Page 88 - AI for Good-Innovate for Impact
P. 88
AI for Good-Innovate for Impact
Use case –18: AI enabled Power efficient Macro Radio Unit
Country: India
Organization: HFCL Limited
Contact person: Sonali, sonali@ hfcl .com
18�1� Use case summary table
Domain Energy Efficient
Problem to be Inefficient resource allocation and energy wastage in telecommunica-
addressed tions infrastructure.
AI based deactivation of RU.
Key solution
(granularity at RF Chain level and at frame level)
Technology Resource allocation, telecommunications, 5G NR, Radio Unit, 6G,
keywords Energy Efficiency.
Data availability Open traffic data (Traffic Data | Mapbox)
Metadata (type of Time series data on network activity (cpu usage), radio unit occupancy
data) (cpu utilization at BS)
Based on the traffic density on working days
Model Training and
fine tuning vs weekends and also between the different time slots of a day, i.e.,
office hours (peak hours) vs late evening time.
Experimentation. With optimized design, energy efficiency can
Testbeds or pilot increase by 50 % for heavy network load conditions and the number
deployments
can scale upto 80 % with low load conditions.
Metrics, KPIs, CPU utilization, energy spent in tackling a particular load amount and
measurements Radio Unit utilization factor.
Role of Trainings, Integration to RU products, and standards.
standards
Open Source Commercial Utilization
18�2� Use case description
18�2�1� Description
The use case involves the deployment of 5G base stations and Customer Premises Equipment
(CPE), focusing on assessing network efficiency and resource allocation through key performance
indicators (KPIs) like CPU utilization. CPU utilization of the base station serves as an indicator
of its operational workload. It tells us the amount of data sent which then tells us the CPU
72
