An extensive review of THz communication in 6G: Facilitating technologies with edge computing and native AI

The transition toward sixth-generation networks is expected to enable unprecedented capabilities such as immersive extended reality, holographic telepresence, large-scale digital twins, and ultra-reliable autonomous communications. Unlocking these services requires exploiting the terahertz spectrum, which provides massive bandwidth but also presents formidable challenges, including severe propagation loss, beam misalignment, blockage sensitivity, and the need for fine-grained synchronization in dynamic environments. To address these, this article advances a Native-AI paradigm where artificial intelligence and edge computing are embedded as intrinsic elements of the THz communication fabric rather than treated as external add-ons. Within this framework, AI methods such as deep reinforcement learning, graph neural networks, and federated learning are harnessed for proactive channel estimation, adaptive beamforming, RIS-assisted propagation control, and distributed mobility management across heterogeneous nodes, including base stations, UAV relays, and edge servers. The proposed architecture emphasizes hierarchical orchestration, fairness-aware resource allocation, robustness against adversarial threats, and zero-touch management to ensure reliable and secure operation in high-mobility and heterogeneous conditions. Our contribution lies in providing a unified architectural blueprint and technical mapping that connects native AI mechanisms to core THz communication functions, highlighting how synchronization, resilience, and scalability can be achieved through distributed coordination and intent-driven orchestration. Beyond architectural design, the paper discusses performance implications and practical deployment considerations, including challenges of non-IID data, energy and compute heterogeneity, and the vulnerability of distributed AI to adversarial manipulation. Strategies such as TinyML-based model compression, Byzantine-resilient federated protocols, and explainable AI are outlined as enablers of sustainable, trustworthy deployment. By synthesizing current research trends with original design insights, this work positions native AI as a foundational principle for 6G—transforming THz systems into intelligent, secure, and autonomous infrastructures capable of meeting the stringent demands of next-generation services.