Spectrum environment learning and prediction in cognitive radio

Using the theory of machine learning to spectrum cognition and management is a necessary requirement of realizing cognitive radio technology. Based on the objective license channel model, two evaluation parameters of packet loss rate and throughput are designed and simulated for analysis in this paper, according with the different service types of cognitive users. Study reveals that channel mean probability of error prediction related with vacancy state probability, which as a result, shows that the spectrum sensing performance of the cognitive user based on machine learning and prediction improves compared with random spectrum sensing both in packet loss rate and throughput.