Adaptive reinforcement learning based joint approach for energy efficiency in ultra dense networks: ARJUN model

The rapid growth of cellular network users poses a significant challenge for providers to maintain Quality of Service and reduce network power consumption. The proposed work presents two power saving strategies, namely BS sleeping and Cell zooming, which are executed based on predictive evaluation of network traffic. The proposed work integrates two distinct cell zooming methodologies i.e. centralized and distributed cell zooming, tailored for various traffic load scenarios. The cell zooming technique is implemented using Reinforcement learning that helps network to learn from the environment and take decisions based on traffic conditions. In the proposed distributed cell zooming the traffic load is balanced using Spider Monkey Optimization algorithm. The proposed system provides the 37% power saving when compared with the always on condition of network. When compared with existing techniques BS utilization is reduced by 3%, spectral efficiency is increased by 1% and Energy Efficiency is increased by 4% with proposed system.