DEVELOPMENT OF A MULTILEVEL RELIABILITY-STATE MONITORING MODEL FOR A BASE STATION OF A WIRELESS TELECOMMUNICATION NETWORK

Abstract

A multilevel reliability-state monitoring model for a base station of a wireless telecommunication network is developed. The purpose of the study is to formalize reliability-state classification when diagnostic data are accumulated sequentially and the duration of monitoring must be coordinated with the required decision confidence. The current technical state is represented by an ordered set of reliability levels described by prior probabilities and probability-density functions of monitored diagnostic indicators. The posterior probability that the state belongs to a particular level is used as the generalized diagnostic measure. The mathematical description combines Bayesian classification, a truncated sequential observation procedure, and a maximum-posterior-probability rule. A decision may be made before the maximum number of observations is reached; if the accumulated information remains inconclusive, the sequence is stopped at a predetermined instant and the state is assigned to the level having the greatest posterior probability. Analytical relations are obtained for the stopping indicator, the minimum sufficient number of measurements, and level-specific decision thresholds. It is shown that, for nondegenerate diagnostic distributions and a finite set of reliability levels, the classification is unique and the monitoring procedure terminates after a finite number of observations. Unlike fixed-sample binary inspection, the proposed approach distinguishes operable, degraded, critical, and failure states while limiting unnecessary measurements. The model can be used for operational monitoring of base stations by availability, failure-rate, thermal, power-supply, radio-path, and transmission-quality indicators and provides a basis for condition-based preventive maintenance.