In this paper, joint resource management for device-to-device (D2D) communication assisted multi-tier fog computing is studied. In the considered system model, each subscribed mobile end user can choose to offload its computation task to either an edge server deployed at the base station via the cellular connection or one nearby third-party fog node via the direct D2D connection. After receiving offloading requests from all end users, the network operator determines the optimal management of the fog computing system, including both computation and communication resource allocations, according to its service agreements with end users, energy cost of edge-server processing and total expense in renting third-party fog nodes. With the objective of maximizing the network management profit, a joint multi-dimensional resource optimization problem, integrating link scheduling, channel assignment and power control, is formulated. An optimal solution algorithm is proposed based on the idea of branch-and-price for addressing this complicated mixed integer nonlinear programming problem. To facilitate the practical implementation in large-scale systems, a suboptimal greedy algorithm with significantly reduced computational complexity is also developed. Simulation results examine the efficiency of the proposed D2D-assisted fog computing framework, and demonstrate the superiority of the proposed resource allocation algorithm over the counterparts.

中文翻译：

---

设备到设备通信辅助雾计算的联合资源分配

在本文中，研究了设备到设备（D2D）通信辅助多层雾计算的联合资源管理。在所考虑的系统模型中，每个订阅的移动终端用户可以选择将其计算任务卸载到通过蜂窝连接部署在基站的边缘服务器或通过直接 D2D 连接的一个附近的第三方雾节点。网络运营商在收到所有终端用户的卸载请求后，根据与终端用户的服务协议、边缘服务器处理的能源成本和总费用，确定雾计算系统的优化管理，包括计算和通信资源分配。租用第三方雾节点。以网络管理利润最大化为目标，联合多维资源优化问题，结合链路调度、信道分配和功率控制，制定。针对这一复杂的混合整数非线性规划问题，提出了一种基于分支定价思想的最优解算法。为了促进大规模系统中的实际实现，还开发了一种显着降低计算复杂度的次优贪婪算法。仿真结果检验了所提出的 D2D 辅助雾计算框架的效率，并证明了所提出的资源分配算法优于同行。为了促进大规模系统中的实际实现，还开发了一种显着降低计算复杂度的次优贪婪算法。仿真结果检验了所提出的 D2D 辅助雾计算框架的效率，并证明了所提出的资源分配算法优于同行。为了促进大规模系统中的实际实现，还开发了一种显着降低计算复杂度的次优贪婪算法。仿真结果检验了所提出的 D2D 辅助雾计算框架的效率，并证明了所提出的资源分配算法优于同行。