Optimal Wireless Caching with Placement Cost

TL;DR

This paper analyzes how including placement costs affects the efficiency of coded caching schemes, revealing scenarios where simple uncoded caching can be optimal under realistic cost considerations.

Contribution

It introduces a general caching framework that incorporates placement costs and analyzes their impact on optimal caching strategies in various network settings.

Findings

Placement costs significantly influence caching scheme optimality.

Uncoded caching can be optimal in certain network configurations.

The framework applies to diverse network architectures and traffic patterns.

Abstract

Coded caching has been shown to result in significant throughput gains, but its gains were proved only by assuming a placement phase with no transmission cost. A free placement phase is, however, an unrealistic assumption that could stand as an obstacle towards delivering the promise of coded caching. In [1], we relaxed this assumption by introducing a general caching framework that captures transmission costs for both delivery and placement phases under general assumptions on varying network architectures, memory constraints, and traffic patterns. Here, we leverage this general framework and focus on analyzing the effect of the placement communication cost on the overall throughput and the structure of the optimal caching scheme, under the assumptions of the worst case traffic pattern and unlimited memory at the end users. Interestingly, we find relevant network configurations where uncoded caching is the optimal solution.