基于隐马尔可夫模型的协作频谱检测策略

摘要

Spectrum sensing is the foundation and key technology of cognitive radios.Hidden Markov Model (HMM) was used to model the spectrum sensing problem, and a cooperative spectrum sensing strategy based on HMM was proposed.First of all, the strategy estimated the HMM system parameters by utilizing a maximum likelihood parameter estimation method called Baum-Welch algorithm.Secondly,the posterior probability of the channel state in the current slot was updated by Bayesian inference, which was based on HMM and deduced from the sensing results of secondary users and the posterior probability in previous slots.Finally, the channel state in the current slot was estimated according to the maximum a posteriori criterion.Furthermore, the system's sensing performance was evaluated through utilizing posterior probability, and then an approach to select the fewer secondary users with better sensing ability was presented so as to satisfy the performance requirements and reduce the overhead and complexity for cooperative sensing.Simulation results showed that the proposed strategy outperforms those strategies such as K-out-of-N, maximum likelihood ratio and Chair-Varshney rule.%频谱检测是认知无线电的基础和关键技术,将其建模为隐马尔可夫模型(hidden Markov model,HMM),并由此提出基于隐马尔可夫模型的协作频谱检测策略.该策略首先使用Baum-Welch法对HMM的系统参数进行最大似然估计;然后基于HMM模型,利用各次用户的检测信息以及过去信道状态的后验概率信息进行贝叶斯推理,更新当前时隙信道状态的后验概率;最后根据最大后验概率准则对当前时隙的信道状态进行最终判决.使用后验概率,该策略可进一步估计系统协作检测的性能,在满足系统协作检测性能要求的前提下,选择尽可能少的、检测性能较优的次用户来参与协作,以节约开销和降低复杂度.仿真实验表明,所提出的策略的系统检测性能优于基于大数判决、似然比和Chair-Varshney准则的协作频谱检测策略.