Abstract
We consider probabilistic amplitude shaping (PAS) as a means of increasing the spectral efficiency of fiber-optic communication systems. In contrast to previous works in the literature, we consider probabilistic shaping with hard decision decoding (HDD). In particular, we apply the PAS recently introduced by B"ocherer et al. to a coded modulation (CM) scheme with bit-wise HDD that uses a staircase code as the forward error correction code. We show that the CM scheme with PAS and staircase codes yields significant gains in spectral efficiency with respect to the baseline scheme using a staircase code and a standard constellation with uniformly distributed signal points. Using a single staircase code, the proposed scheme achieves performance within $0.57$--$1.44$ dB of the corresponding achievable information rate for a wide range of spectral efficiencies.
Introduction
To meet the insatiable demand for data capacity, fiber-optic systems must maximize the spectral efficiency of their links. Standard modulation schemes, such as Quadrature Amplitude Modulation (QAM), typically employ "uniform signaling," where all constellation points are transmitted with equal probability. This approach, however, leaves a fundamental "shaping gap" of approximately 1.53 dB compared to the theoretical Shannon limit.
Closing this gap generally requires Probabilistic Amplitude Shaping (PAS) combined with Soft-Decision Decoding (SDD). While SDD is highly effective, its implementation at the multi-terabit speeds required for modern optical networks is often prohibited by extreme power consumption and hardware complexity. Hard-Decision Decoding (HDD) offers a significantly more energy-efficient alternative, but its compatibility with PAS to maximize spectral efficiency has remained an open question until now.
Methods
The authors propose a Coded Modulation (CM) scheme that integrates PAS with bit-wise hard-decision metrics and staircase codes. The system architecture is designed to maintain low hardware complexity while capturing shaping gains.
Key components of the architecture include:
- Probabilistic Amplitude Shaping (PAS): The framework is adapted to work with bit-wise hard-decision metrics rather than the soft-decision metrics for which it was originally designed.
- Constant Composition Distribution Matching (CCDM): This block transforms uniform information bits into a shaped Maxwell-Boltzmann distribution, prioritizing lower-energy signals to improve efficiency.
- Staircase Codes: These high-performance Forward Error Correction (FEC) codes are specifically tailored for high-speed optical communications. The scheme maps shaped bits into the staircase code structure while maintaining systematic encoding.
- Optimization: The input distribution is mathematically optimized to maximize the Achievable Information Rate (AIR) specifically for a hard-decision receiver.
Results and Discussion
A major finding of this work is that probabilistic shaping provides substantial gains even when the receiver utilizes "dumb" hard-decision bits. This challenges the common assumption that shaping's benefits are strictly tied to soft-decision processing.
As illustrated in the figure above, the PAS-HDD scheme achieves significant shaping gains over standard uniform signaling. For a 256-QAM constellation (corresponding to 16-ASK per dimension), the shaping gain reaches up to 2.88 dB. At a spectral efficiency of 3 bits per channel use, the gain is approximately 2 dB.
Furthermore, the proposed scheme demonstrates remarkable versatility. A single staircase code can cover a broad range of spectral efficiencies by simply adjusting the shaping distribution in software, rather than redesigning FEC hardware for different link conditions. The system operates within 0.57–1.44 dB of the theoretical AIR across its operating range and maintains a predictable gap of about 1.9 dB to the absolute soft-decision Shannon capacity.
Conclusion
This study demonstrates a practical path toward high-efficiency, low-power optical communications by combining PAS with HDD-based staircase codes. While the current evaluation focuses on the Additive White Gaussian Noise (AWGN) channel, it provides a robust foundation for future work addressing fiber nonlinearities. Future research may also explore "shaping-light" architectures to reduce the computational overhead of the distribution matcher or investigate non-binary code structures to further close the gap to the Shannon limit.
Summary updated