Superimposed Training based Estimation of Sparse MIMO Channels for Emerging Wireless Networks

Multiple-input multiple-output (MIMO) systems constitute an important part of todays wireless communication standards and these systems are expected to take a fundamental role in both the access and backhaul sides of the emerging wireless cellular networks. Recently, reported measurement campaigns have established that various outdoor radio propagation environments exhibit sparsely structured channel impulse response (CIR). We propose a novel superimposed training (SiT) based up-link channels' estimation technique for multi-path sparse MIMO communication channels using a matching pursuit (MP) algorithm; the proposed technique is herein named as superimposed matching pursuit (SI-MP). Subsequently, we evaluate the performance of the proposed technique in terms of mean-square error (MSE) and bit-error-rate (BER), and provide its comparison with that of the notable first order statistics based superimposed least squares (SI-LS) estimation. It is established that the proposed SI-MP provides an improvement of about 2dB in the MSE at signal-to-noise ratio (SNR) of 12dB as compared to SI-LS, for channel sparsity level of 21.5%. For BER = 10(-2), the proposed SI-MP compared to SI-LS offers a gain of about 3dB in the SNR. Moreover, our results demonstrate that an increase in the channel sparsity further enhances the performance gain.