Terahertz Links: Channel Modeling and Hardware Constraints

Abstract

Terahertz (THz) communication, spanning roughly 0.1 to 10 THz, is poised to unlock ultra-high data rates in next-generation systems. However, its deployment is constrained by unique channel characteristics and stringent hardware limitations. This paper presents an integrated review of THz channel modeling techniques and hardware obstacles, drawing on research published prior to 2022. Channel modeling approaches at THz frequencies include frequency-domain measurements using vector network analyzers (VNAs), time-domain techniques such as sliding correlation and THz time-domain spectroscopy (THz-TDS) arXivACM Digital Library. These methods support three modeling paradigms—deterministic, stochastic, and hybrid— that capture path loss, molecular absorption, scattering, and spatio-temporal nonstationarity arXivTechRxiv. Recent models also incorporate geometry-based stochastic frameworks suitable for ultramassive MIMO and dynamic environments arXiv. Additionally, high-resolution indoor models for mmWave and sub-THz (up to 150 GHz) environments offer useful statistical insights for wider THz design arXiv. On the hardware front, the THz "gap" presents a core challenge: conventional electronics struggle with these frequencies, and optical sources cannot easily scale down. This necessitates novel oscillators or multipliers with very low output power (<–10 dBm) graphyonline.com. High propagation and molecular absorption losses demand highly directional and compact antennas, such as horn or plasmonic nano-antennas made of graphene, which also bring fabrication and beam-steering complexities IET Research JournalWikipedia+1. Material-specific scattering mechanisms, especially at rough surfaces and high frequencies, further influence channel reliability and modeling accuracy arXiv. Despite these formidable challenges, advances in modeling pave the way for effective THz link design, while hardware innovation—such as graphene antennas—offers a path forward. This paper synthesizes these findings, identifies gaps, and outlines future directions toward bridging channel and device-level limitations.