针对单通道的同调制方式、同调制参数的时频重叠信号的盲分离问题，提出了一种新的卷积混合盲分离算法。首先，基于时间分集的过采样技术的信道多样性性质，将单通道混叠的MISO模型转化为多通道混叠的MIMO模型进行研究。然后，结合MIMO模型混叠矩阵的结构特点，讨论了混叠信号分量的时延差对MIMO模型混叠矩阵结构和混叠矩阵秩的影响。最后，数学推导证明：当信号分量时延差为符号速率的整数倍时，从信号时延的角度来讲混叠信号不具有盲可分离性，并通过仿真实例予以验证。

Considering the blind signal separation problem of the convolutive mixture communication signals with same carrier frequency and modulation by a single channel,a new algorithm was presented for convolutive blind signal separation.Firstly,a single channel mixing signal model of MISO was transformed into a multi-channel mixing signal model of MIMO,based on the characteristics of oversampling.Then,the influence of the time-delay differences between mixing signal components to the structure and rank of mixing matrix of MIMO was studied.Finally,mathematical derivation proved that:if the time-delay difference between mixing signal components was the integral multiples of symbol velocity,the mixing signal is not blindly separable.And simulation results showed the validation of the algorithm.