Predictive learning model in cognitive radio using reinforcement learning

There is a sweeping change in the technology of wireless devices and application that tends to increase the demand of wireless spectrum. Conventional fixed spectrum allocation method is unable to satisfy the huge amount of spectrum needs resulting in bottleneck condition. To defeat from this problem Cognitive Radio is comes into the picture. Cognitive Radio is a wireless smart radio that can detect available idle channels in a wireless spectrum automatically and share these channels to secondary users and also get better radio operating behavior without interfering with primary users. In this paper we are studying a reinforcement learning technique to predict the throughput using the CACLA algorithm which is modified version of actor critic learning. The performance metrics of learning such as execution time, prediction accuracy and prediction error has been observed using MATLAB.