In this paper, we propose an energy efficient passive optical network (PON) architecture for backhaul connectivity in indoor visible light communication (VLC) systems. The proposed network is used to support a fog computing architecture designed to allow users with processing demands to access dedicated fog nodes and idle processing resources in other user devices (UDs) within the same building. The fog resources within a building complement fog nodes at the access and metro networks and the central cloud data center. A mixed integer linear programming (MILP) model is developed to minimize the total power consumption associated with serving demands over the proposed architecture. A scenario that considers applications with intensive demands is examined to evaluate the energy efficiency of the proposed architecture. A comparison is conducted between allocating the demands in the fog nodes and serving the demands in the conventional cloud data center. Additionally, the proposed architecture is compared with an architecture based on state-of-art Spine-and-Leaf (SL) connectivity. Relative to the SL architecture and serving all the demands in the cloud, the adoption of the PON-based architecture achieves 84% and 86% reductions, respectively.

翻译：本文提出了一种用于室内可见光通信（VLC）系统回传连接的节能无源光网络（PON）架构。所提出的网络用于支持一种雾计算架构，该架构允许有处理需求的用户能够访问专用雾节点以及同一建筑内其他用户设备（UD）中的空闲处理资源。建筑内的雾资源是对接入网和城域网中的雾节点以及中央云数据中心资源的补充。我们开发了一个混合整数线性规划（MILP）模型，以最小化在该架构中服务需求所消耗的总功率。通过考察一个包含高需求应用的场景，评估了所提架构的能效。我们将需求在雾节点中进行分配的情况与在传统云数据中心中服务需求的情况进行了比较。此外，我们所提架构还与一种基于最新脊叶（Spine-and-Leaf, SL）连接架构进行了比较。相对于SL架构以及将所有需求在云中服务的情况，采用基于PON的架构分别实现了84%和86%的能耗降低。