中文摘要：

令授权用户在超时隙边界处抢占频谱，认知用户基于时隙传输数据，提出了一种基于超时隙的集中式频谱分配策略。结合当前时隙在超时隙中的编号及系统中认知用户数据包的数量，建立了一个二维Markov链模型。利用矩阵几何解方法求出系统的稳态分布，给出了认知用户的数据包平均延迟、吞吐量以及信道切换率等性能指标的表达式。实验结果表明，所提出的集中式认知无线网络频谱分配策略有效地提高了认知用户的吞吐量，降低了信道切换率。考虑不同的性能指标之间存在着折中关系，建立了一个系统收益函数，给出了超时隙大小的优化设置方案。

英文摘要：

Letting primary users access spectrum with preemptive priority at the boundary of a super-slot, cognitive users transmit their data packets in a slot structure, a novel centralized spectrum allocation strategy for cognitive radio net- works is proposed. Combining the sequence number of the current slot in a super-slot and the number of data pack- ets for the cognitive users in the system, a two-dimensional Markov chain model is established. By using the method of matrix geometry soluti nitive users, such as the ons, the system model in ~eady state is analyzed, and the performance measures for the cog- throughput, the average latency of data packets and the spectrum switching ratio, are given accordingly. The experimental results show that the new centralized spectrum allocation strategy can improve the throughput,lower the spectrum switching ratio. Considering the trade-off between different performance measures, a profit function is constructed to optimize the number of slots in a super-slot.