The convergence of molecular communication and terahertz (THz) electromagnetic signaling is emerging as a promising paradigm for enabling seamless information exchange in nanoscale networks. However, one of the critical challenges in such hybrid nano communication systems is efficient demultiplexing of heterogeneous signals under severe constraints of noise, interference, and energy availability. This paper proposes an AI-driven adaptive demultiplexing framework designed specifically for hybrid molecular–terahertz nano communication environments. The proposed approach leverages deep learning models to dynamically distinguish and separate overlapping signal modalities by learning complex temporal and spectral patterns inherent in both molecular diffusion channels and THz propagation channels. Unlike conventional static demultiplexing techniques, the adaptive framework continuously updates its parameters based on channel conditions, thereby improving signal fidelity, reducing error rates, and enhancing overall system throughput. Furthermore, the model incorporates lightweight architectures suitable for nanoscale implementation, ensuring energy efficiency and computational feasibility. Simulation results demonstrate that the proposed AI-driven strategy significantly outperforms traditional methods in terms of bit error rate, latency, and robustness against environmental variability. This work not only bridges the gap between bio-inspired molecular communication and high-frequency electromagnetic communication but also opens new avenues for intelligent nanoscale networking in applications such as targeted drug delivery, in-body sensing, and nano-Internet of Things (IoNT). The findings highlight the transformative potential of integrating artificial intelligence into next-generation nano communication systems.
Kumar et al. (Sat,) studied this question.