Topological Content Delivery with Feedback and Random Receiver Cache

TL;DR

This paper investigates how random receiver caches and feedback influence content delivery rates in interference channels with changing topologies, establishing bounds and achievable rates that highlight the benefits of side-information.

Contribution

It introduces new outer-bounds on achievable rates considering random receiver caches and delayed channel knowledge, revealing the potential rate improvements from side-information.

Findings

Outer-bounds on achievable rates with receiver caches

Potential rate boost from small side-information

Matching achievable rates in specific scenarios

Abstract

We study the problem of content delivery in two-user interference channels with altering topology, random available cache at the receivers, and delayed channel knowledge at the transmitter. We establish a new set of outer-bounds on the achievable rates when each receiver has access to a random fraction of the message intended for the other receiver, and when each transmitter is aware of which part of its own message is known to the unintended receiver. The outer-bounds reveal the significant potential rate boost associated with even a small amount of side-information at each receiver. The key in deriving the bounds is to quantify the baseline entropy that will always become available to the unintended receiver given the altering topology and the already available side-information. We will also present matching achievable rates in certain scenarios and outline the challenges in more general settings.