A Deep Reinforcement Learning-Based Contract Incentive Mechanism for Mobile Crowdsourcing Networks

Mobile crowdsourcing network (MCN) is a promising paradigm to leverage mobile users (MUs) to provide sensing services by data collection. However, owing to limited sensing-computation resource and data security risk, MUs may be unwilling to complete sensing tasks without sufficient incentive. Therefore, a deep reinforcement learning (DRL)-based contract incentive mechanism is proposed by jointly considering participation contribution and sensing-computation cost of MUs. Specifically, considering the heterogeneous sensing and computation abilities of MUs, we formulate a task-reward contract incentive mechanism to achieve the maximum utility of mobile crowdsourcing platform. Moreover, based on the individual rationality and incentive compatibility constraints, we design the optimal contract under the partial information asymmetry scenario. Furthermore, considering the network dynamics and privacy protection, we formulate the contract incentive problem as an Markov decision process. Due to the high-dimensional continuous action and state spaces, we develop the twin-delayed deep deterministic method to obtain the efficient sensing task and incentive reward policy under complete information asymmetry scenario. Simulation results demonstrate the feasibility of the proposed DRL-based contract incentive mechanism.