Real-time issues of predictive modeling for industrial cognitive radios

Cognitive radios (CR) can sense and detect temporarily available spectral holes for an opportunistic operation to improve the spectral efficiency and coexistence of industrial radio systems. It will be of particular interest for a CR system to apply predictive modeling in order to forecast the behavior of the coexisting environment. A secondary cognitive user shall use preemptive tuning of its operating parameters following the predictive model. However, a considerable challenge is to generate an accurate model and predict efficiently in order to meet strict time related requirements of industrial applications. Such predictive modeling has already gained some attention but real-time results are never reported. In this contribution we investigate real-time aspects of predictive modeling for its application in industrial CR systems.