中文摘要：

认知无线电频谱分配中为了追求绝对公平而采用随机分配策略,然而由于次用户业务服务质量要求不同,会导致用户中断概率增加.在主用户行为预测模型中应用二状态马尔科夫链,主用户出现概率随机产生,不能反映信道真实状态.为此,提出时频联合检测与实时监测相结合的主用户行为预测,在频谱分配中区分业务类型.仿真结果显示,所提算法次用户平均中断概率优于随机分配机制.

英文摘要：

In pursuit of absolute justice, random allocation policy is adopted in spectrum decision of cognitive radio, but because of the different requirements of service quality which will lead to the increase of user interruptions. The primary user behavior predic- tion model uses 2 state Markov chain, as the primary users generated at random, it can not reflect the real state of the channel. For this, the prediction that time-frequency joint detection and real-time monitoring of the main user combined to distinguish the type of business in the spectrum decision. The result of simulation shows that the average user interrupt probability of the proposed algo- rithm is better than random allocation policy.