Theoretical Foundations and Experimental Validation of Compressive Sensing-Based Channel Estimation for Massive MIMO in 5G Networks in West Africa

Massive MIMO channel estimation overhead scales proportionally with antenna count using conventional pilot-based methods, creating a fundamental training overhead bottleneck that limits the spectral efficiency gains achievable in practical systems with finite coherence intervals. Compressive sensing exploits the sparse structure of wireless channels in angular and delay domains to enable accurate channel reconstruction from significantly reduced pilot measurements. This doctoral study develops original theoretical foundations and experimental validation for compressive sensing-based channel estimation in massive MIMO systems targeting 5G deployment in West African propagation environments. A novel sparse channel representation basis is derived analytically for the angular-delay domain channel structure observed in Nigerian urban environments, based on directional channel measurement campaigns conducted in Abuja and Lagos at the 3.5 GHz 5G pioneer band. Theoretical recovery conditions for the proposed compressive channel estimator are derived using restricted isometry property analysis adapted for the proposed basis. An original mixed-norm convex optimization formulation that jointly exploits angular and temporal sparsity achieves minimum pilot overhead while meeting the theoretical recovery guarantee. Experimental validation using a 64-antenna testbed in Abuja demonstrates channel estimation normalized mean square error within 1.8 dB of the Cramer-Rao lower bound using 34% fewer pilot symbols than conventional MMSE estimation. The doctoral contribution advances compressive sensing theory for wireless channels by deriving the first recovery conditions analytically grounded in empirically characterized West African massive MIMO channel sparsity structure. Keywords: compressive sensing, massive MIMO, channel estimation, 5G, sparsity.