Enhancing 5G Massive MIMO Systems Using a Compressive Sensing-Based Approach

Abstract

The capacity of massive multiple-input multiple-output (MIMO) systems is significantly enhanced through the use of abundant antennas at the base station (BS), supporting a multitude of users. However, in Time Division Duplex (TDD) mode, the multiplexing of pilots inevitably leads to Pilot Contamination (PC). In this study, a novel approach to interference alignment and pilot purification in a large MIMO system was proposed, utilizing the principles of compressive sensing and deep learning. The primary aim of the Compressive Sensing-Based Location Scheduling System (CSLSS) is to mitigate PC issues while optimizing the pilot sequence for the user. By introducing an additional sequence, the pilot set is enhanced using our location-based decontamination technique. Furthermore, a Feed-Forward Convolutional Neural Network (FFCNN) is employed for interference alignment in the MIMO system. Experimental results indicate significant improvements in the parameters under consideration. The sum rate showed an increase of 74.6%, the Bit Error Rate (BER) improved by 57.8%, the Signal-to-Interference-plus-Noise Ratio (SINR) rose by 79.8%, and the spectral efficiency of the channel improved by 98% in terms of interference alignment. In addition, a compressive sensing-based parameter showed an enhancement of 98%. The Mean Square Error (MSE) for CSLSS was reduced by 64.5% for the Signal-to-Noise Ratio (SNR), and the BER for the proposed compressive sensing method improved by 41.2% for SNR. The throughput also improved, with an increase of 97.5% for SNR and 97.8% for the number of antennas used in the BS.