Research and optimization of spectral efficiency in heterogeneous 4G/5G networks using massive MIMO technology

DOI: 10.31673/2412-9070.2026.028314

Abstract

The article presents a comprehensive study on methods for increasing spectral efficiency and throughput in modern mobile communications networks, specifically addressing the transition from LTE-Advanced Pro to 5G NR within dense urban environments. The research relevance is driven by the exponential growth of mobile data traffic (up to 45% annually) and the critical scarcity of available frequency resources, necessitating the deployment of advanced spatial multiplexing technologies. The study focuses on radio resource management processes in the Kyivstar heterogeneous network, analyzing a cluster of 145 base stations (comprising 435 sectors) in the Pechersk district of Kyiv, an area characterized by complex "street canyon" propagation conditions and high-rise density. 
The methodology is based on analyzing large datasets (Big Data) gathered from network management systems (OSS Huawei U2020) and validating them through field drive test results using Rohde & Schwarz scanners. Data processing and predictive modeling were performed in the Jupyter Notebook environment using the Python 3.9 programming language. The study utilized specialized libraries such as Pandas and NumPy for statistical analysis, as well as XGBoost machine learning algorithms to forecast network behavior and interference patterns under varying Beamforming parameters.
It was established that the deployment of Massive MIMO 64T64R Active Antenna Systems (AAS) provides a significant advantage over classical 2T2R and 4T4R schemes, yielding a 2.7-fold increase in average spectral efficiency to reach 4.8 bps/Hz. Experimental results demonstrate that adaptive beamforming control, with optimized vertical tilt (D-Tilt) in the range of 6-8 degrees, increases signal levels on high-rise building floors (20+ stories) by 18% while simultaneously minimizing inter-sector interference (overshooting). Furthermore, the study reveals a 270% increase in throughput at the cell edge, ensuring stable Ultra HD (4K) video streaming even during peak load hours.
An important finding regarding energy efficiency indicates that the high sensitivity of the 64-element antenna array allows for a reduction in user terminal transmission power (UE Tx Power) by 4 dB, thereby extending the battery life of subscriber devices. The research also validates the efficacy of Dynamic Spectrum Sharing (DSS) in the 2100 MHz band, showing a 12-15% increase in total segment capacity despite the signaling overhead required for coordinating LTE and NR standards. Statistical analysis confirmed that the usage of high-order modulation (256QAM) reached 35%, achieving a median data transfer rate of 45 Mbps. Practical recommendations for configuring handover thresholds and beam coordination are formulated to minimize mutual interference. The findings can be utilized by mobile operators for planning and optimizing 5G NR networks in large cities and laying the groundwork for future Self-Organizing Networks (SON).

Keywords: 5G NR, Massive MIMO, Beamforming, spectral efficiency, heterogeneous networks, QoS, SINR, OSS statistics.