Artificial intelligence driven digital twin for dynamic reconfigurable intelligent surface control in terahertz internet of things mesh networks

Terahertz communication is a key enabler for the sixth-generation wireless networks, offering ultra-high data rates and massive connectivity for the Internet of Things applications. However, its practical deployment is very limited due to severe propagation losses, molecular absorption, and high sensitivity to blockages, which further degrade link reliability and system performance, especially in dense and multi-user environments. Reconfigurable Intelligent Surfaces give an effective means to overcome these problems by enabling the programmable management of the wireless channel. The joint optimization of the RIS phase configuration and the transmit power in the THz systems leads to a high-dimensional and non-convex problem, which is further complicated by the dynamic channel conditions and multi-user interference. To address these challenges, this paper presents a Digital Twin–assisted AI-driven framework for adaptive RIS control in the THz IoT mesh networks. A physics-aware system model is developed that captures THz-specific propagation effects, including frequency-dependent path loss and molecular absorption. Based on this model the RIS-assisted communication problem is formulated as a multi-objective optimization task which jointly considers throughput, latency, and energy efficiency under quality-of-service constraints. The proposed framework uses a Digital Twin to simulate the network environment and enables the predictive evaluation of the RIS configurations while an AI agent learns efficient control policies through interaction with the virtual model, thus reducing dependence on instantaneous channel state information. A comprehensive simulation study is conducted to evaluate the proposed approach against the baseline, passive RIS, and Digital Twin–assisted greedy schemes. The results demonstrate that the proposed framework achieves performance gains in terms of signal quality, spectral efficiency, energy efficiency, and reliability, also ensuring faster convergence and improved scalability with increasing user density and RIS size. These findings show the potential of combining Digital Twin modelling with AI-based optimization to enable intelligent and adaptive control of RIS-assisted THz communication systems, thus paving the way for robust and scalable 6G IoT networks.