Coded caching gains at low SNR over nakagami fading channels

The promising gains provided by coded caching in ideal settings were thought to vanish in more realistic scenarios. This drawback was due to two main reasons: First, the limitation

on the maximum number of chunks in which a given file can be split. Second, the so-called worst-user bottleneck, which refers to the fact that the transmission is limited by the user with the worst channel quality at each time, and which is exacerbated at low SNR. Recent works have shown that the first problem can be exploited to reduce the worst-user bottleneck through the use of shared caches and by dropping the use of XORbased

transmissions. This work analyzes for first time how this promising result impacts the delivery time of coded caching at low SNR. Motivated by the fact that the delivery time under Rayleigh fading does not have an expectation, Nakagami-m fading is assumed to model the physical-layer channel. We derive the analytical expression of the delivery time at low SNR, and we consider also the regime of large number of users in order to

draw some valuable insights. Finally, some numerical results are used to validate the correctness of the derived expressions.