feedforward phase noise compensation

This work addresses the performance degradation of high-order modulation signals caused by phase noise in dispersion-managed fiber channels. To mitigate this impairment, the authors propose performing phase noise compensation prior to chromatic dispersion compensation and develop efficient feedforward and iterative algorithms within the Expectation Propagation framework. Evaluated in a 100 GBaud 64-QAM long-haul transmission system spanning 10,000 km, the proposed approach significantly suppresses phase noise effects, achieving information rates approaching those of an ideal channel without phase noise. This strategy overcomes the performance limitations inherent in conventional post-compensation schemes, thereby establishing a new benchmark for phase noise resilience in high-capacity coherent optical systems.

This work addresses the performance degradation caused by phase noise in intersymbol interference (ISI) channels and proposes a non-iterative feedforward phase noise compensation method based on the sum-product algorithm (SPA). The approach models the received signal as independent Gaussian random variables and incorporates the von Mises distribution to characterize phase mismatch. To the best of our knowledge, this is the first application of non-iterative SPA to phase noise compensation in ISI channels. Under comparable computational complexity, the proposed scheme significantly outperforms conventional linear filtering techniques across various channel conditions—including ISI-free, standard single-mode fiber, and OFDM multipath channels—achieving higher information transmission rates.

This study addresses the prohibitive channel equalization complexity in satellite and UAV communications caused by high Doppler shifts and fractional delays. To mitigate this, it constructs orthogonal waveform bases using the Generalized Discrete Affine Fourier Transform (GDAFT) and leverages maximal abelian subgroups of the Heisenberg–Weyl group to align with channel characteristics. A subgroup selection strategy is proposed to maximize the diagonal energy of the channel matrix, which, combined with Zak-OTFS modulation and a Neumann series approximation algorithm, enables efficient signal processing. The primary contribution lies in achieving robust single-tap equalization under both linear time-invariant and dual-path line-of-sight scenarios. This approach significantly reduces the required number of equalizer taps, effectively suppresses inter-carrier interference, and adaptively compensates for fractional offsets.

该研究针对多用户MIMO系统中由相位噪声引起的问题，提出了一种基于变分贝叶斯框架的方法进行联合相位噪声估计和数据检测，有效提升了通信系统的性能。

In 5G/6G short-block transmissions, quasi-coherent (QC) detection suffers significant performance and sensitivity degradation—especially at low spectral efficiencies—due to neglect of non-coherent terms; meanwhile, long-code decoding incurs prohibitive complexity. Method: This paper proposes a low-complexity receiver integrating adaptive DMRS/data power allocation with block-wise first-order Reed–Muller coding. Contribution/Results: We quantitatively characterize, for the first time, the critical impact of non-coherent terms in short-block QC detection. A unified QC-and-non-coherent joint detection framework is established, incorporating block-wise fast Hadamard transform (FHT) and power-coordinated optimization. Decoding complexity is reduced from *O*(*N*²) to *O*(*N* log *N*), while block error rate approaches the maximum-likelihood (ML) bound. The scheme substantially outperforms conventional LS+QC detection, achieving notable sensitivity gains—making it well-suited for small-payload, ultra-low-latency communication scenarios.

This work addresses the challenges posed by nonlinearities and memory effects in power amplifiers, which cause spectral regrowth, reduced energy efficiency, and degraded reliability—particularly under stringent adjacent channel leakage ratio (ACLR) constraints where efficiency and performance are difficult to balance. The authors propose a fully digital, transceiver-cooperative distortion compensation framework based on transfer learning: the transmitter integrates iterative clipping and filtering (ICAF) with static digital pre-distortion (SDPD), while the receiver employs a lightweight digital post-distortion (DPoD) network. Leveraging prior knowledge from an APTBM model, the DPoD is weakly supervised via modulation-structure-induced constraints for pre-training and further adapted online with few-shot learning to efficiently compensate residual distortions, aided by clipping noise cancellation. Evaluated under a 30-dBc ACLR constraint, the approach achieves reliable transmission with approximately 2 dB of input back-off, outperforming existing learning-based DPoD schemes by over 2 dB while substantially reducing online training overhead and computational complexity.

本文解决了连续时间AFDM波形的不连续性问题，该问题导致高带外发射。通过提出一种新的连续时间波形SFDM，保持频率在奈奎斯特采样间隔内不变，从而消除不连续性并减少OOBE。

This study addresses the ambiguity in the digital signal processing (DSP) chain between physical transceivers and periodic sensing models in communication-centric integrated sensing and communication (ISAC). We construct a unified DSP transceiver framework for single-antenna ISAC that integrates modulation, cyclic prefix insertion, and pulse shaping, thereby establishing a symbol-rate equivalent channel and a periodic matched filtering model. By revealing the DSP mapping mechanism from physical waveforms to the periodic model, the proposed framework achieves linearized communication reception and cyclic-shift equivalence for sensing reception. Numerical experiments validate the superiority of this framework in terms of target range estimation accuracy and bit error rate performance over Rayleigh fading channels under CP-OFDM transmission.

This study addresses the lack of a quantified relationship between constellation order and the optimal acceleration factor in faster-than-Nyquist (FTN) signaling. To bridge this gap, the authors compute finite-alphabet symbol-level achievable information rates (AIR) for BPSK through 16-QAM within an MMSE framework, cross-validated against Ungerboeck BCJR benchmarks, thereby establishing a quantitative model linking modulation order to temporal compression. The work reveals an intrinsic principle that higher-order modulations must approach the Nyquist limit to sustain spectral efficiency. Furthermore, it identifies the optimal acceleration factors at 6 dB for each modulation scheme—0.65 for BPSK and 0.90 for 16-QAM—and demonstrates their robustness across coded systems and varying detector memory lengths. These findings provide a theoretical foundation for operating point design in FTN systems.

This work addresses the failure of conventional channel estimation in orthogonal time frequency space (OTFS) systems under ultra-wide Doppler shifts, where severe aliasing degrades performance. To overcome this challenge, the authors propose a novel delay–Doppler (DD) domain training frame that integrates cosine pilots with pilot symbols, along with a two-stage channel estimation method. In the first stage, coarse estimates of path delays, aliased Doppler shifts, and channel gains are obtained directly in the DD domain. The second stage leverages time–frequency transformation and cosine pilot detection to identify true Doppler peaks and performs parameter pairing via thresholding. This approach uniquely combines DD-domain and frequency-domain information to resolve ultra-wide Doppler aliasing, significantly reducing normalized mean square error (NMSE) and improving bit error rate (BER) performance.