Resource Optimization For Cognitive Radio Networks Under The Nonlinear Energy Harvesting Model

Cognitive radio networks combined with wireless energy harvesting is a key technology to alleviate the problem of spectrum scarcity and energy shortage.However,it is more difficult to estimate the channel state information and guarantee the system robustness due to the energy harvesting mechanism and the increasing network density.Moreover,the robust resource allocation for cognitive radio networks based on linear energy harvesting model cannot be appropriate in practice due to the nonlinear characteristic of the practical energy harvesting circuit.In order to efficiently harvest wireless energy and improve the robustness of the system,this paper studies the resource optimization of cognitive radio under the practical nonlinear energy harvesting models,and proposes optimal multi-objective robust beamforming schemes and an optimal resource allocation strategy based on deep reinforcement learning.The main contributions are summarized as follows:In this paper,a robust resource allocation problem is studied to minimize the consumed power and maximize the harvested power in a multiple-input single-output cognitive radio network with simultaneous wireless information and power transfer under imperfect channel state information.An optimal robust beamforming is designed under two practical non-linear energy harvest models.Simulation results show that there is a tradeoff between the transmit power and the harvested energy and compare the performance of energy harvesting under two different non-linear energy harvest models.The results also quantify the effect of the sensitivity threshold of the energy harvesting circuit on the harvested power.The resource allocation strategy mentioned above relies on the mathematical model of channel state information estimation error,it cannot achieve optimal performance under actual channel conditions.The resource optimization technique based on deep reinforcement learning does not depend on the error model of channel state information estimation,which effectively reduces the influence of the performance degradation caused by the inaccuracy of the channel estimation error model.In this paper,a resource allocation problem including power allocation and dynamic channel select is studied in a cognitive radio network under the nonlinear energy harvesting model.By modeling the problem into Markov decision process,an optimal resource allocation strategy based on deep reinforcement learning is proposed.Simulation results show that the performance of cognitive networks under the nonlinear energy harvesting model is better than under the linear energy harvesting model.