An Iterative Massive-MIMO NOMA Receiver with Superimposed Pilots based Estimator for Sparse Channels

Abstract

The application of time division duplex (TDD) operation in massive multiple-input multiple-output (MIMO) systems make the resources required for channel estimation independent of the number of antennas at the base station (BS), while its dependence on the number of users causes pilot contamination. The more pilot slots available in superimposed pilot (SIP) arrangement compared to the conventional time-multiplexed pilot (TMP) arrangement allows higher amount of generatable orthogonal pilots; therefore, SIP arrangement have recently gained significant attentions for pilot decontamination. An SIP based method, exploiting first-order statistics of the received signals, is proposed for the estimation of frequency selective sparse massive-MIMO channels. A multi-cell radio communication scenario is considered, where the data and SIP associated to the users within a certain cell are taken non-orthogonal and orthogonal, respectively. An iterative receiver algorithm is proposed, which eliminates the contribution of information sequences from the received signals prior to estimating the channels and then eliminates the contribution of pilots from the received signal prior to decoding the users’ information sequences. The proposed work focuses on the propagation scenarios which lead to a sparse structured channel impulse response (CIR); while the theory of compressed sensing (CS) is adopted for the reconstruction of CIR. Mathematical expressions to study the behaviour of signal to noise and interference ratio (SINR), observed at different stages of the receiver, are derived. A thorough analysis on the effect of training-to-information power ratio (TIR), number of BS antennas, and channels’ sparsity level on bit error rate (BER), SINR, user sum-rate, normalized channel mean square error (NCMSE) performance of the proposed method is conducted. A performance comparison of the proposed method with notable methods from the classes of TMP and SIP arrangements is conducted. In a scenario, when the BSs antennas are taken as 128, channel sparsity set at 4/14, and studied after 5 iterations, the proposed SIP based method outperforms the existing TMP based method for the values of TIR in the range [0.112, 0.419]; while the optimal value of TIR is observed as 0.24.