DefinePK

Abstract

The escalating demand for high-capacity, low-latency wireless communication in densely populated environments necessitates robust and adaptive mobility management techniques. Traditional handover mechanisms in cellular networks, typically governed by static signal threshold rules, often fail to account for the dynamic nature of user mobility and heterogeneous network conditions, leading to suboptimal performance outcomes. This study presents a simulation-based analysis of a machine learning (ML)-assisted handover decision model employing logistic regression for real-time optimization of user equipment (UE) association in a densely deployed cellular environment.
A comprehensive MATLAB-based framework is developed to emulate UE mobility, dynamic BS distribution, SINR computation, power control, and network interference. The ML model is trained using synthetic features derived from current and candidate base station distances and SINR gain metrics to predict handover decisions adaptively. Performance evaluation over multiple simulation intervals highlights improvements in handover accuracy, reduction in packet loss during transitions, and enhancement in throughput and latency distributions. Spatial analysis via heatmaps further elucidates network behavior under varying UE density and mobility patterns. The results substantiate the efficacy of ML integration in mobility management protocols, offering a scalable approach to next-generation wireless network optimization.