput.