Adaptive Time Window Linear Regression for Outage Prediction in Q/V Band Satellite Systems

Abstract

Accurate prediction of link outage events is of paramount importance in very high throughput satellite systems with feeder links operating in ${Q}/{V}$ frequency bands, where severe propagation impairments (e.g., rain fading) can dramatically hamper data communications. To cope with these performance degradations, this letter proposes a novel feeder link outage event prediction scheme based on time-adaptive linear regression. The conducted performance analysis proves the validity of the proposed concept and its superiority over other candidate solutions available from the current literature (e.g., probabilistic prediction).