一种基于黏性隐马尔可夫模型的多频带频谱感知方法

A multi-band spectrum sensing method based on sticky hidden Markov model

现有多频带频谱感知方法经常利用宽带频谱的稀疏性来实现检测,当频谱占用率较高时具有较差的性能。针对这一问题,提出了一种基于相邻频带状态的多频带频谱感知方法。首先,通过引入黏性因子,建立了多频带状态和观测值的黏性隐马尔可夫模型。接着,详细分析了黏性隐马尔可夫模型中参数的迭代更新方式。最后,通过估计各频段观测值的后验均值实现了多频带频谱感知。仿真结果表明,不管宽带频谱是否具有稀疏性,所提方法的检测性能都优于传统方法,且在虚警概率为0.1、频带平均占用率为50%、平均信噪比为−12 dB时能达到接近0.99的检测概率,比其他方法的检测概率提升了约30%。另外,所提方法的收敛速度快于已有方法,因此具有更低的计算复杂度。

Existing multi-band spectrum sensing methods often use the sparsity of broadband spectrum.However,high spectrum occupancy rate of primary users degrades their performance severely.To address this issue,a novel multi-band spectrum sensing method was proposed by exploiting the state correlation among adjacent frequency bands.Firstly,a sticky hidden Markov model(SHMM)was established by regarding the multi-band states and measured energies as hidden and observed variables.In the SHMM,a sticky factor was introduced to represent the state correlation among adjacent frequency bands.Secondly,iterative expressions for the parameters of SHMM were derived.Finally,multi-band spectrum sensing was implemented by obtaining the posterior mean of observations from every frequency bands.Simulation results show that the proposed method outperforms existing methods,and when the false alarm probability is 0.1,the average frequency band occupancy rate is 50%,and the average signal-to-noise ratio is-12 dB,the detection probability can reach close to 0.99,which is about 30%higher than the detection probability of other methods.In addition,the proposed method had a faster convergence rate than existing methods and therefore has lower computational complexity.