Path Evolution Model for Endogenous Channel Digital Twin towards 6G Wireless Networks

TL;DR

This work introduces Path Evolution Model (PEM), a new endogenous Channel Digital Twin for 6G that models the temporal evolution of individual channel paths from consecutive measurements. By extracting per-path features and using a neural network to propagate their evolutions, PEM creates digital replicas whose sum yields accurate, low-overhead CSI without relying on external sensors. It formalizes environment-invariant target functions and addresses distribution shifts across environments via per-path modeling, data augmentation, and continual learning, demonstrating robust generalization and pilot overhead reduction in simulations. The approach offers a practical, lightweight CDT that can be deployed with existing systems and shows potential for simultaneous CSI prediction and multi-functional sensing in dynamic 6G scenarios.

Abstract

Massive Multiple Input Multiple Output (MIMO) is critical for boosting 6G wireless network capacity. Nevertheless, high dimensional Channel State Information (CSI) acquisition becomes the bottleneck of 6G massive MIMO system. Recently, Channel Digital Twin (CDT), which replicates physical entities in wireless channels, has been proposed, providing site-specific prior knowledge for CSI acquisition. However, external devices (e.g., cameras and GPS devices) cannot always be integrated into existing communication systems, nor are they universally available across all scenarios. Moreover, the trained CDT model cannot be directly applied in new environments, which lacks environmental generalizability. To this end, Path Evolution Model (PEM) is proposed as an alternative CDT to reflect physical path evolutions from consecutive channel measurements. Compared to existing CDTs, PEM demonstrates virtues of full endogeneity, self-sustainability and environmental generalizability. Firstly, PEM only requires existing channel measurements, which is free of other hardware devices and can be readily deployed. Secondly, self-sustaining maintenance of PEM can be achieved in dynamic channel by progressive updates. Thirdly, environmental generalizability can greatly reduce deployment costs in dynamic environments. To facilitate the implementation of PEM, an intelligent and light-weighted operation framework is firstly designed. Then, the environmental generalizability of PEM is rigorously analyzed. Next, efficient learning approaches are proposed to reduce the amount of training data practically. Extensive simulation results reveal that PEM can simultaneously achieve high-precision and low-overhead CSI acquisition, which can serve as a fundamental CDT for 6G wireless networks.

Figures (6)

• Figure 1: Structure of PEM, where digital replicas of physical path evolutions are created.
• Figure 2: PEM operation framework, where the path extraction, path update and path evolution steps are illustrated in the bottom part.
• Figure 3: Analysis of environment-invariant target function in PEM is shown on the top. Then, an overview environment generalizable PEM is illustrated in the middle, which progressively addresses the distribution shift. Next, data augmentation and continual learning that address the varying user mobility are detailed at the bottom.
• Figure 4: Illustration of PEM applications (top), along with simulation scenario in Wireless Insite (bottom).
• Figure 5: An example of path evolution during UE mobility in env-1. Three physical paths are illustrated with a top view on the top side. Delay and AoA evolution of these paths are plotted on the bottom side: (a) path-1: LoS path; (b) path-2: reflection path from building-2; (c) path-3: reflection path from building-3.