Cache-Aided MIMO Communications: DoF Analysis and Transmitter Optimization

TL;DR

This work addresses the problem of efficiently delivering cacheable content over cache-enabled MIMO networks by jointly leveraging coded caching and spatial multiplexing under linear processing. It develops a unified DoF framework that optimizes the number of users served per transmission $\Omega$ and the per-user streams $\beta$, achieving a DoF of $\Omega\beta$ subject to $\beta \le \min\big(G, \frac{L\binom{\Omega-1}{t}}{1+(\Omega-t-1)\binom{\Omega-1}{t}}\big)$, thereby surpassing prior $Gt+L$ DoF results and removing the $L/G$ divisibility constraint. For finite-SNR, the paper introduces bit-level interference cancellation, full-size XOR transmissions, and hypergraph-based scheduling to design a new class of linear multicast schemes, complemented by a practical linear beamforming strategy that handles partially overlapping multicast groups. Numerical results show significant DoF gains and improved symmetric rates across SNR regimes, demonstrating both theoretical and practical value for CC-enabled MIMO systems. Overall, the work offers a scalable pathway to integrate CC into next-generation MIMO networks with enhanced delivery rates and manageable subpacketization under linear processing.

Abstract

Cache-aided MIMO communications aims to jointly exploit both coded caching~(CC) and spatial multiplexing gains to enhance communication efficiency. In this paper, we analyze both the achievable degrees of freedom~(DoF) under linear processing constraint and the finite-SNR performance of a MIMO-CC system with CC gain \(t\), where a server with \(L\) transmit antennas communicates with \(K\) users, each equipped with \(G\) receive antennas. We first demonstrate that the enhanced DoF of \(\max_{β, Ω} Ω\times β\) is achievable with linear processing, where the number of users \(Ω\) served in each transmission is fine-tuned to maximize DoF, and \(β\le \min\big(G, \nicefrac{L \binom{Ω-1}{t}}{\big(1 + (Ω- t - 1)\binom{Ω-1}{t}}\big)\big)\) represents the number of parallel streams decoded by each user. Then, we propose a new class of MIMO-CC schemes using a novel scheduling mechanism leveraging maximal multicasting opportunities to maximize delivery rates at given SNR levels while still adhering to linear processing constraints. This new class of schemes is paired with an efficient linear multicast beamformer design, resulting in a more practical, high-performance solution for integrating CC in future MIMO systems.

Figures (8)

• Figure 1: MIMO-CC system model and user selection for different $\Omega$.
• Figure 2: Behavior of the solution to \ref{['eq:total_DoF']} in Corollary \ref{['remark1-Dof']} for $L$=16, $t$=1.
• Figure 3: MIMO-CC multicast scheduling: a base scheduling block of size $B_0 \times S_0$ is repeated $\delta$ times, and $\eta$ columns are selected from the resulting table for each interval, with two arbitrary options for $\eta$.
• Figure 4: Achievable DoF of UC and MU-MIMO.
• Figure 5: The achievable DoF of UC, MC schemes, $(G,t)=(8,2)$.

Theorems & Definitions (24)

• Remark 1
• Example 1
• Theorem 1
• proof
• Theorem 2
• proof
• Remark 2
• Corollary 1
• Lemma 1
• Example 2