Hybrid ARQ in Multiple-Antenna Slow Fading Channels: Performance Limits and Optimal Linear Dispersion Code Design

Cong Shen, Michael P. Fitz

Introduction

This research explores the synergy between Multiple-Input Multiple-Output (MIMO) antenna systems and Hybrid Automatic Repeat Request (HARQ) protocols. While MIMO technology enhances data rates through spatial multiplexing, HARQ ensures communication reliability by retransmitting failed packets. This paper provides a rigorous framework for designing space-time codes optimized for retransmission-based systems in slow-fading environments. The authors demonstrate that while Incremental Redundancy (IR) is theoretically superior in throughput, well-designed Linear Dispersion Codes (LDCs) can achieve near-optimal performance with significantly reduced receiver complexity.

System Model and Problem Statement

In wireless communications, "slow fading" describes a scenario where the channel remains constant over several transmission attempts. Consequently, retransmissions do not inherently benefit from time diversity; if a channel is poor during the initial transmission, it remains poor for subsequent attempts.

Traditional HARQ protocols, such as Chase Combining (CC) or Incremental Redundancy (IR), were originally developed for single-antenna channels. When applied to MIMO systems, these protocols often fail to exploit the spatial degrees of freedom provided by multiple antennas. Much of the existing literature focuses on high-SNR asymptotics, leaving a gap in the design of fixed-rate codes that perform well at realistic, finite signal-to-noise ratios (SNR).

Methodology: Linear Dispersion Code Design

The authors utilize Linear Dispersion Codes (LDC) as a generalized mathematical structure to represent various space-time block codes. Their design approach is based on two primary criteria:

  1. Capacity Losslessness (Criterion 1): Derived from information-theoretic analysis, this criterion suggests that an ideal LDC should create an equivalent MIMO channel that preserves the full capacity of the physical channel in every ARQ round.
  2. Diversity Order (Criterion 2): Based on error probability analysis, this ensures the code achieves the maximum possible diversity gain ($L_r \cdot \min\{T^{(n)}, L_t\}$) at each stage of retransmission.

The study evaluates several prominent codes—including the Alamouti code, the Golden Code, and Cyclic Delay Diversity (CDD)—against these criteria using analytical proofs and Monte Carlo simulations of i.i.d. Rayleigh fading channels across various antenna configurations (2x1 MISO, 2x2 MIMO, and 4x2 MIMO).

Key Insights and Results

The research yields several critical insights into the design of MIMO-HARQ systems:

  • The "Aggressive" Strategy: A successful design philosophy involves sending independent data symbols in the first round to maximize rate. If the transmission fails, subsequent rounds send symbols that, when combined with the first, form a robust space-time code (e.g., Alamouti).
  • IR vs. LDC-HARQ: While IR is the "gold standard" for throughput due to its ability to achieve ergodic capacity, it requires complex code combining. LDC-based HARQ can match IR's performance if it satisfies the capacity losslessness criterion, using much simpler packet-level combining.
  • Suboptimality of Chase Combining: CC is shown to be strictly suboptimal in MIMO environments. By simply repeating the same signal, it fails to spread symbols across different spatial dimensions in subsequent rounds.
  • Throughput vs. Reliability: In coded systems with long block lengths, HARQ primarily boosts throughput (average rate) rather than packet error rate. The reliability gain is most significant in uncoded or short-block systems.

Specific Code Performance

  • 2x1 MISO: The Alamouti code is proven optimal, satisfying both capacity and diversity criteria.
  • 2x2 MIMO: The Golden Code and Hassibi-Hochwald (HH) codes are shown to be average-rate optimal.
  • 4x2 MIMO: Existing codes, such as Orthogonal STBC, are found to be suboptimal due to their low rates. This identifies a significant need for new code designs in higher-dimensional systems.

Conclusion and Future Work

The framework presented assumes a static channel over retransmissions. Extending this to time-varying channels remains a challenge. Furthermore, while the paper establishes criteria for optimal codes, it does not provide a constructive method for generating codes in high-dimensional systems (e.g., 4x4 or 8x8). Future research should also consider "fast-decodable" properties to reduce the computational burden on mobile hardware.

Summary updated

BibTeX
@article{shen2011hybrid,
  author = {Cong Shen and Michael P. Fitz},
  title = {{Hybrid ARQ in Multiple-Antenna Slow Fading Channels: Performance Limits and Optimal Linear Dispersion Code Design}},
  journal = {IEEE Trans. Information Theory},
  year = {2011},
  doi = {10.1109/tit.2011.2161949},
}
Figure