A Fuzzy Neural Network based reasoning and learning approach for efficient spectrum management in cognitive radio

Cognitive radio (CR) is a software defined radio with artificial intelligence (AI) i.e. it can learn from and adapt to ambient radio environment. Most of the research in the field of CR has been centered around policy-based systems that are hard-coded with certain rules for reasoning and learning capabilities for very specific applications. In CR networks, multiple interacting capabilities are required for practical implementation of spectrum management. This paper discusses a Fuzzy Neural Network (FNN) based reasoning and learning approach for efficient spectrum management abilities in CR networks. Neural network in the feedback configuration is utilized to incorporate the results of the learning engine into a fuzzy based reasoning engine so that radios can remember lessons learned in the past and act quickly in the future for efficient spectrum management.