Hardware alarms reduction in Radio Base Stations by forecasting Power Supply Units headroom

Power Supply Units (PSUs) are the components that supply Radio Base Stations (RBSs) with electric power, making it an essential unit for mobile communication networks. A power supply malfunction is critical and must be resolved, but immediate troubleshooting could result in additional operational costs. Based on the prediction of PSU usage by an advanced generalized additive model for time series forecasting called Prophet, the PSU power headroom is derived and the maintenance schedule is proposed, which depends on the severity of demand for power. The safety criteria are derived from the critical operational conditions for several consecutive worst-case scenarios. A PSU alarm will only be generated under the most critical circumstances for an RBS, having multiple PSUs, and sent to the Network Operating Center (NOC). This approach not only allows us to better schedule maintenance and keep operational continuity but also to improve the overall energy efficiency and costs. •Proposing a method for alarm handling in power system for Radio Base Stations.•Generating alarms based on a new triggering logic in the severity of the power headroom variation.•Deriving power headroom based on Machine Learning prediction of PSU usage.•Improving the maintenance schedule and overall energy efficiency.