Continuous Incentive Mechanism for D2D Content sharing: A Deep Reinforcement Learning Approach

Device-to-device (D2D) communication-based content sharing is regarded as a promising way to offload traffic from cellular networks, where incentive mechanisms are required to motivate mobile user equipment (UE) to participate in content sharing. In this paper, we firstly propose an improved scoring mechanism to provide continuous incentive and then study the impact of historical behavior on continuous motivation. Furthermore, to maintain continuous motivation while keeping the service quality of content sharing, we investigate the weights setting of historical behavior and current status in scores calculating, which is formulated as a stochastic dynamic programming (SDP) problem due to the long-term performance and the randomness of the network. To tackle the curse of dimensionality, a deep reinforcement learning (DRL) algorithm is employed for optimization. Simulation results show that with DRL, the mechanism is effective in motivating content-sharing continuously, improving the quality of service (QoS), and cutting down the sharing cost as well.