<p>Backscatter communication is a new low-power communication technique to exploit reflected or backscattered signals to transmit data, where backscattered signals can be the reflection of ambient radio frequency (RF) signals. The major challenge for backscatter communications is the time scheduling of different operation modes. Driven by this observation, this paper studies a time scheduling problem by jointly optimizing energy harvest, backscatter and active communications. Based on the <i>normalized constant elasticity of substitution bargaining solution</i> (<i>NCESBS</i>), we propose a novel time scheduling scheme for the <i>ambient backscatter communication</i> (AmBC). In our proposed scheme, the communication time period is divided into two phases through the idea of <i>NCESBS</i>, and each time phase is scheduled for devices by using a special case of <i>NCESBS</i>. To effectively handle the current AmBC system, the <i>frequency adjusted Q-learning</i> (<i>FAQ</i>-learning) is adopted to adjust the control parameter of <i>NCESBS</i>. The main novelty of this study is our collaborative control paradigm to reach reciprocal advantages. Through the cooperation of AmBC system agents, three cooperative game models are developed; they are jointly combined, and work together to effectively share the time period. The effectiveness of our proposed method supported by simulation results confirms its superiority compared with existing state-of-the-art AmBC control methods. Especially, we increase the system throughput, normalized payoff and fairness of IoT devices by about 10%, 10% and 15%, respectively. Finally, concluding remarks are discussed, and the findings, which are significant for the future work, have been presented.</p>