Intelligent Adaptive Data Transmission Strategy for CRNs

Due to the time-varying characteristics of the wireless channels in cognitive radio networks, how to set the length of data packets for transmission is a challenge. It’s because that the packet length is too large, which may cause link congestion and transmission failure, and the packet length is too small to make full use of the spectrum resources. To solve this problem, we propose an adaptive transmission strategy based on Multi Armed Bandit. We assume the packet length as the arm, and the transmitter as the player. The goal of our strategy is to improve data transmission efficiency as far as possible by choosing the suitable arm with the change of channel quality. Furthermore, in order to improve the convergence speed, we innovatively propose using the belief factor to balance the exploitation and exploration among arms, so as to avoid falling into local optimum. The experiments show that our strategy can select the optimal arm when the channel quality changes dynamically. It means that the packet length is intelligent adaptive in varying environment of cognitive radio networks. Compared with other traditional algorithms, our strategy has better throughput.