A novel access network selection scheme using Q-learning algorithm for cognitive terminal

In a B3G/4G wireless communication system, the users will connect to the network using one of several available radio access technologies. In this paper, we proposed a Q-learning based algorithm for terminals´ independent access network selection with the aim of improving the resource utilization and providing the best quality of service with respect to the wireless environment status, network performance and user´ requirement. In particular, for the first time we introduced the concept of low-carbon as one of the evaluation indicators of wireless communication performance, in order to reduce the power consumption and achieve a balance between quality and consumption. The proposed scheme is based on the concept of cognitive network, which has been proposed recently by the motivation of complexity, heterogeneity and reliability requirements of tomorrow´s network and the cognitive pilot channel used in it. The performance of the access network selection algorithm is shown in the simulation and it can be seen that this algorithm significantly reduced the blockrate and power consumption as well as increased the throughput compared with random accessing approach. In future work, we will continue to research on the effective access network selection algorithm and try to introduce the low-carbon indicator to other aspects of the wireless communication system.