Massive MIMO is Very Useful for Pilot-Free Uplink Communications

Pilot-free (or non-coherent) communications are of great significance for short-packet communications in beyond 5G applications. The absence of pilots brings a fundamental challenge when decoding a message because of no knowledge of channel state information at a receiver. In this paper, we aim to show that a massive number of antennas at a base station (BS) is very useful for pilot-free uplink communications. Specifically, for a single-input multiple-output channel, we show that the spectral efficiency linearly scales with channel coherence time, provided that the number of antennas is infinite. We prove this result by both a variant of binary sparse superposition codes and a compressive covariance sensing-based decoding method. We also present a novel covariance matching pursuit (CMP) decoding method that is computationally efficient yet achieving a nearoptimal decoding performance. By simulations, we demonstrate the proposed decoding algorithm significantly outperforms the existing approximated message-passing based algorithm.