Analytical Study of Channel Estimation Techniques for Massive MIMO Systems Under Nigerian Propagation Conditions

Massive multiple-input multiple-output antenna systems are a foundational technology for fifth generation mobile networks, promising significant improvements in spectral efficiency and user capacity through highly directional beam steering. The accuracy of channel state information acquired through channel estimation directly determines the performance gains achievable by massive MIMO precoding and detection algorithms. This study analytically evaluates channel estimation techniques for massive MIMO systems under propagation conditions representative of Nigerian urban and suburban environments, using data from field measurement campaigns conducted in Enugu and Abuja. Channel sounding measurements were collected at 3.5 GHz using a custom-built virtual antenna array measurement system, capturing spatial channel characteristics across 64-element virtual array configurations at twelve outdoor locations per city. Least squares, minimum mean square error, and deep learning-based channel estimators were implemented and evaluated against the measured channel datasets. Results demonstrate that the deep neural network estimator achieves normalized mean square error that is 7.4 dB lower than conventional MMSE estimation under the low-to-moderate signal-to-noise ratio conditions prevalent at cell edges in measured environments. Performance degradation from pilot contamination was more severe in Enugu's compact urban grid environment than in Abuja's wider street geometries. The study contributes locally measured channel model parameters for massive MIMO system design in West African propagation environments, filling a significant gap in the international channel modeling literature. Keywords: massive MIMO, channel estimation, propagation, 5G, Nigeria.