针对用户呼叫层和分组层的不同运行机制，建立了认知无线电网络呼叫层和分组层相互关系的跨层马尔可夫模型，推导了次用户的呼叫阻塞概率和分组时延的表达式，提出了一种基于跨层建模的随机频谱接入新机制，并对认知无线电网络进行了系统级仿真． 结果表明，文中模型能够准确地评估认知无线电网络的性能，随机频谱接入新机制可同时保证次用户呼叫层和分组层的性能要求．

In view of different operation mechanisms of call level and packet level，a cross-layer Markov model is established to describe the relationship between call level and packet level，the formulas of call blocking probability and packet delay for secondary users are derived，and a new random spectrum access (RSA) scheme on the basis of cross-layer model is proposed． Furthermore，a system level simulation is conducted for cognitive radio networks．Simulated results show that the proposed model helps evaluate the performance of cognitive radio networks accurate-ly，and that the random spectrum access scheme can guarantee the performance of both the call level and the packet level for secondary users．