Using DSM and LDPC coding, we successfully achieved error-free transmission for the PDM-1024QAM signal and high-quality transmission for the 2048QAM signal in the 400G photonic-assisted terahertz communication system. Without using LDPC, the BER of 1024-QAM signals can be reduced to 3.8e-3, while the BERs for 2048-QAM and 4096-QAM signals can be decreased to below 4.2e-2. When LDPC is employed for error correction, 1024-QAM signals achieve error-free transmission, and the BER of 2048-QAM signals can be further reduced to 9.8e-5.

This paper proposes a physical-layer enhancement scheme for photonics-aided terahertz (THz) system based on block-wise statistical permutation probabilistic shaping (PS) and phase masking. The scheme employs a three-level encryption strategy to sequentially encrypt the bit sequence, the mapped symbols, and the matrix carrying PS information, thereby enhancing overall security. Through blocking operations, the uniform distribution of the original sequence is disrupted, introducing dynamic fluctuations in the occurrence probabilities of symbols within each data block. This enables the establishment of a dynamic mapping mechanism, which maps high-frequency symbols to low-energy constellation points, thereby reducing the average transmission power while improving bit error rate (BER) performance. Transmission experiments were conducted in a 320 GHz photonics-aided THz system over a channel comprising 10 km of standard single-mode fiber (SSMF) and a 5 m wireless link. The results show that, compared with conventional 16QAM, the proposed scheme achieves maximum BER gains of 6.08 dB, 5.91 dB, and 5.60 dB at symbol rates of 1, 2, and 4 Gbaud, respectively. Furthermore, the scheme provides a key space of approximately $10^{63}$, making it robust against brute-force attacks. The experiments validate the effectiveness of the proposed approach in enhancing both system security and error performance.

This study investigates trellis shaping (TS) for mitigating nonlinear impairments in photonic terahertz (THz) communication systems. A parallel shaping architecture integrating hierarchical modulation (HM) with adjustable power allocation is proposed. The scheme applies TS to structure the data stream hierarchically and employs superposition coding (SC) to generate non-uniformly distributed TS-16QAM signals, enhancing transmission performance. Experimental validation over a 5-m link at 320 GHz demonstrates that the proposed scheme achieves a 1.65 dB improvement in input power sensitivity and a 1.39 dB gain in Q-factor compared to the Maxwell–Boltzmann (MB) distribution. Furthermore, the system provides enhanced power allocation flexibility, expanding the tunable range of the hierarchical power ratio by a factor of 1.37.

We propose a probabilistic shaping (PS) scheme based on a single-layer lookup table (LUT) that employs only one LUT for symbol mapping while achieving favorable system performance. This scheme reduces the average power of the signal by adjusting the symbol distribution using a specialized LUT architecture and a flexible shaping proportion. The simulation results indicate that the proposed PS scheme delivers performance comparable to that of the conventional constant-composition distribution-matching-based probabilistic shaping (CCDM-PS) algorithm. Specifically, it reduces the bit error rate (BER) from 1.2376 $\times {10}^{-4}$ to 6.3256 $\times {10}^{-5}$, corresponding to a 48.89% improvement. The radial basis function neural network (RBFNN) effectively compensates for nonlinear distortions and further enhances transmission performance due to its simple architecture and strong capacity for nonlinear learning. In this work, we combine lookup-table-based probabilistic shaping (LUT-PS) with RBFNN-based nonlinear equalization for the first time, completing the transmission of 16-QAM OFDM signals over a photonic terahertz-over-fiber system operating at 400 GHz. Simulation results show that the proposed approach reduces the BER by 81.45% and achieves a maximum Q-factor improvement of up to 23 dB.

This paper proposes a dual-side hybrid embedding network (DHEN) to mitigate gain degradation in terahertz amplifiers at frequencies near $f_{\max }$. The proposed approach employs a pre-embedding network for parasitic absorption, followed by Y-embedding for gain enhancement. A theoretical analysis is conducted to derive the embedding conditions for process-constrained circuit synthesis. In this architecture, a capacitive base-side pre-embedding provides intrinsic DC isolation, while a defected-ground-structure (DGS) inductor realizes the Y-embedding inductive element with reduced layout area. Based on the DHEN, a four-stage amplifier is designed in a 130 nm SiGe BiCMOS process. Electromagnetic co-simulation results demonstrate a power gain of 19.3 dB at 280 GHz, corresponding to an 11.5 dB improvement over a conventional unboosted amplifier. The proposed approach provides a unified synthesis methodology that simultaneously addresses parasitic absorption, DC isolation, and gain enhancement for near-$f_{max}$ THz amplifier design.