Performance Analysis of Mixed-ADC MIMO Systems Over Rayleigh Channels Based on Random Matrix Theory

The mixed analog-to-digital converters (ADC) architecture is a promising solution to the problem of high energy consumption of multiple-input multiple-output (MIMO) systems. In such a scheme, part of the received signals at the base station are quantized by high-resolution ADCs, while the others are quantized by one-bit ADCs. The signals quantized by the mixed-ADCs architecture are correlated and non-identically distributed, which is different from the one-bit quantized MIMO systems. Moreover, unlike the previous works focusing on the achievable rate of mixed-ADC MIMO systems under linear detectors, we derive a closed-form expression of the ergodic mutual information between the transmit signals and the quantized outputs of the mixed-ADC architecture over Rayleigh channels, where the statistical property of the equivalent channel is characterized by random matrix theory. The results show that in uplink mixed-ADC MIMO systems, reducing the transmit power will not sacrifice the transmission efficiency when the number of receive antennas increases. Furthermore, constrained by a given hardware energy consumption, the theoretical mutual information shows obvious gains over the one of unquantized MIMO systems for low user transmit power.