Scaling Analysis for Massive MIMO Systems With Hardware Impairments in Rician Fading

Abstract

In this paper, we propose a framework for rate analysis in the multicell massive multiple-input multiple-output (MIMO) system which models the channel within one cell as Rician fading and considers different kinds of hardware impairments. Closed-form achievable uplink rates with common and separate oscillators are derived, respectively. Based on them, we find that line-of-sight (LoS) components can benefit the rate performance under any configurations, but this improvement decays as the phase drift aggravates. Utilizing the excessive degrees-of-freedom offered by massive MIMO, we can tolerate the hardware constraints without rate reduction. In particular, as the number of base station (BS) antennas (M) grows, we can relax the hardware constraint on additive noise by M in LoS-Rayleigh mixed channel, which is different from the relaxation by √M in pure Rayleigh channel, whereas the hardware constraint on phase drift can only be relaxed logarithmically with M. Furthermore, we also utilize the benefit of LoS components to reduce the required antenna size and transmit powers. Via simulation, we find that with sufficient LoS propagations, both the BS antenna and transmit powers can be cut down remarkably while maintaining a non-reduction rate performance as the Rayleigh fading environment, which is useful for the ultra-dense network in future.