Analytical Investigation of Self-Organizing Network Algorithms for Automatic Neighbor Relation Optimization in Nigerian LTE Networks

Self-organizing network technology enables automated optimization of mobile network configuration parameters, reducing the operational burden on network engineers while improving coverage and capacity performance through adaptive parameter management. Automatic neighbor relation optimization, a core SON function, maintains accurate neighbor cell lists that are critical for handover success rates. This study analytically investigates SON algorithms for automatic neighbor relation optimization in Nigerian LTE network contexts, where frequent topology changes from new site additions and informal network expansion create rapidly outdating neighbor relations. An analytical study combining mathematical modeling of the ANR decision logic with event-driven simulation using LTE network topology data representing MTN's network configuration in Anambra and Imo States was conducted under a research collaboration agreement. The simulation environment modeled 230 eNodeBs with realistic handover event generation and missing neighbor relation scenarios. Three ANR algorithm variants were compared: standard 3GPP Release 9 ANR, a modified penalty-based ANR, and a reinforcement learning-enhanced ANR implementation. Results demonstrate that the reinforcement learning-enhanced ANR reduced missing neighbor relation-induced handover failures by 38% compared to the standard 3GPP algorithm, with convergence achieved within 72 hours of deployment. The modified penalty-based ANR achieved intermediate performance at 24% improvement over standard ANR with lower computational overhead. The study provides locally relevant algorithm performance benchmarks for SON deployment planning in south-eastern Nigerian LTE networks, contributing to the limited body of SON evaluation literature for West African network environments. Keywords: self-organizing network, ANR, LTE, handover optimization, Nigeria.