Enhancing robustness and perturbation tolerance of cognitive radio engines with metacognition

A cognitive radio engine (CE) is an intelligent agent which observes the radio environment and chooses the best communication settings that best meet the application's goal. In this process, providing reliable performance is one of the major tasks in designing CEs for wireless communication systems. The main purpose of this work is providing predictable performance and controlling the cost of intelligent algorithms based on the CE's experience and complexity analysis respectively. In this work, we extend our meta-CE design to control the cost of computations and provide more reliable performance for providing the minimum requirement of the radio applications in different scenarios. To achieve this, we use robust training algorithm (RoTA) in two different levels alongside of the individual CE algorithms. The RoTA, enables radio to guarantee some minimum output performance based on the learning stages. RoTA uses confidence interval approximation for standard normal distribution to calculate the lower and upper bounds of CE's expected performance to analyze the reliability of decisions. Moreover, in the case of non-stationary environments, RoTA is facilitated by forgetfulness factor to provide minimum performance guarantees. The second level of RoTA operates in meta-level to control the amount of computation complexity of intelligent algorithms in all levels with respect to the obtained performance and complexity analysis.