Physical layer security protects the transmitted data by trying to prevent the information from reaching the adversary. Adding artificial noise (AN) to the transmitted data helps to achieve that goal. In this paper, AN technique is used to protect the transmission of the secondary user (SU) from an eavesdropping attack in an energy harvesting cognitive radio network with hybrid channel access. At the beginning of each time slot, the SU is faced with multiple alternatives concerning the energy harvesting/transmission mode. The uncertainty in the PU’s activity adds another challenge to the decision of the SU. The mixed observable Markov decision process (MOMDP) is selected to model SU’s decision process in order to handle these challenges. The optimal policy is solved using the point-based value iteration algorithm (PBVI) to mitigate the time complexity of the basic value iteration algorithm. Moreover, the fraction of power allocated to AN (AN ratio) is optimized and an adaptive policy is introduced. The results show the superiority of the proposed model compared to other models in literature in terms of secrecy capacity and throughput. The effect of multiple system parameters on the model performance is studied and the optimal AN ratio is identified for different system scenarios. Moreover, an insight of the resultant policy is provided and the proposed adaptive model is proved to be superior compared to the fixed AN ratio models.

物理层安全性通过尝试阻止信息到达对手来保护传输的数据。在传输的数据中添加人工噪声（AN）有助于实现该目标。在本文中，AN技术用于保护次要用户（SU）的传输，使其免受混合信道访问的能量收集认知无线电网络中的窃听攻击。在每个时隙的开始，SU面临着与能量收集/传输模式有关的多种选择。PU活动的不确定性给SU的决策增加了另一个挑战。选择混合可观察马尔可夫决策过程（MOMDP）来建模SU的决策过程，以应对这些挑战。使用基于点的值迭代算法（PBVI）解决了最优策略，以减轻基本值迭代算法的时间复杂性。此外，优化了分配给AN的功率比例（AN比率），并引入了自适应策略。结果表明，相对于文献中的其他模型，该模型在保密容量和吞吐量方面具有优势。研究了多个系统参数对模型性能的影响，并针对不同的系统场景确定了最佳AN比率。此外，提供了对结果策略的见解，并且与固定的AN比率模型相比，所提出的自适应模型被证明是优越的。结果表明，相对于文献中的其他模型，该模型在保密容量和吞吐量方面具有优势。研究了多个系统参数对模型性能的影响，并针对不同的系统场景确定了最佳AN比率。此外，提供了对结果策略的见解，并且与固定的AN比率模型相比，所提出的自适应模型被证明是优越的。结果表明，相对于文献中的其他模型，该模型在保密容量和吞吐量方面具有优势。研究了多个系统参数对模型性能的影响，并针对不同的系统场景确定了最佳AN比率。此外，提供了对结果策略的见解，并且与固定的AN比率模型相比，所提出的自适应模型被证明是优越的。