Network Energy Saving Techniques Aided by AI/ML in 4G/5G Networks

28/11/2024

This thesis aims to study the aspects and characteristics of 4G and 5G mobile networks, their energy consumption and potential optimization using energy efficiency techniques with the aid of machine learning. A traffic dataset was provided by Vodafone, which was carefully analyzed to identify potential energy-saving opportunities that could be simulated. A time-based energy efficiency technique is developed aimed at turning off components of the base station during periods of low or no traffic. To predict these periods, an algorithm was developed, leveraging traffic predictions from an LSTM model to guide energy-saving actions. The algorithm is designed to deliver substantial energy efficiency while also being customizable to meet the specific requirements of network operators and remaining easy to implement and maintain. The LSTM model effectively predicts future sector usage by accurately learning and adapting to daily network usage patterns, significantly enhancing the effectiveness of energy-saving actions performed by the algorithm. The proposed algorithm is able to achieve high predictive accuracy, whose outcomes demonstrate substantial results, achieving up to 2,451,624 kWh of saved energy, translating to total monetary savings of €403,047 and total CO2 emissions saved of 405.7 tons in a year of simulated use across the whole 5G network of the country in study.

Autores da comunidade :

• 

  A

  André Hilário Cunha

  ist196156

Orientadores desta instituição:

• 

  L

  Luís Manuel De Jesus Sousa Correia

  ist12076

• 

  A

  António Manuel Raminhos Cordeiro Grilo

  ist14017

electrical-engineering-electronic-engineering-information-engineering - Engenharia Eletrotécnica, Eletrónica e Informática

eng - Inglês

Instituto Superior Técnico