Fig. 8
From: Reinforcement learning-based dynamic band and channel selection in cognitive radio ad-hoc networks
Utilization efficiency reward according to DRE. To realize the mechanism in Fig. 7, the reward for channel utilization efficiency (RUtil) is designed as shown. The x-axis for each band group represents the DRE, and the y-axis represents RUtil. For the band usage maintenance range, RUtil is set to 0. For the band usage change range where the DRE is low, RUtil increases from −1 to 0 since the DRE represents better value as DRE increases. The band usage change range where the DRE is high is divided into [r2, 1) and [1, ∞) to distinguish the insufficient transmission rate provided by the channel. RUtil represents a more rapid decrease rate in [1, ∞). RUtil decreases from −r2 to −1 in [r2, 1) range and from −1 to ∞ in [1, ∞) range