[Paper Review] Structural Properties of Uncoded Placement Optimization for Coded Delivery

Sian Jin, Ying Cui+2 more|arXiv (Cornell University)|Jul 22, 2017

Caching and Content Delivery13 references34 citations

TL;DR

This paper proposes an optimized uncoded placement strategy for centralized coded caching under arbitrary file popularity, minimizing average load via convex optimization. It derives structural properties leading to equivalent linear programs with reduced variables and provides a closed-form solution under uniform popularity, matching the Maddah-Ali–Niesen scheme.

ABSTRACT

A centralized coded caching scheme has been proposed by Maddah-Ali and Niesen to reduce the worst-case load of a network consisting of a server with access to N files and connected through a shared link to K users, each equipped with a cache of size M. However, this centralized coded caching scheme is not able to take advantage of a non-uniform, possibly very skewed, file popularity distribution. In this work, we consider the same network setting but aim to reduce the average load under an arbitrary (known) file popularity distribution. First, we consider a class of centralized coded caching schemes utilizing general uncoded placement and a specific coded delivery strategy, which are specified by a general file partition parameter. Then, we formulate the coded caching design optimization problem over the considered class of schemes with 2^K2^N variables to minimize the average load by optimizing the file partition parameter under an arbitrary file popularity. Furthermore, we show that the optimization problem is convex, and the resulting optimal solution generally improves upon known schemes. Next, we analyze structural properties of the optimization problem to obtain design insights and reduce the complexity. Specifically, we obtain an equivalent linear optimization problem with (K+1)N variables under an arbitrary file popularity and an equivalent linear optimization problem with K+1 variables under the uniform file popularity. Under the uniform file popularity, we also obtain the closed form optimal solution, which corresponds to Maddah-Ali-Niesen's centralized coded caching scheme. Finally, we present an information-theoretic converse bound on the average load under an arbitrary file popularity.

Motivation & Objective

• To minimize the average delivery load in a centralized coded caching system under arbitrary, known file popularity distributions.
• To design a coded caching scheme that leverages non-uniform file popularity by optimizing uncoded placement via a general file partition parameter.
• To reduce computational complexity by revealing structural properties of the optimization problem.
• To derive equivalent linear programs with significantly fewer variables for efficient computation.
• To establish an information-theoretic converse bound on the average load for arbitrary popularity.

Proposed method

• Formulates the coded caching design as a convex optimization problem over a class of schemes with general uncoded placement and a specific coded delivery strategy.
• Introduces a file partition parameter to represent the fraction of each file stored in each user's cache, enabling optimization over 2^K * N^K variables.
• Transforms the original non-linear, non-convex problem into an equivalent linear program with (K+1)N variables under arbitrary popularity and K+1 variables under uniform popularity.
• Applies Karush-Kuhn-Tucker (KKT) conditions to prove uniqueness and derive the optimal solution, showing it reduces to the Maddah-Ali–Niesen scheme under uniform popularity.
• Derives a closed-form optimal solution for the uniform popularity case, corresponding to the known order-optimal scheme.
• Establishes an information-theoretic converse bound on the average load to validate optimality in the uniform case.

Experimental results

Research questions

• RQ1 Can the average load in coded caching be minimized under arbitrary file popularity distributions using optimized uncoded placement?
• RQ2 What structural properties emerge in the optimization problem that allow for complexity reduction?
• RQ3 Does the optimal solution under uniform popularity reduce to the known Maddah-Ali–Niesen scheme?
• RQ4 Can the original high-dimensional optimization problem be transformed into a lower-dimensional linear program?
• RQ5 What is the fundamental limit on the average load under arbitrary popularity, and how close is the proposed scheme to it?

Key findings

• The optimization problem for minimizing average load under arbitrary file popularity is convex, ensuring global optimality and improved performance over existing schemes.
• Under arbitrary popularity, the problem reduces to an equivalent linear program with (K+1)N variables, significantly reducing computational complexity.
• Under uniform popularity, the problem reduces to a linear program with only K+1 variables, enabling efficient computation.
• The closed-form optimal solution under uniform popularity matches the Maddah-Ali–Niesen centralized coded caching scheme, confirming its optimality in this regime.
• The optimal solution is unique and derived via KKT conditions, with explicit expressions for the optimal file partition parameters.
• An information-theoretic converse bound is established, showing the proposed scheme achieves the minimum average load under uniform popularity.