Abstract:

In order to solve the problem of channel and power allocation in distributed cognitive radio networks (CRN),a case-based reasoning (CBR) and cooperative Q-learning algorithm was proposed.In order to optimize the Q initialization of Q-learning algorithm,the current problem and the historical case were matched according to the similarity function,the Q value of the matching case was extracted and normalized as the initial value.Cooperative Q-learning was based on the total reward value,and each agent integrates the Q values of other agents with higher reward values with different weights to gain learning experience to reduce unnecessary exploration.Simulations show that the proposed algorithm can improve the energy efficiency of the cognitive system’s channel and power allocation,and accelerate the convergence speed of the system.

Key words: cognitive radio, cooperative Q-learning, case-based reasoning, channel and power allocation, energy efficiency, convergence speed