Optimising Open Loop Power Control in Ultra-Dense Wireless Network Using Bayesian Learning Approach

The evolving growth of the telecommunications sector has intensified the need for efficient power control strategies, particularly in Ultra-Dense Wireless Networks (UDNs), which are integral to 5G and beyond. The networks encounter substantial obstacles because they demand increased data speeds, reduced delays, and reduced interference impacts. Control methods based on traditional models, supervised learning, and reinforcement learning approaches do not perform well under network dynamics in dense areas because they demonstrate complex structures and slow convergence rates. This research adopts the Bayesian learning approach to optimise open-loop power control by processing wireless network uncertainties, resulting in better power allocation effectiveness. The research implements an advanced predictor from the Gaussian process model, which updates prior information from observed data and gathered knowledge during each cycle to improve prediction precision. The iterative process helps decision systems optimise performance during varying network scenarios using newly acquired data. The approach delivers an improved network efficiency of 25% and a signal quality enhancement of 15% when applied versus traditional power control methods through simulation testing experiments.