US 12745109 B2 / Sep 22, 2026

Neural network based signal-to-noise ratio prediction in mobile networks

Predicting link outages in a mobile network. A method includes obtaining a first array of time series data comprising magnitude values of a first communication property beginning at a first time. A second array of time series data comprising angle values of a second communication property beginning at the first time, correlated to the first array is obtained. The first array is modulated with the second array to obtain a first modulated array. The first modulated array is normalized to obtain a first normalized modulated array. The first normalized modulated array is provided to a trained time-series forecasting model. As a result, one or more predicted SNR values are obtained from the trained time-series forecasting model.