Learning Optimal Fronthauling and Decentralized Edge Computation in Fog Radio Access Networks

Fog radio access networks (F-RANs), which consist of a cloud and multiple edge nodes (ENs) connected via fronthaul links, have been regarded as promising network architectures. The F-RAN entails joint optimization computing well interactions, is challenging for traditional techniques. This paper proposes Cloud-Enabled Cooperation-Inspired Learning (CECIL) framework, structural deep learning mechanism handling generic problem. proposed solution mimics cloud-aided cooperative policies by including centralized at the cloud, distributed decision ENs, their uplink-downlink interactions. A group neural (DNNs) are employed characterizing computations ENs. forwardpass DNNs carefully designed such that impacts practical channel noise signling overheads, can be included in training step. As result, operations ENs jointly trained an end-to-end manner, whereas real-time inferences carried out decentralized manner means coordination. To facilitate cooperation among optimal schemes designed. Training algorithms robust to impairments also presented. Numerical results validate effectiveness approaches.