A Subspace Estimation Method Based on Eigenvalue Decomposition for Multi-Target Constant Modulus Algorithm

The multi-target constant modulus algorithm (MT-CMA) can blindly separate multiple co-channel signals captured by an antenna array. It sequentially calculates the unknown weight vectors by executing CMA on a subspace spanned by these vectors, and estimates the weight vectors of all users. Thus, it needs to precisely estimate the subspace by using the calculated weight vectors. The conventional estimation method is to estimate the subspace assuming that the array output signals are orthogonal. However, when the signal-to-noise ratio (SNR) is low, the subspace estimation accuracy is degraded due to the failure of orthogonality. To overcome this problem, the paper proposed a new subspace estimation method for MT-CMA. It estimates the subspace by using eigenvalue decomposition of the received signals' covariance matrix, which is independent of signal orthogonality. Computer simulation results show that the proposed algorithm achieves a 10% improvement of user capacity or 5.0dB improvement in required CNR. Moreover, the BER performance of the proposed algorithm is close to that of MMSE.