Reliability analysis of energy management subsystem in a high altitude platform station using hierarchical stochastic models and machine learning

The increasing need for faster data rates and better performance has prompted wireless network providers to consider energy management in High Altitude Platform Station (HAPS) over Internet of Things (IoT) enabled devices in 5G New Radio (5G NR) [22]. Energy management is a basic requirement for solar-powered HAPS, which has not been paid attention in recent times. This HAPS architecture suffers from limited and unreliable energy management resources. Although there are many competent technologies working for the purpose of capacity enhancement and reliability of the Energy Management Subsystem (EMS), such as optimization of propulsion energy and energy optimization technologies, there is still room and need for further improvement. In this article, we introduce a three-level hierarchical model to evaluate the reliability characteristics of EMS in HAPS. Two different methods, continuous-time Markov chains (CTMCs) and reliability block diagrams (RBDs), are used to model this hierarchical model of EMS in HAPS. CTMC-based modeling is used to evaluate performance metrics, such as the reliability function and Mean Time To Failure (MTTF), at two different levels by using $ k $-out-of-$ n $ stochastic modeling technique. Further, the proposed $ k $-out-of-$ n $ models for repairable and non-repairable batteries and solar cells are subject to an analytical investigation, and their validity is confirmed using discrete event simulation. The precision in estimating reliability and temporal attributes for diverse input data, employing important parameters, is attained through the widely recognized ML models such as K-nearest neighbors (KNN) regression, polynomial regression, etc.