Online Caching with Wireless Fronthauling and Delivery in Fog-Aided Networks

TL;DR

This paper investigates online caching strategies in fog radio access networks with wireless fronthauling, focusing on arbitrary ENs and time-variant content popularity, revealing limits set by fronthaul power scaling on delivery performance.

Contribution

It extends existing models to arbitrary ENs and dynamic content popularity, providing analytical insights into the impact of fronthaul power scaling on performance.

Findings

Fronthaul power scaling limits delivery performance.

Analytical results for arbitrary ENs with time-variant popularity.

Proactive online caching strategies analyzed.

Abstract

Fog Radio Access Network (F-RAN) exploits cached contents at edge nodes (ENs) and fronthaul connection to the cloud for content delivery. Assuming dedicated fronthaul links between cloud and each EN, previous works focused on analyses of F-RANs using offline or online caching depending whether the content popularity is time-invariant or time-variant. Extension has been done for multicast fronthaul link connecting cloud to only two ENs and time-invariant popularity. In contrast, the scope of this work is on the case where multicast fronthaul link connects arbitrary number of ENs to the cloud and content popularity is time-variant. Normalized Delivery Time (NDT) is used as a performance measure and by investigating proactive online caching, analytical results reveal that the power scaling of fronthaul transmission sets a limit on the performance of F-RAN.