A Joint Combiner and Bit Allocation Design for Massive MIMO Using Genetic Algorithm

I. Zakir Ahmed,Hamid Sadjadpour,Shahram Yousefi
DOI: https://doi.org/10.48550/arXiv.1711.06706
2017-11-18
Abstract:In this paper, we derive a closed-form expression for the combiner of a multiple-input-multiple-output (MIMO) receiver equipped with a minimum-mean-square-error (MMSE) estimator. We propose using variable-bit-resolution analog-to- digital converters (ADC) across radio frequency (RF) paths. The combiner designed is a function of the quantization errors across each RF path. Using very low bit resolution ADCs (1-2bits) is a popular approach with massive MIMO receiver architectures to mitigate large power demands. We show that for certain channel conditions, adopting unequal bit resolution ADCs (e.g., between 1 and 4 bits) on different RF chains, along with the proposed combiner, improves the performance of the MIMO receiver in the Mean Squared Error (MSE) sense. The variable-bit-resolution ADCs is still within the power constraint of using equal bit resolution ADCs on all paths (e.g., 2-bits). We propose a genetic algorithm in conjunction with the derived combiner to arrive at an optimal ADC bit allocation framework with significant reduction in computational complexity.
Signal Processing,Information Theory
What problem does this paper attempt to address?