Reduced Space and Time-Delay Channel State Representation for 5G FDD Massive MIMO Systems

Channel State Information (CSI) feedback is a very relevant topic for Frequency Division Duplexing (FDD) Multiple Input/ Multiple output (MIMO) on 4G/LTE systems and it will be also the case for the development of future Massive MIMO 5G standards. In the following article, we present a novel CSI compression technique based on the potential of Karhunen-Lo`eve Transform (KLT) to reduce the spatial domain representation by means of a description based on a number of principal eigenvector projections, which is combined with the channel sparsity in time-delay after exploiting frequency domain correlation. The reduced spatial domain information obtained from KLT can be further compressed by using a truncated Fourier component representation of the principal Eigendirections obtained. The algorithm is compared with respect to other CSI description schemes including modern standards like 3GPP Rel 14Adv CSI for FDD. Results show the feasibility of implementation and the benefits of this technique for the future 5G standards.