TL;DR
This paper introduces a novel "symbol-wise puncturing" technique to improve the integration of Probabilistic Amplitude Shaping (PAS) with Hybrid Automatic Repeat Request (HARQ) in wireless communications. By puncturing symbols at a finer granularity, the method preserves the desired probability distribution of PAS across retransmissions, leading to more stable performance gains compared to traditional methods.
What Problem Does This Paper Address?
Modern wireless systems strive for higher data rates and reliability. Probabilistic Amplitude Shaping (PAS) is a technique that can bridge the gap between theoretical Shannon limits and practical system performance by shaping the distribution of transmitted symbols to be more Gaussian-like. However, integrating PAS with Hybrid Automatic Repeat Request (HARQ), a crucial mechanism for ensuring reliability through retransmissions, presents a challenge. Traditional HARQ schemes often use "sequential puncturing," where only redundant information (sign bits) is sent in retransmissions. This can severely distort the carefully crafted probability distribution of PAS, diminishing its benefits, especially after multiple retransmissions. The paper aims to solve this by proposing a method that maintains the PAS distribution's integrity throughout the HARQ process.
Methods
The core innovation is the "symbol-wise puncturing" scheme. Instead of puncturing bits sequentially, the authors propose pairing sign bits with amplitude bits into a "label sequence" for shaped modulation symbols. This sequence is then punctured. This means that in each retransmission, a portion of both the amplitude and sign bits are sent, preserving the overall desired probability distribution.
Key steps include:
- PAS Encoder: A standard PAS encoder (using a Constant Composition Distribution Matcher - CCDM) transforms input bits into shaped amplitude symbols.
- Symbol Labeling: These shaped symbols are then represented by a sequence of bits that includes both sign bits (for quadrant selection) and amplitude bits (for position within the quadrant).
- Symbol-Wise Puncturing: This combined bit sequence is then punctured. The paper defines how to determine the number of bits to transmit in each retransmission to maintain a certain throughput.
- Inverse Puncturing at Receiver: At the receiver, the inverse process reconstructs the codeword. Crucially, for amplitude bits that were not transmitted in a given retransmission, the receiver uses their a-priori information (derived from the desired distribution) to enhance the decoding process, rather than simply assuming zeros.
Datasets
Not discussed in the paper. The evaluation relies on simulations using defined channel models (AWGN, Rayleigh fading) and parameters.
Key Insights and Ideas
- Preserving the PAS Gain: The most significant insight is that traditional sequential puncturing fundamentally undermines PAS by distorting its distribution. Symbol-wise puncturing offers a way to "have your cake and eat it too" – achieving high data rates with PAS while maintaining reliability through HARQ.
- The Power of A-Priori Information: The clever use of a-priori information for missing amplitude bits at the receiver is a key enabler. This allows the system to leverage the knowledge of the desired distribution even when some bits are not explicitly transmitted in a given retransmission.
- Stable Performance: Unlike sequential puncturing, which sees its gains erode rapidly with retransmissions, the symbol-wise approach provides a stable shaping gain across various signal-to-noise ratios (SNRs) and retransmission cycles.
Key Results
- Stable Shaping Gain: The proposed symbol-wise puncturing scheme achieves a stable shaping gain of at least 0.6 dB compared to uniform signaling across the SNR range, under the same throughput. This gain remains consistent even with retransmissions.
- Reduced Distribution Distortion: In each transmission, the symbol-wise method reduces distribution distortion, leading to a gain of over 1.2 dB at a Block Error Rate (BLER) of $10^{-3}$ compared to uniform signaling.
- Improved BLER: For sequential puncturing, the distribution becomes severely distorted in retransmissions, leading to worse BLER performance than uniform signaling. The symbol-wise method, however, maintains better BLER performance.
- MIMO Performance: The benefits extend to MIMO channels, where the symbol-wise puncturing still provides a stable shaping gain of about 1 dB, while sequential puncturing offers no gain.
Limitations and Future Work
- Complexity: While the paper mentions low implementation complexity for PAS itself, the symbol-wise puncturing adds a layer of complexity to the HARQ process. The paper doesn't deeply analyze the computational overhead of the proposed inverse puncturing with a-priori information.
- HARQ Scheme Focus: The current work focuses on Incremental Redundancy HARQ (IR-HARQ). The authors suggest that their approach could be extended to other HARQ schemes (like Type-I or Type-III) and potentially different Forward Error Correction (FEC) codes.
- Optimization: Future work could explore optimizing the puncturing patterns and the distribution of amplitude and sign bits across transmissions for even greater efficiency.
Summary updated