Device to device (D2D) communication, as a proximity communication technique that leverages the spatial–temporal locality of mobile data usage to achieve one to many simultaneous transmission, provides an effective solution for unloading heavy traffic. However, due to the sharing of spectrum resources, interference management has become a major challenge for D2D communication. In this paper, a game-based robust power allocation strategy with quality of service (QoS) guarantee is designed for D2D multicast network. A Stackelberg game is proposed to characterize the actions of the base station (BS) and D2D users (DUEs). In the game, the BS is seen as the leader, whereas the D2D users are followers. On the one hand, the purpose of BS is to maximize the leader’s profit with the constraint of the maximum tolerable interference. On the other, the DUEs attempt to maximize the followers’ profit in the network system while guaranteeing the QoS requirements for DUEs, which are modeled as probability constraints. Spectrum sharing is assumed among the cellular users (CUEs) and DUEs. Because of the dynamic characteristics of wireless channel, it is difficult and expensive to acquire accurate channel state information (CSI). On this account, the uncertain channel state information is considered in the formulated problem, and the probability threshold method is utilized to convert the uncertain constraints into tractable ones. Furthermore, a distributed algorithm to determine the optimal solutions is proposed, which significantly reduces the information exchanges. Finally, numerical results validate the performances of the proposed scheme in the aspects of convergence, utility of uses, and power consumption.

设备到设备（D2D）通信，作为一种利用移动数据使用的时空局部性来实现一对多同时传输的邻近通信技术，为卸载繁重的流量提供了有效的解决方案。但是，由于频谱资源的共享，干扰管理已成为D2D通信的主要挑战。本文针对D2D多播网络设计了一种基于游戏的，具有服务质量（QoS）保证的鲁棒功率分配策略。提出了Stackelberg游戏来表征基站（BS）和D2D用户（DUE）的行为。在游戏中，BS被视为领导者，而D2D用户则是跟随者。一方面，BS的目的是在最大可容忍干扰的约束下最大化领导者的利益。在另一，DUE尝试在确保DUE的QoS要求（被建模为概率约束）的同时最大化网络系统中的追随者的利润。假定在蜂窝用户（CUE）和DUE之间进行频谱共享。由于无线信道的动态特性，获取准确的信道状态信息（CSI）既困难又昂贵。因此，在所提出的问题中考虑了不确定的信道状态信息，并且利用概率阈值方法将不确定的约束转换为易于处理的约束。此外，提出了一种确定最优解的分布式算法，该算法大大减少了信息交换。最后，数值结果从收敛性，使用效用，