Optimizing Incremental Redundancy in Hybrid Automatic Repeat Request

31 concepts27 papers1 proposal

In wireless communications, channels fluctuate unpredictably due to fading, mobility, and environmental interference. Traditional Automatic Repeat Request (ARQ) protocols handle transmission errors by discarding corrupted packets and requesting an identical retransmission. However, as Vaze notes, simply sending identical packets fails to exploit advanced error correction capabilities.

To maximize throughput and reliability, modern networks rely on Hybrid ARQ (HARQ) with Incremental Redundancy (IR). Instead of treating a failed transmission as a total loss, the receiver stores corrupted packets in a buffer to combine their partial information with subsequent attempts (Li et al.).

The Intuition of Information Accumulation

The core mechanism of IR-HARQ relies on the progressive accumulation of mutual information. When a transmission fails, the transmitter does not repeat the exact same bits. Instead, it sends new parity bits generated from the original data.

As Graell i Amat et al. and Makki et al. explain, this allows the receiver to aggregate mutual information over multiple attempts. Each retransmission provides unique redundancy, effectively lowering the code rate over time (Makki et al.). This accumulation transforms a series of independent failures into a single, successful joint decoding process. While simple network models often rely on binary success/failure feedback, Aggarwal et al. emphasize that capturing this soft-combining and incremental information gain is essential for true resource efficiency.

Coding Gain vs. Diversity Gain

Why is sending new parity bits better than repeating the original packet (a method known as Chase Combining)? The answer lies in the difference between diversity gain and coding gain.

Repetition merely increases the received signal-to-noise ratio (SNR) by averaging out noise across identical copies. Incremental redundancy, however, expands the underlying error-correcting code space.

Zedini et al. demonstrate that this coding gain makes IR significantly more resilient to deep signal fades—such as those caused by atmospheric turbulence in optical links—than simple code combining. Furthermore, Makki et al. show that by providing cumulative coding gains rather than just relying on diversity, IR protocols consistently achieve superior throughput and lower outage probabilities under strict constraints.

Optimizing the Retransmission Profile

To fully exploit IR-HARQ, system designers must optimize how resources are allocated across transmission rounds. This involves two main levers: power allocation and codeword structure.

Power Allocation

For power allocation, Graell i Amat et al. show that an "aggressive-first" strategy—allocating more power to the initial transmission—frequently yields the highest efficiency under reliability constraints. Guo et al. further establish that this coding gain provides superior power efficiency over repetition-based methods, particularly at high data rates where it lowers the SNR required to maintain target outage probabilities.

Codeword Structure

For codeword structure, Arafa et al. prove that the optimal strategy in highly noisy environments balances a short initial codeword with significantly longer subsequent incremental redundancy sequences. This minimizes total transmission time while ensuring robust error correction if the first attempt fails.

Designing these patterns is challenging; Wiesmayr et al. note that optimizing initial puncturing patterns must carefully account for how they interact with subsequent HARQ retransmission stages. Furthermore, as Anand et al. point out, mathematically modeling these unique redundancy stages remains highly complex compared to simple repetition, posing a challenge for closed-form low-latency optimization.

The Freshness Dilemma: Age of Information

When a network prioritizes data freshness—measured as the Age of Information (AoI)—IR-HARQ introduces a compelling strategic tradeoff.

As Gunduz et al. explain, because the receiver accumulates information with each failed attempt, the probability of a successful decode increases with each subsequent retransmission. This forces the scheduling logic to choose between two paths:

  1. Retransmit a stale packet: Higher probability of success due to accumulated info, but the delivered data will be older.
  2. Send a fresh packet: Lower initial success chance, but if successful, the delivered data will be perfectly fresh.

Gunduz et al. demonstrate that integrating HARQ into age-aware scheduling via reinforcement learning allows systems to dynamically navigate this choice, utilizing accumulated information to drastically reduce average data age compared to standard retransmission schemes.

Masking Physical Layer Impairments

Beyond standard fading channels, IR-HARQ serves as a robust architectural layer that masks physical and hardware imperfections. For instance, Makki et al. show that IR-HARQ can effectively compensate for hardware impairments like inefficient power amplifiers in millimeter-wave networks.

Similarly, Makki et al. and Makki et al. show that accumulating mutual information over successive parity rounds mitigates severe physical layer disruptions, such as pointing errors and transmitter misalignments in hybrid radio-optical links. By shifting the burden of reliability from perfect hardware execution to mathematical information accumulation, IR-HARQ provides a predictable path to high-throughput, low-latency wireless communication.

Key Foundational Work

Performance of 1xEV-DO third-generation wireless high-speed data systems (2003) — contributed 2 concept elements to this direction.
Throughput-Delay-Reliability Tradeoff with ARQ in Wireless Ad Hoc Networks (2010) — contributed 1 concept elements to this direction.
History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 27 papers
  2. Version 1 · 2026-10-04 · 38 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_incremental_redundancy_optimization_a3dfb55a,
  title = {incremental redundancy optimization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/incremental-redundancy-optimization/rdr_a3dfb55a0bc04520b82304f4a71933f0},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 2, frozen text):

@misc{lacuna_incremental_redundancy_optimization_a3dfb55a_v2,
  title = {incremental redundancy optimization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/incremental-redundancy-optimization/rdr_a3dfb55a0bc04520b82304f4a71933f0/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake