Limited Feedback-Based Unitary Precoder Design in Rician MIMO Channel via Trellis Exploration Algorithm

It is well known that in practice, MIMO channels usually exhibit strong line-of-sight (LoS) components. However, common codebook designs in precoded MIMO rely heavily on the assumption of i.i.d Rayleigh fading. Therefore, designing practical unitary precoders that are suited for realistic channel condition (such as Rician MIMO channel) becomes an interesting topic to investigate. In this paper, we propose to use trellis exploration algorithm to accomplish this objective by adapting each entry of the precoding matrix to instantaneous channel realizations. Since trellis search is an on-line learning approach, our proposed precoding scheme is more adaptive to variations in channel propagations than off-line designed codebooks. Specifically, we use simulation results to show that in Rician MIMO channel, our proposed approach results in better performance than Grassmannian line/space packing and typical DFT codebook design in terms of ergodic capacity, under the same amount of feedback bits.