基于BCJR的GMSK混合信号分离算法研究

Research on GMSK Mixing Signals Separation Algorithm Based on BCJR

针对GMSK混合信号单通道盲分离问题,提出将BCJR算法用于GMSK混合信号分离。根据GMSK信号连续相位的特点,通过建立混合信号模型的有限状态马尔科夫性,推导了利用BCJR算法实现GMSK混合信号分离过程。在不增加算法复杂度的情况下,选择每个时刻后验概率的多个极大值对分离算法进行了改进。仿真实验表明,在调制参数比较相近的条件下,即使相对幅度、相对相位、相对时延估计不够准确,改进算法仍能取得较好的分离效果。

The BCJR algorithm was proposed for single-channel blind separation of GMSK mixing signals. Based on the characteristics of continuous phase signals,by establishing finite state Markov property of GMSK mixing signals model,GMSK mixing signals seperation process using BCJR algorithm is deduced. The separation process was further improved by selecting each moment multiple maximum posteriori probability without increasing the algorithm complexity. Simulation experiments show that under similar modulation parameters,the improved algorithm can achieve better performance even when the estimation of amplitude ratio,phase difference and time delay is not accurate enough.