Papers

PACT:解码器置信度引导的亚THz OFDM信道与相位噪声联合估计接收机

PACT: A Decoder-Confidence-Guided Two-Round Receiver for Joint Channel and Phase-Noise Estimation in Sub-THz OFDM

Why we picked it

做 6G 或太赫兹通信仿真的朋友可以看看这篇:它把译码置信度反哺回信道估计,比纯学习型基线在 30 dB 时码块错误率低了 57% 到 70%,而且权重冻结就能换频段换带宽,工程迁移成本不高。

论文提出 PACT 接收机,针对亚THz OFDM 系统中频率选择性衰落与振荡器相位噪声(PN)联合损伤的问题,将 pilot-aware complex transformer 估计器、带状 LMMSE 均衡和神经归一化 min-sum 译码组合成两轮接收流程。PACT 通过交互的复数值分支,同时估计频域信道和等效时域相位噪声;第一轮译码后,由译码器导出的伪导频和连续的星座感知置信度,经 PACT 输入与注意力机制引导第二轮精化,两轮之间参数共享。

Body

在 Eb/N0=30 dB 的两个基准设置下,PACT 接收机的码块错误率分别比表现最好的学习型基线降低约 70% 和 57%。冻结权重测试显示,该接收机无需重训练即可跨信道/PN 模型、载波频率以及 FFT/带宽配置工作。

Original · arXiv cs.LG

PACT: A Decoder-Confidence-Guided Two-Round Receiver for Joint Channel and Phase-Noise Estimation in Sub-THz OFDM

Frequency-selective fading and oscillator phase noise (PN) jointly impair coded sub-THz OFDM reception. We propose a decoder-confidence-guided two-round receiver combining a {pilot-aware complex transformer} (PACT) estimator, banded linear minimum mean-square error equalization, and neural normalized min-sum decoding. PACT jointly estimates the frequency-domain channel and effective time-domain PN through interacting complex-valued branches. Decoder-derived pseudo-pilots and continuous constellation-aware confidence guide second-round refinement through PACT inputs and attention, with parameters shared across rounds. At $E_b/N_0=30$~dB, the PACT receiver achieves approximately $70\%$ and $57\%$ lower code-block error rates than the best-performing learning-based baseline in the two benchmark settings, respectively. Frozen-weight tests further demonstrate applicability across channel/PN models, carrier frequencies, and FFT/bandwidth configurations.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include PACT, OFDM, 6G.

Source attribution: curated from arXiv cs.LG (@Ruilin Wang, Xiaodai Dong).