Knowledge management framework for robust cognitive radio operation in non-stationary environments

To increase cognitive radio operation efficiency, this paper proposes a new knowledge management functional architecture, based on the fittingness factor concept, for supporting spectrum management in non-stationary environments. It includes a reliability tester module that detects, based on hypothesis testing, relevant changes in suitability levels of spectrum resources to support a set of heterogeneous applications. These changes are captured through a set of advanced statistics stored in a knowledge database and exploited by a proactive spectrum management strategy to assist both spectrum selection and spectrum mobility functionalities. The results reveal that the proposed reliability tester is able to disregard the changes due to the intrinsic randomness of the radio environment and to efficiently detect actual changes in interference conditions of spectrum pools. Thanks to this support, the proposed spectrum management strategy exhibits substantial robustness when the environment becomes non-stationary, obtaining performance improvements of up to 75% with respect to the reference case that does not make use of the reliability tester functionality.