Energy efficient learning based auction process for cognitive radio systems

This paper proposes a framework for learning in an auction based cognitive radio network using the concept of Bayesian and Q learning. The learning process is used to aid energy efficiency in the system. By using Q learning to learn the bid price, this paper shows that for the learning users, the amount of energy consumed per file sent can be reduced compared to the non-learning users. Furthermore, the paper shows that to overcome the deficiencies of traditional Q learning the exploration process can be biased with Bayesian learning to help the exploration process to converge faster. It is shown that this further reduces the energy consumption and the system delay of the learning users in the system.