Metacognition and the next generation of cognitive radio engines

Much of the previous research on cognitive radio has focused on developing algorithms based on artificial neural networks, the genetic algorithm, and reinforcement learning, each with its pros and cons. In this research, we present a new approach based on metacognition. We believe that the metacognitive framework can be the foundation for the next generation of CRs and further the performance improvements in CR. In this work, we present the elements involved in metacognitive radio, discuss the challenges in their development, present solutions to the challenges along with a possible meta-CR architecture, and show results from our implementation. Each cognitive engine (CE) algorithm has strengths and limitations that make it more suitable for certain operating scenarios (channel conditions, operating objective, available hardware, etc.) than other algorithms. A meta-CE can adapt faster and improve performance by exploiting the characteristics and expected performance of the individual CE algorithms. It understands the operational scenarios and utilizes the most appropriate algorithm for the current operational scenario by switching between the algorithms or adjusting them as necessary.